U.S. Department of Commerce Volume 98 Number 1 January 2000 Hshery Bulletin U.S. Department of Commerce William M. Daley Secretary National Oceanic and Atmospheric Administration D. James Baker Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Penelope D. Dalton Assistant Administrator for Fistierles Scientific Editor Dr John V. Merriner Editorial Assistant Sarah Shoffler Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 101 Pivers Island Road Beaufort, NC 28516 ^ATE5 OV The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA98115- 0070. Periodicals postage is paid at Se- attle, WA, and at additional mailing of- fices. POSTMASTER: Send address changes for subscriptions to Fishery Bul- letin, Superintendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washington, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this De- partment. Use of funds for printing of this periodical has been approved by the Di- rector of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $50.00 domestic and $62.50 foreign. Cost per single issue: $19.00 domestic and $23.76 foreign. See' back for order form. Managing Editor Sharyn Matriotti National Manne Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C 15700 Seattle, Washington 98115-0070 Editorial Committee Dr Andrew E. Dizon National Marine Fisheries Service Dr. Harlyn O. Halvorson University of Massachusetts, Boston Dr Ronald W. Hardy University of Idaho, Hagerman Dr. Richard D. Methot National Marine Fisheries Service Dr Theodore W. Pietsch University of Washington, Seattle Dr Joseph E. Powers National Marine Fisheries Service Dr Harald Rosenthal Universitat Kiel, Germany Dr Fredric M. Serchuk National Marine Fisheries Service The Fishery Bulletin carries original research reports and technical notes on investi- gations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisher- ies in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Begin- ning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 98 Number 1 January 2000 Fishery Bulletin J* . % Contents The National Marine Fisheries Ser\^ce iNMFSl does not approve, recommend, or endorse any proprietary product or proprietary material mentioned m this publication- No reference shall be made to NMFS. or to this publication furnished by NMFS. in any advertising or sales promotion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the adver- tised product to be used or purchased because of this NMFS publication. Articles 1-13 Arkhipkin, Alexander I. Intrapopulation structure of winter-spawned Argentine shortfin squid, lllex argentinus (Cephalopoda, Ommastrephidae), during its feeding period over the Patagonian Shelf 14-24 Beacham, Terry D., Khai D. Le, Monique R. Raap, Kim Hyatt, Wilf Luedke, and Ruth E. Withler Microsatellite DNA variation and estimation of stock composition of sockeye salmon, Oncorhynchus nerka, in Barkley Sound, British Columbia 25-40 Clear, Naomi P., John S. Gunn, and Anthony J. Rees Direct validation of annual increments in the otoliths of juvenile southern bluefin tuna, Thunnus maccoyii, by means of a large-scale mark-recapture experiment with strontium chloride 41-63 Francis, Malcolm P., and John D. Stevens Reproduction, embryonic development, and growth of the porbeagle shark, Lamna nasus, in the southwest Pacific Ocean 64-74 Govoni, John Jeffrey, Bruce W. Stender, and Oleg Pashuk Distribution of larval swordfish, Xiphias gladius, and probable spawning off the southeastern United States 75-85 Hayes, Daniel B. A biological reference point based on the Leslie matrix 86-95 Lenihan, Hunter S., and Fiorenza Micheli Biological effects of shellfish harvesting on oyster reefs: resolving a fishery conflict by ecological experimentation 96-117 Love, Milton S., Jennifer E. Caselle, and Linda Snook Fish assemblages around seven oil platforms in the Santa Barbara Channel area Fishery Bulletin 98(1), 2000 118-126 Nemerson, David, Steven Berkeley, and Carl Safina Spawning site fidelity in Atlantic bluefin tuna, Thunnus thynnus: the use of size-frequency analysis to test for the presence of migrant east Atlantic bluefin tuna on Gulf of Mexico spawning grounds 127-138 Seyoum, Seifu, Michael D. Tringali, Theresa M Bert, Doug McElroy, and Rod Stokes An analysis of genetic population structure in red drum, Sciaenops ocellatus, based on mtDNA control region sequences 139-152 Shields, Jeffrey D., and Christopher M. Squyars Mortality and hematology of blue crabs, Callinectes sapidus, experimentally infected with the parasitic dinoflagellate Hematodinium perezi 153-166 Stevenson, Jill T, and David H. Secor Age determination and growth of Hudson River Atlantic sturgeon, Acipenser oxyrinchus 167-188 Steves, Brian P., Robert K. Cowen, and Mark H. Malchoff Settlement and nursery habitats for demersal fishes on the continental shelf of the New York Bight 189-198 Wetherbee, Bradley M. Assemblage of deep-sea sharks on Chatham Rise, New Zealand 199-218 Wyanski, David M., D. Byron White, and Charles A. Barans Growth, population age structure, and aspects of the reproductive biology of snowy grouper, Epinephelus niveatus, off North Carolina and South Carolina Notes 219-221 Balart, Eduardo F., Jeanette Gonzalez-Garcia, and Carlos Villavicencio-Garayzar Notes on the biology of Cephalurus cephalus and Parmaturus xaniurus (Chondrichthyes: Scyliorhinidae) from the west coast of Baja California Sur, Mexico 222-225 Galindo-Bect, Manuel S., Edward P. Glenn, Henry M. Page, Kevin Fitzsimmons, Luis A. Galindo-Bect, Jose M. Hernandez-Ayon, Robert L. Petty, Jaqueline Garcia-Hernandez, David Moore Penaeid shrimp landings in the upper Gulf of California in relation to Colorado River freshwater discharge 226 Subscription form II 1 1 Abstract.— Stock structure dynamics of the important commercial squid Illcx argentinus were studied by using bio- logical data from about 25 thousand squid caught January-April 1991 by Rus- sian trawlers in three fishery regions: 51-52°S; 47-49°S within the exclusive economic zone of Argentina (EEZA); and 45-47°S outside the EEZA. A total of 2664 statoliths were read to prepare age-length keys for each 10-day interval of the period studied. It was found that between January and April, the Patago- nian shelf south of 45°S was a feeding ground of two intraspecific groups of winter-hatched /. argentinus: a shelf group that matured at medium sizes (ShG) and a slope group that matured at large sizes (SIG). After massive immi- gration of /. argentinus from the north in January-February into the two fish- ery regions within 45-49°S, the stock structure remained rather stable until April, composed predominantly of June- and July-hatched squid. Squid grow and mature rapidly, and males mature at younger ages (from one to two months) than do females. During feeding, some redistribution of the stock was observed: maturing and mature SIG squid (mainly females) tended to shift ft-om the shelf (130-150 m depth) in a northeast direc- tion and concentrate over the shelf edge (160-170 m depth). In April, mature SIG squid began to shift to the continental slope around 45-47°S and migrated to depths >600 m where they then mixed with schools of SIG squid that had fed in the region 51-52°S and that were already migrating northwards along the slope. ShG squid remained on the shelf and made their prespawning northward migrations along the shelf edge. Intrapopulation structure of winter-spawned Argentine shortfin squid, ///ex argentinus (Cephalopoda, Ommastrephidae), during its feeding period over the Patagonian Shelf Alexander I. Arkhipkin Atlantic Research Institute ol Marine Rshenes and Oceanography (AtlantNIRO), 5 Dm. Donskoy street Kaliningrad, 236000 Russia Present address, Fishenes Department, Falkland Islands Government PO Box 598, Stanley, Falkland Islands E-mail address fish figiaihonzon.co.fk Manuscript accepted 13 Julv 1999. Fish. Bull. 98:1-13(2000). The Argentine shortfin squid, Illex argentinus (de Castellanos, 1960), is a common neritic species occurring in waters ofFBrazil, Uruguay, Argen- tina, and the Falkland Islands in the Southwest Atlantic (Nesis, 1987). This squid is an important world fish- ery resource. According to the FAO (1997), since 1978, its total annual catch has varied firom 180 to 250 thousand metric tons (t), achieving 300-330 thousand tons in 1993-95. However, actual total annual catch of/, argentinus could reach up to 700 thousand tons (Uozumi and Shiba, 1993). Illex argentinus is captured by the international fleet consisting of both jigging light vessels (mainly fi-om Asian countries) and trawlers (mostly fi-om European countries: Poland, Spain, and Russia (former USSR) in two fishery regions off the Argentine Exclusive Economic Zone (EEZA): at 42°S and 45-47°S. In the 1970s and 1990s, /. argen- tinus was also caught in consider- able numbers within the EEZA and Falkland Islands Interim Conserva- tion Zone (FICZ) (Csirke, 1987; FAO, 1997 ). Such an extensive fishery has induced detailed studies of different biological aspects of /. argentinus in order to monitor and forecast its stock structure dynamics. Originally, /. argentinus was con- sidered to be a single stock (Sato and Hatanaka, 1983; Csirke, 1987). Then it was found that the species consisted of two populations differ- ing both by season and place of their spawning: an abundant winter- spawning population (more than 95% of the total stock) and a small sum- mer-spawning population (Hatanaka et al., 1985; Hatanaka, 1988). Bru- netti ( 1988) divided winter-spawning squid into two stocks: the bonaerensis north Patagonian stock (BNPS) and the south Patagonian stock (SPS), dif- fering both by feeding groimds and size of adults (medium and large, respec- tively). On the basis of occurrence of mature females in different seasons, Nigmatullin (1989a) revealed that /. argentinus spawn throughout the year, and proposed to subdivide the total stock into four seasonal spawning groups. Although Tsygankov (1987) found qualitative differences in three loci of esterases extracted from the buccal muscles of various intrapopu- lational groups of 7. argentinus, the taxonomic status of these groups, however, still remained imclear. Anal- ysis of dynamics in length-frequency compositions showed that comple- tion of the life cycle of/, argentinus populations took one year (Hatanaka et al., 1985; Hatanaka, 1986) and this length of time was then con- firmed by statolith aging investiga- tions (Arkhipkin, 1990; Rodhouse and Hatfield, 1990). The life cycle of the most abundant winter-spawned group can be subdi- vided into five stages: a postlarval Fishery Bulletin 98(1) period that takes place in waters off Brazil and Uru- guay in August-September (Leta, 1987; Santos and Haimovici, 1997); a juvenile period that takes place in shelf and oceanic waters off Uruguay and Argentina in September-December (Biiinetti, 1988; Parfeniuk et al., 1992); a feeding period that takes place on the Patagonian and Falkland Islands shelves in January- April (Brunetti, 1988; Hatanaka, 1988); a prespawn- ing period that takes place on the shelf edge and slope off Argentina and Uruguay in May^uly (Hatanaka, 1986, 1988; Arkhipkin, 1993); and a spawning period that takes place in shelf and slope waters off north- em Argentina, Uruguay, and Brazil in July-August (Brunetti, 1988; Santos and Haimovici, 1997). Squid aggregate and are fished mainly during their feeding period on the shelf, as well as during their prespawn- ing period on the shelf edge and slope (Nigmatullin, 1989b). Stock structure dynamics of /. argentinus during their feeding period on the Patagonian Shelf were studied both in the fishery region of 45^7°S outside the EEZAand within the FICZ by using data obtained fi-om Japanese jigging vessels (Rodhouse and Hat- field, 1990; Uozumi and Shiba, 1993). It was found that the age composition of /. argentinus catches changed between January and April owing to the grad- ual migrations of feeding schools of these squid. Ear- lier hatched squid immigi'ated to and emigrated fi-om the fishery region earlier than later hatched groups (Uozumi and Shiba, 1993). Squid of the former group had slower growth rates than those of the latter group (Rodhouse and Hatfield, 1990). Data from trawling vessels showed that, during the prespawning period, winter-spawned /. argentinus made active northward migrations from the southern Patagonian Shelf along the continental slope of Argentina. Squid migrated in waves of abundance, consisting of 2-4 successive monthly generations. Males moved 2-3 weeks earlier than females of the same monthly group (Arkhipkin, 1993). An analysis of length-frequency distributions of /. argentinus showed that the jigging fishery had a higher selectivity for squid than did the trawl fishery (Koronkiewicz, 1995), and therefore data from the trawl fishery reflected the natural population distri- bution far better than those from the jigging fishery. In the present report I examined the stock structure dynamics of/, argentinus during the January-April feeding period, using both data from two research vessels and two commercial trawlers (both fishing within the fishery) and statolith aging techniques, and comparing these stock structure data with those obtained by the Japanese jigging fishery both within and outside the EEZA. Together with the data obtained during the prespawning period (April-June) (Arkhip- -40°S -4i°S • -44°S -46°S -48°S -50°S -52°S -54°S -64''W -62°W -60°W -58°W -56°W Figure 1 Sampling locations of Illex argentinus off (circles I and within ( triangles ) the Exclusive Economic Zone of Argen- tina (EEZA) in the southwest Atlantic in January-April 1991. kin, 1993), the results of the present study make it possible to reconstruct the full picture of the stock structure dynamics of/, argentinus during the entire fishery period. Materials and methods Data on the Ai'gentine shortfin squid, Illex argentinus, for the present study were collected during four exper- imental surveys carried out in the fishery region of 45-47°S outside the EEZA by the Soviet research ves- sels A«c/mr and Volzhanin (2700 GRT) and within the EEZA (in 47^9°S and 51-52°S) by the fishing trawl- ers Petropavlovskaya krepost and Batiliman (4000 gross registered tonnage [GRT] ) between January and April 1991 (Fig. 1). Trawls were conducted with differ- ent types of rope trawls with a mean horizontal open- ing of 60-75 m and mean vertical opening of 40-50 m. Trawls were made in the superficial water layer at night and near bottom in the daytime at bottom depths ranging from 140 to 190 m in January-March Arkhipkin: Intrapopulation structure of ///ex argentinus during its feeding period over the Patagonian Shelf and from 190 to 660 m in April. The duration of trawls ranged from 4 to 8 hours, and the average towing speed was 6-8 km/h. Every ten days, average daily catch per unit of effort (CPUE) was calculated as a mean of daily CPUEs of the Soviet fishing trawlers that caught squid in each fishery region. Mean CPUEs were calculated separately for the fishery region of 45-47°S outside the EEZA (where a majority of fishing vessels are 2700-GRT trawlers) and for the fisheiy regions within the EEZA (where all fishing vessels were 4000-GRT trawlers). To obtain an objective picture of the squid fisheiy, CPUEs of the 2700 GRT trawlers may be adjusted to those of 4000-GRT trawlers by a coeffi- cient of 0.7 (Arkhipkin, 1993). Length-frequency sampling A random sample of one hundred squid was taken by scientific observers from each of two catches (at night and day) everyday on board each of the four ves- sels. Dorsal mantle length (ML) was measured to the nearest 1 mm, total body weight (BW) was weighed to the nearest 1 g. Sex and maturity stages were iden- tified according to the maturity scale elaborated for Illex argentinus (Nigmatullin, 1989a). Sex ratio was determined. A total of 16,436 squid were analyzed in the fishery region of 45^7°S outside the EEZA and 9893 squid were analyzed within the EEZA. Every ten days, three length-frequency curves of males and females were constructed for three maturity periods; immature ( maturity stages 1-2 ), maturing ( maturity stages 3-5 J ) and mature (maturity stages 5.,-5g). Age sampling and statolith processing Eveiy ten days, from January to April, statoliths were dissected from 100-150 individuals of Illex argenti- nus from two successful catches on board each of the four vessels. The length-frequency distribution of the 10-day age sample was proportional to the length-frequency distribution of squid caught during these ten days. Statoliths were washed in distilled water and stored in oil-paper envelopes in 969f eth- anol. A total of 1700 statoliths were sampled in the fishery region of 45^7°S outside the EEZA, and 1150 statoliths were collected within the EEZA. All statoliths sampled were processed by statolith aging techniques in the Laboratory of Commercial Invertebrates ofAtlantNIRO( Arkhipkin, 1991). Stato- hth terminology follows Clarke (1978) and Lipinski et al. (1991). Statoliths were attached to the micro- scopic slides with Pro-texx mounting medium and were ground on both sides on a wet waterproof sand- paper of 1000-grit gi-ade. During grinding, the stato- lith rostrum was completely removed, so that growth increments could be easily distinguished from the nucleus to the edge of the dorsal dome. Ground stato- liths were embedded in Canada balsam and covered with glass covers. Ready preparations were placed in an oven at 90-100°C for one hour to dry the balsam and improve the readability of growth increments. Statoliths were read under a Biolam Rl light micro- scope at 450-500X magnification. To avoid possible counting errors, each statolith was counted twice by two observers using the gradation of an eye-piece micrometer The total number of growth increments for each specimen was obtained as a mean of these rep- licate counts if the deviation between the two counts was less than 57(. If deviation exceeded 5%, the stato- lith was recounted by the two observers once more. If such deviation did not decrease after the recounting, the statolith was rejected from further analysis. From the whole sample, 1597 statoliths fi-om the region outside the EEZA (93.97^) and 1067 statoliths from the region within the EEZA (92.7% ) were prepared and read. Length-at-age data analysis Deposition of putative growth increments within /. argentinus statoliths has not yet been validated. However, incorporation of either tetracycline or stron- tium marks into statoliths of the congeneric species I. illecebrosus kept in captivity has shown that growth increments are formed daily ( Dawe et al., 1985). Stato- lith microstioicture in both species is similar; therefore growth increments within statoliths of/, argentinus are considered to form daily in the present paper. Hence, their total number was considered to repre- sent squid age in days. Hatching dates were backcal- culated. Month classes of hatching were defined by- pooling squid into each month of hatching ( Ai'khipkin, 1990; Rodhouse and Hatfield, 1990). Length-at-age data were analyzed separately for both sexes. The 10-day age stnacture was determined by construction of age-length keys. Age-length keys were constructed by using numbers of squid for each month class sepa- rately for each sex and maturity period (Arkhipkin et al., 1996). Results CPUE dynamics Region 45-47°S Illex argentinus were caught in all trawls of the research vessels (Fig. 2). In January, schools of squid aggregated mainly north of the region. Trawl catches were variable ( from 1 to 20 t per vessel day, \Jd ), and mean January CPUEs were low (8-9 t/d ). Fishery Bulletin 98(1) During the first 10-day period of February, concen- trations of /. argentinus were observed in all parts of the region. Squid concentrated near the bottom during the daytime and ascended to the upper water layers at night. The CPUE was twice as high as in January, and the squid fishery stabilized at 16-17 t/d. However, aggregations of /. argentinus quickly dis- persed, and during the second 10-day period of Febru- ary, the CPUE fell sharply to the January level (Fig. 2). Catches of squid were low but stable (around 10 t) imtil the second 10-day period in April, when fish- ing vessels shifted from the shelf edge ( 170-190 m depth) to the continental slope (440-660 m depth) and changed fishing tactics. The vessels performed near-bottom trawls at the shelf edge during the day- time, and on the continental slope at night. These tac- tics considerably increased CPUE (up to 15-18 t/d) for the third 10-day period of April and in the begin- ning of May because the fleet began to target not only shelf aggregations but also the slope aggregations of /. argentinus that were beginning to appear at that time. Region AT-AQ^S Fishing trawlers operated in this region between February and the second 10-day period of March. The fishing tactics were the same as those used in the previous region. Abundance of/, argenti- nus was considerably greater than that in the region outside the EEZA (Fig. 2). The peak of CPUE was observed in the third 10-day period of February (51 t/d). Region SI-SZ'S All fishing vessels that had oper- ated in the 47-49°S region moved to the fishing region west of the Falkland Islands during the second 10-day period of March and fished there until the middle of May (Fig. 2). The fishing tactics were different fi-om those used in the two previous regions. Trawlers fished for squid in midwater both at night and in the day- time. CPUE in March-April was high, similar to those in the 47-49°S region with a prominent peak (53 t) during the second 10-day period of April. At the begin- ning of May, CPUE decreased sharply and the fleet ceased fishing for squid in the region. Sex ratios Region AS-AT'S The proportion of females was the highest during the second 10-day period of January (ca 80% of the total sample). From the end of Janu- ary through the beginning of February, the proportion of females decreased sharply (to 55—60%). Sex ratio was close to 1:1 between the second 10-day period of February and the first 10-day period of April, when the proportion of females occurring on the shelf edge decreased to 30%. However during the third 10-day period of April on the continental slope at depths of 480 m, the sex ratio was found to be close to 1:1, and females prevailed in catches (65%) at deeper depths (630 m) (Fig. 3). Region 47-49°S Except the first 10-day period of February (when the sex ratio was close to 1:1), males always predominated in catches at a ratio of 2:1 (Fig. 3). Region 51-52°S The sex ratio was close to 1:1, but the proportion of females tended to decrease from 55% to 45% (Fig. 3). 60 - 50 40 30 - LU 20 Z) a. o 10 -• 45-47°S I - - ■- • 47-49°S i— ■* — 51-52°S u -* — — ^ 12312312312312 Jan Feb Mar Apr May Figure 2 Mean CPLIE (tons per vcs.sel day i of the Russian trawlers in the lllex argentinus fi.shcry in three fishery regions of the Southwest Atlantic in 1991: 4.5-47°S ibold solid line), 47-49°S (dotted linei and .51-52°S (dashed line). Length and hatching-month composition Region 45-47°S During the second 10-day period of January, June-hatched maturing females (modal sizes 210 mm ML) and mature males (200 mm) predominated in catches. The proportion of mature May-hatched squid was low. Hatching-month com- position changed considerably during the third 10-day period of January owing to an appearance of maturing and mature May-hatched squid (males of 200 mm and females of 220-250 mm ML) in the region (Fig. 4). Massive appearances of dense schools of immi- grating June and July-hatched immature and maturing females (200-220 mm ML) and mature males (200-210 mm ML) were observed in the first 10-day period of February which resulted in a double increase in CPUE of the fishing fleet (Fig. 2) and in corresponding changes in the hatching- month compositions of squid: June-hatched squid Arkhipkin; Intrapopulation struaure of I/lex argentlnus during its feeding period over the Patagonlan Shelf 1] (A 0) |0 75 ° 0.5- \ « •■-■-•■••- v^ Proportio o en I--A-A ■- B O D <> E 1 2 3 12 3 12 3 12 3 12 Jan Feb Mar Apr May Figure 3 Proportion of fern ales in sex ratio of Illex argentinus in three fishery regions of the Southwest Atlantic in 1991: (A) 51-52°S; | (B)47^9°S;and( C-E) 4.5-47°S. Depths for the last region were 160-190 m(C 1.48 Om(D), and 630 m (E). became the most abundant in catches, as in the middle of January. During the second and third 10-day peri- ods of February, hatching-month composition was the same as that of January, with June- and July-hatched squid predominating. However during the third 10-day period, the number of large mature females (280 mm ML) increased considerably (Fig. 4). In March, the hatching-month composition of the /. argentinus catch was approximately similar to that of the second 10-day period of February; June- and July- hatched squid were the most abundant. The pro- portion of July-hatched squid increased by the end of March with a corresponding decrease in June-hatched squid. Modal sizes of males increased from 220-230 mm ML in the beginning to 230-240 mm ML at the end of the month. Length composition of females was bimodal (240-250 and 290-300 mm ML). The pro- portion of immature females decreased, whereas the proportion of mature females increased by the end of March (Fig. 5). In April, hatching-month composition of/, argenti- nus caught over the shelf edge (170-190 m depths) remained almost similar to that of the second and third 10-day periods of March; July-hatched squid were predominant. Almost all males were mature. Immature females disappeared from catches in the first and second 10-day periods of April, but were caught in small numbers in the third 10-day period. At the beginning of the month, the length composition of both maturing and mature females was bimodal (260-270 and 300 mm ML). During the second 10-day period, large maturing and mature females (310 mm ML) began concentrating over the shelf edge, whereas medium-size females (270-280 mm ML) were still dispersed (Fig. 6). These concentrations caused another increase in CPUE for the fishing fleet (Fig. 2). During the third 10-day period of April, large mature squid (females of 310-320 mm ML and males of 270 mm ML) appeared in deeper waters over the continental slope, and they became most abundant at 480-630 m depths. Medium-size squid (maturing females of 280 mm ML and mature males of 250 mm ML) remained over the shelf edge (Fig. 6). Region 47-49°S During the first 10-day period of February, the length composition of males was uni- modal (230 mm ML) and most of these males were mature. Among females, two different modal groups occurred in the catches: immature June-hatched females (220 mm ML) and maturing and mature April- and May-hatched females (260 mm ML). Large catches of June- and July-hatched squid were evident between the second 10-day period of Feb- ruary and first 10-day period of March (Fig. 2). Hatching-month compositions did not change sig- nificantly in this period, June- and July-hatched squid were caught almost in equal proportions. Mature males increased slightly in length from 230 to 240 mm ML. The proportion of immature females decreased and that of maturing and mature females increased by the second 10-day period of March (Fig. 7). Region 51-52°S Hatching-month composition was similar between the third 10-day period of March and second 10-day period of April; July-hatched males and females were predominant in catches. Length compo- sitions were unimodal for both sexes. Except during the third 10-day period of March when about a third of males were maturing, most of the males were mature, and their sizes increased from 260 mm ML at the end of March to 280 mm ML at the end of April. Females grew more rapidly in length than did males (from 280 to 320 mm ML). They matured quickly; imma- ture females prevailed at the end of March, whereas maturing females were predominant at the end of April. During the third 10-day period of April, age composition of/, argentinus changed owing to a high proportion of August-hatched squid (Fig. 8). Comparative comments Simultaneous sampling in the regions of 45^7°S and 47^9°S between February and March and in the regions of 45-47°S and 51-52°S between March and April enabled a comparison of both length and hatching- month compositions of/, argentinus in these regions. Hatching-month compositions were almost similar in the regions of 45-47°S and 47^9°S during the same 10-day periods (Figs. 4, 5, and 7). Modal sizes Fishery Bulletin 98(1) of mature males were about 10-20 mm greater in the portion of large mature females was much higher at southern region than those in the northern region. 47-49°S (Fig. 7) than at 45-47°S (Fig. 4). After the During the first 10-day period of February, the pro- second 10-day period of February, the opposite situa- Females Jan, H 100 160 2O0 2S0 aOO 300 400 Jan, M no 160 200 260 aOO 360 400 Feb, I it • 100 160 200 260 900 360 4O0 Feb, N 100 160 200 260 300 360 400 Feb, HI 100 160 200 260 300 380 400 Mantle length (mm) Males Jan. I no I6O 200 260 300 360 400 Jan, wo 160 200 260 300 360 400 Feb, I no 160 200 260 900 960 400 Feb, N no I60 200 260 300 360 400 Feb, III no 160 200 260 300 960 Mantle length (mm) Age structure Jan, n Jan, HI Feb, I Feb, n ^ mn{ Fab, III Hatching month Figure 4 Length-frequency compositions of immature (dotted line), maturmg (solid line) and mature (bold solid line) squid and hatching month compositions (age structures) of females (black bars) and males (dashed bars) of ///t'.v argenliniia m the fisher\- region of 45-47°S outside the EEZAin January-February 1991. Number of males in parentheses. Arkhlpkin: Intrapopulation structure of ///ex argentlnus during its feeding penod over the Patagonian Shelf tion was observed; the proportion of mature females decreased at 47-49°S and increased at 45-47°S. Both the length and hatching-month compositions of/, argentinus catches were different over the Pata- gonian Shelf at 45-47°S and 51-52°S. In the southern region, squid were 20-30 mm larger, about a month younger, and less mature than in the northern region, except during the third 10-day period of April when the Females Males Ntar. I Mar, I no tw too leQ 3oo aeo «oo too wo ioo i«o aoo aso Mar. a Mar, N 100 160 aoo 260 aoo aao 400 100 160 too 26o aoo aao 400 s £ Mar, HI Mar, III 100 I6O too 260 aoo 360 400 100 160 800 860 300 360 400 Apr. Apr, I 100 wo too t60 aoo a«o 400 100 160 too too aoo a6o 400 Apr. II Apr. II no lao too a60 aoo aao 400 Mantle length (mm) wo 160 too 860 300 360 400 Mantle length (mm) Age structure ao 40 M P Mar, 1 SO B P to 10 kl J n • M (ui m A M J J A • O Mar, n 1 n ' M (••) A M J J A 8 O Mar, III n • M iM> A M J J A 6 O Apr, I A M J J A S O Apr, II A M J J A » O Hatching month Figure 5 Length-frequency and hatching month compositions (age structures) of Illex argentinus in the fishery region of 45-^7°S outside the EEZAin March-April 1991. Symbols are the same as in Figure 4. Fishery Bulletin 98(1) Females 190 m 100 leO loo 360 300 360 400 480 m Ida aoo >6o 300 630 m 100 wo 200 260 SCO 360 400 Mantle length ^mm) Males Age structure 190 m 190 m 200 260 300 360 480 m 480 m 630 tn 630 m wo 160 100 260 300 360 400 1 Mantle lengtti (mm) Hatching month Figure 6 Length-frequency and hatching month compositions (age structures) of Ulex argentinus at different depths of the fishery region of 45-47°S outside the EEZA between 21 and 30 April 1991. Symbols are the same as in Figure 4. hatching-month compositions were practically similar to those for the continental slope (630 m) at 45-47°S and shelf (190-210 m) of 51-52°S. However, in spite of the similarity in modal length both in males and females in the last case, most of the females were mature in 45-47°S, whereas those at 51-52°S were still maturing (Figs. 5 and 8). Discussion Stock structure dynamics Studies of the length-at-age structures for immature, maturing, and mature squid (separately! of both sexes, with a 10-day interval, revealed in detail the intrapop- ulation structure dynamics and migratory patterns of /. argentinus during the January-April feeding period on the Patagonian Shelf Previous investigations, in which length-at-age data were pooled separately for each sex, revealed only general patterns in the age structure dynamics of/, argentinus (Rodhouse and Hatfield, 1990; Uozumi and Shiba, 1993). Stock structure of winter-hatched /. argentinus was rather stable during the feeding period (January- April). After massive immigration of June- and July- hatched squid into the region 45^7°S from the end of January through the beginning of February, prob- ably from an area farther north on the Patagonian Shelf (Hatanaka, 1988; Parfeniuk et al, 1992), the age structure of squid remained rather stable until the middle of April. During each 10-day period, from four to five month classes were observed, similar to the number obtained from the jigging fishery data (Uozumi and Shiba, 19931. Predominance of monthly classes changed gradually fi'om June-hatched squid in February to July-hatched squid in March-April. A considerable portion of the June and July-hatched squid continued their southward feeding migrations and reached 47-49°S by the end of February, which Arkhipkin: Intrapopulation structure oU/lex argentinus dunng its feeding period over the Patagonian Shelf Females Males Feb, I !00 WO ZOO 2S0 300 360 40O Feb, I Feb. I wo wo 200 2B0 300 3S0 400 Feb, II wo wo aoo MO *oo 3so 400 Feb, III o c 600 m), where they mixed with the already migrat- ing schools of July- and August-hatched squid that had fed in the southern part of the Patagonian Shelf and around the Falkland Islands (Arkhipkin, 1993). Such a redistribution of /. argentinus aggregations caused a rather sharp decrease in the CPUE of jig- ging vessels on the shelf in April-May and a simulta- neous regi'ouping of trawlers — a shift from the shelf to one over the continental slope (Hatanaka, 1988; Nig- matullin, 1989b). Medium-size squid remained on the shelf and probably made their prespawning migra- tions along the shelf edge. Stock structure of winter-hatched ///ex argentinus From the complex of biological characteristics (stato- lith microstructure, modal length in different months, sizes at maturity, types of feeding, and prespawning migrations), it is possible to consider the Patagonian Shelf south of 45°S as a feeding ground for two intra- specific groups of winter-hatched /. argentinus: the "shelf group that matures at medium sizes" (ShG) and the "slope group that matures at large sizes" (SIG). These two groups correspond well with the bonaeren- sis north Patagonian stock (BNPS) and south Pata- gonian stock (SPS) distinguished by Brunetti (1988) by using length-frequency analysis. Later, Brunetti et al. (1998) postulated that the spawning of both groups takes place near the shelf edge and over the continental slope; the BNPS squid spawn north of 43°S in winter, whereas the SPS squid spawn south of 43°S in autumn. It was shown however that the SPS squid definitely migrated from the southern part of the Patagonian and Falkland shelves along the con- tinental slope farther north at 41^2°S (Arkhipkin, 1993), but location of their spawning grounds is still unknown (Haimovici et al., 1998). The shelf group also corresponds to the winter shelf group (WSG), and the slope group corresponds well to the winter oceanic group (WOG), both (WSG and WOG) of which were distinguished by different loca- tions of juvenile feeding and by type of life cycle (Parfeniuketal., 1992; NigmatuUin and Laptikhovsky, 1996). The shelf group of /. argentinus has a neritic life cycle, characterized by the following features: spawn- ing in warm waters of the northern part of the spe- cies range (27-36°S); southward feeding migrations of juveniles < 100-150 mm ML over the Patagonian Shelf; a "shelf type dark zone within the statolith micro- structure (Arkhipkin, 1993); fast juvenile growth but rather slow growth of immature squid; medium sizes at maturation (males at 160-220 mm ML, females at 180—240 mm ML); medium maximum sizes for mature squid (males of 180-260 mm ML, females of 220-320 mm ML); and northward prespawning migrations over the shelf The slope group of /. argentinus has an oceanic-slope life cycle characterized by the following features: slope spawning in the northern part of the species area (27-36°S); southward feeding migrations of juveniles < 100-150 mm ML in the open part of the Argentine Basin; an "oceanic" type dark zone within the statolith microstructure (Arkhipkin, 1993); slow juvenile growth but rather fast growth of immature squid; large sizes at maturation (males at 180-240 mm ML, females at 240—340 mm ML); large maximum sizes for mature squid (240-340 mm ML, females up to 280—400 mm ML); and northward prespawning migrations over the continental slope. The taxonomic status of the two groups of winter-spawned /. argenti- nus remains unclear (Arkhipkin and Scherbich, 1991; Parfeniuk et al., 1992; NigmatuUin and Laptikhovsky, 1996; Santos and Haimovici, 1997). Interannual changes in stock structure It has been shown that growth rates of/, argentinus from the same hatching month vary to a lesser extent between different years from those of the dif- ferent months of hatching within one year (Arkhip- kin and Laptikhovsky, 1994). Thus it is possible to make comparisons of modal lengths of squid from the same month of hatching but in different years. The results of this study (based on data collected by the trawl fishery in 1991) are somewhat different from those obtained from the Japanese jigging fish- ery in 1989-1990 (Uozumi and Shiba, 1993). Gener- ally, during the same month and in the same region of sampling, a majority of males and females caught by jigs in 1989 were about a month younger and correspondingly 20-30 mm smaller than those sam- pled by the trawl fishery in 1991 (Figs. 8 and 9 in Uozumi and Shiba, 1993; and Figs. 4 and 5 of the pres- ent study). Unfortunately, there are no data on the length-frequency composition of trawl-caught /. argen- tinus in 1989 (Arkhipkin and Laptikhovsky, 1994), and Fishery Bulletin 98(1) thus it is difficult to explain the reasons for such a dif- ference in length composition between the two years. It was found that /. argentinus caught by jigging gear were significantly larger and more mature (especially females) than trawl-caught squid fished in the same location and time (Koronkiewicz, 1995). Thus, differ- ences in age and length compositions observed in 1989 and 1991 can be explained by interannual changes in population structure of/, argentinus rather than by various selectivity of the two different sampling gears. The results of the present study show that in Janu- ary-April, the international squid fishery in the south- west Atlantic catches aggregations of both groups of winter-spawned/, argentinus. Squid of the shelf group are captured by trawlers and jigging vessels over the depth range of 150—200 m mainly in the region of 45^9°S. Squid of the slope group are caught by trawl- ers and jigging vessels mainly in the southern part of the Patagonian Shelf within the EEZA(47-51°S) over the depth range of 150-250 m in February-March, and by trawlers over the continental slope (45^7°S) at depths of 600-700 m in April. Acknowledgments I gratefully acknowledge the generous help of scientific observers of the trawlers Anc/ior, Volzhanin, Petropav- lovskaya krepost and Batiliman for data sampling. I would like to thank L. A. Vavilova and A. B. Mikheev for processing statoliths, Ch. M. Nigmatullin and A. Z. Sundakov for discussions and comments on an ear- lier version of the manuscript. The edtiorial work of Emma Hatfield (NOAA, Woods Hole) and Emma Jones (FIFD, Stanley, Falkland Islands), who helped with the text in English, is most appreciated. Literature cited Arkhipkin, A. I. 1990. Edad y crecimiento del calamar Illex argentinus. Frente Maritimo 6 (sec. A):25-35. 1991. Methods for cephalopod age and growth studies with emphasis on statohth ageing techniques. In P. Jereb, S. Ragonese, and S. von Boletzky (eds.l, Squid age deter- mination using statohths. p. 11-17. Note Tecniche e Preprints deiristituto di Tbcnologia della Pesca e del Pescato (N.T.R,- IT.PP.) Special Publ. 1993. Age, growth, stock structure and migratory rate of pre- spawning short-finned squid Illex argentinus based on stato- lith ageing investigations. Fish. Res. 16:31.3-3.38. Arkhipkin, A.I., V. A, Bizikov, V. V. Krylov, and K. N. Nesis. 1996. Distribution, st;us. In T. Okutani (ed. i. Contributed papers to international sympo- sium on large pelagic squids (July 18-19, 1996), p. 217-232. JAMARC, Tokyo. Clarke, M. R. 1978. The cephalopod statolith — an introduction to its form. J. Mar. Biol. Assoc. UK. 58:701-712. Csirke, J. 1987. Los recursos pesqueros patagonicos y las pesquerias de altura en el Atlantico Sud-occidental. FAO Doc. Tec. Pesca 280. Rome, 78 p. Dawe, E. G., R. K. O'Dor, P. H. O'Dense, and G. V. Hurley. 1985. Validation and application of an ageing technique for short-finned squid i Illex illecebrosus). J. Northwest Atl. Fish. Sci. 6:107-116. FAO Yearbook. 1997. Fishery statistics. Catches and landings, 1995, vol. 80. FAO Fisheries Series 48. 714 p. Haimovici, M., N. Brunetti, P. G. Rodhouse, J. Csirke, and H. R. Leta. 1998. Illex argentinus. In P. G. Rodhouse. E. G. Dawe, and R. K. O'Dor (eds.), Squid recruitment dynamics: the genus Illex as a model, the commercial Illex species and influences on variabihty, p. 27-.58. FAO Fish. Tech. Pap. 376. FAO, Rome. Hatanaka, H. 1986. Growth and life span of short-finned squid Illex argen- tinus in the waters off Argentina. Bull Jap. Soc. Sci. Fish., 52:11-17. 1988. Feeding migration of short-finned .squid ///t'.v argenti- nus in the waters off Argentina. Nippon Suisan Gakkaishi ,54(8):1343-1349. Hatanaka, H., S. Kawahara, Y. Uozumi, and S. Kasahara. 1985. Comparison of life cycles of five ommastrephid squid fished by Japan: Todarodes pacificus. Illex illecebrosus, Illex argentinus, Nototodarus sloani sloani and Nototodarus sloani gouldi. NAFO Sci. Counc. -Studies 9:59-65. Koronkiewicz, A. 1986. Growth and life cycle oflllex argentinus from the Pata- gonian shelf and Polish squid fishery in the region. 1978-85. ICES C.M./K 27:1-25. 1995. Size and maturity differences between trawl and jigger caught short-finned squid Illex argentinus (Cephalopoda: Ommastrephidae). Acta Ichtyol. Pise. 25 (1):97-112. Leta, H. R. 1987. Descripcion de los huevos yjuveniles de Illex argentinus (Ommastrephidae) y juveniles dei«/(go 6ra.s/fe«s(s( Loligini- dae) en la Zona Comun de Pe.sca Argentino-Uruguaya. Publ. Cient. Institute Nacional de Pe.sca (INAPE), Montevideo, Uruguay 1(11:1-8. Arkhipkin: Intrapopulation structure of ///ex argentinus during its feeding period over the Patagonian Shelf 13 Lipinski, M. R., E. G. Dawe, and Y. Natsukari. 1991. Practical procedures of squid ageing using statoliths. A lab- oratorv- manual. Introduction. In P. Jereb, S. Ragonese, and S. von Boletzky (eds.). Squid age determination using statoliths, p. 77-81. Mazara del Vallo, N.T.R.-I.T.PP Special publication 1. Nesis, K. N. 1987. Cephalopods of the world. T.F.H. Publications, Nep- tune City, NJ, 351 p. Nigmatullin, Ch. M. 1989a. Las especies del calamar mas abundantes del Atlan- tico sudoeste y sinopsis sobre ecologia del calamar Illex argen- tinus. Frente Maritime 5 (sect. A):7 — 81. 1989b. Squid of the world ocean. In V. V. Ivchenko (ed.), Development of the fishery in the open oceanic waters, p. 26-48. Kaliningrad Book Press, Kaliningrad. [In Russian]. Nigmatullin, Ch. M., and V. V. Laptikhovsky. 1996. Soviet/Russian fishery of the Argentine squid (///c.v argen- tinus) in the Southwest Atlantic, its ecological fishing periodicity and probable reasons of anomalous fishery conditions in 1993-1995. In V A. Sushin (ed. ), Fishery and biological research carried out by AtlantNIRO in 1994-1995, p. 13S-156. Atlant- NIRO Press, Kaliningrad. [In Russian with English summary] . Parfeniuk, A. V., Yu. M. Froerman, and A. N. Golub. 1992. Particularidades de la distribucion de los juveniles de Itlex argentinus en el area de la depresion argentina. Frente Maritimo 12 (sect. A):105-lll. Rodhouse, P. G., and E. M. C. Hatfield. 1990. Dynamics of growth and maturation in the cephalopod Illex argentinus de Castellanos, 1960 (Tfeuthoidea, Omma- strephidae). Phil. Trans. Roy. Soc. Lond. (sen B) 329:229- 241. Santos, R. A,, and M. Haimovici. 1997. Reproductive biology of winter-spring spawners o{ Illex argentinus (Cephalopoda: Ommastrephidae) off southern Brazil. Scient. Man 61(1 ):53-64. Sato, T., and H. Hatanaka. 1983. A review of assessment of Japanese distant-water fish- eries for cephalopods. In J. F. Caddy (ed.). Advances in assessment of world cephalopod resources, p. 14—180. FAO Fish. Tech. Pap. 231. Tsygankov, V. Y. 1987. Squid polymorphous protein systems and an analysis of intraspecific differentiation of Illex argentinus (Castella- nos). In Genetic investigations of marine hydrobionts, p. 243-251. Annals of the III All-Union symposium. Moscow, VNIRO. [In Russian]. Uoziuni, Y., and C. Shiba. 1993. Growth and age composition of Illex argentinus (Ceph- alopoda: Oegopsida) based on daily increment counts in statoliths. In T. Okutani, R. K. O'Dor, and T. Kubodera (eds.), Recent advances in cephalopod fishery biology, p. 591-605. Tbkai Univ Press, TDkyo. 14 Abstract.— Microsatellite DNA varia- tion at six microsatellite loci (Omy77, Ots3, OtslOO, OtslOS, Otsl07, and OtslOS) was examined in approximately 900 sockeye salmon, Oncorhynchus nerka, collected between 1987 and 1995 from three stocks on the west coast of Van- couver Island, British Columbia, Canada. Variation in allele frequencies among stocks was, on average, about 12 times greater than temporal variation within stocks. Individual locus Fgj estimates ranged from 0.013 to 0. 107 among stocks, with an overall value of 0.056. Analysis of simulated mixed-stock samples indi- cated that data from four to six of the microsatellite loci surveyed would enable relatively accurate and precise estimates of stock composition for mixtures com- posed offish from the three stocks. Appli- cation of the mixture analysis to 1100 fish sampled in Barkley Sound and Albemi Inlet fisheries during 1997 indicated that sockeye salmon from Great Central Lake constituted about 70% of the commercial catch. The later time of return of sock- eye salmon from Henderson Lake than of those from Great Central or Sproat Lake as previously indicated by analysis of parasite frequencies was confirmed in the 1997 fishery sampling. Stock com- position of catches varied among gears, presumably owing to gear selectivity. Microsatellite DNA variation and estimation of stock composition of sockeye salmon, Oncorhynchus nerka, in Barkley Sound, British Columbia - Terry D. Beacham Khai D. Le Monique R. Raap Kim Hyatt Wilf Luedke Ruth E. Withler Pacific Biological Station Department of Fishenes and Oceans Nanaimo, British Columbia Canada V9R 5K6 E-mail address (for TD Beacfiam) beachamtig pac.dfo-mpo gc ca Manuscript accepted 27 April 1999. Fish. Bull. 98:14-24(20001. In the sockeye salmon (Oncorhyn- chus nerka ) fishery in Barkley Sound on the west coast of Vancouver Island, three stocks (Sproat Lake, Great Central Lake, and Henderson Lake) account for all of the catch in the mixed-stock fishery (Hyatt and Steer, 1987) (Fig. 1). These stocks have been exploited for over 100 years, but the area of the fishery has changed. The present fishery is conducted over a wide area in Bar- kley Sound. Lake fertilization has been used to increase production of Barkley Sound sockeye salmon ( LeB- rasseur et al., 1978; Hyatt and Stock- ner, 1985). Of the lakes sampled in our study, Great Central Lake has been fertilized most extensively, with annual applications of fertilizer between 1970 and 1973, and from 1977 to the present. Sproat Lake was fertilized between 1985 and 1987, and Henderson Lake has been fertil- ized from 1976 to the present. Assessment of the effects of ferti- lization on the productivity of Great Central and Henderson lakes re- quired accurate and reasonably pre- cise estimates of stock composition in the Barkley Sound sockeye salmon catch. The frequency of occurrence of two myxosporean parasites, Myxo- bolus arcticus in the brain and Hen- neguya salmonicola in the muscle, differed substantially among sock- eye salmon in the three lakes during 1977-84 (Quinn et al., 1987), and these differences in prevalence were used to provide estimates of stock composition in the fishery until 1984 (Steer et al., 1986, 1988). Sockeye salmon from Sproat Lake and Great Central Lake accounted for 95% of the catch from 1980 to 1984 (Hyatt and Steer, 1987). In the 1990s, it became apparent that the frequency of occurrence of the two parasites had changed in Great Central Lake sock- eye salmon (Beacham et al., 1998), and fishery managers no longer con- sidered estimates of stock compo- sition derived from parasites to be reliable for management decisions. The timing of the change in para- site frequency of occurrence between 1984 and the 1990s was unknown, rendering post-1984 estimates of stock composition and associated estimates of individual lake produc- tivity uncertain. It became impera- tive to develop a reliable alternative method of stock identification that could be applied to fishery samples for accurate estimation of both catch and productivity by stock. Beacham et al : Microsatellite DNA variation and estimation of stock composition of Oncorhynchus neika 15 A preliminary survey of DNA variation at microsatellite loci indicated that there was some differentiation among the Bark- ley Sound sockeye salmon stocks (Nelson et al., 1998). Evaluation of alternative methods of stock identification indicated that mixture analysis based on micro- satellite allele frequencies would likely provide reliable estimates of stock compo- sition (Beacham et al., 1998). In the pres- ent study, we expanded the analysis of variation at microsatellite loci of Barkley Sound sockeye salmon to six polymorphic loci, examined the differentiation among and within stocks at each locus, evalu- ated the precision of data and accuracy of stock composition estimates for a range of mixture sample sizes based on data from three to six loci, and finally used the microsatellite variation to estimate stock compositions from 1997 fishery samples. Materials and methods Collection of DNA samples and amplification by PCR Scales were collected ft'om sockeye salmon returning to spawn in the Sproat Lake and Great Central Lake drainages in 1987, 1990, and 1992. Scales were collected from Henderson Lake sockeye salmon in 1988 and 1993, and liver samples preserved in 95% ethanol were collected in 1995. Scales or operculum punches were collected from sockeye salmon sampled in fisheries in 1997. DNA was extracted fi-om scales as outlined by Nelson et al. ( 1998). For the operculum or liver samples, approximately 0.3 g of tissue was placed in each well of a 96-well plate containing 0.2 mL of 5% chelex in TE buffer ( 10 mM Tris pH 7.4, 1 mM EDTA pH 8.0, 0.10 mg/mL proteinase K, and 0.1% SDS) and incubated for 15 min at 50°C, and then incubated for an additional 15 min at 95°C. The supernatant from each well was collected and placed in a fresh 96-well plate and stored at -20°C. About 1 mL of this extract was required for each amplification of the sample by the polymerase chain reaction (PCR). Loci amplified by PCR were the dinucleotide repeats Omy77 and Ots3 and the tetranucleotide repeats OtslOO, Otsl03, Otsl07, and Otsl08 (Table 1). For all primer sets used in this study, PCR was conducted in 25-juL reactions containing 12 pmol (0.48 ^M) of each primer, 80 ^M of each nucleotide, 20 mM Tris-pH ^4a40 — Figure 1 Location of Barkley Sound on Vancouver Island. Sockeye salmon are produced in Great Central Lake and Sproat Lake, both part of the Somass River drainage, as well in Henderson Lake. 8.8, 2 mM MgS04, 10 mM KCl, 0.1% Triton X-100, 10 mM (NH4)S04, and 0.1 mg/mL of nuclease-free bovine serum albumin. Each PCR reaction was pre- ceded by an initial denaturation step of three min at 94°C. All cycle extension (30 cycles for all loci except OtslOS which was 35 cycles) steps were for 60 sec at 72°C and all cycle denaturation steps were for 20 sec at 94 C. PCR of Omy77, Ots3, OtslOO, Otsl03, Otsl07, and OtslOS was accomphshed with anneal- ing temperatures of 48°C, 50°C, 57°C, 55°C, 48°C, and 46°C, respectively. Annealing times were 30 sec for Omy77 and OtslOO, and 60 sec for the other loci. Gel electrophoresis and band analysis PCR products were size fractionated on 16 cm x 17 cm nondenaturing polyacrylamide gels and visualized by 16 Fishery Bulletin 98(1) staining with 0.5 mg/mL ethidium bromide in water and ultraviolet light illumination. Nelson et al. (1998) pro- vide a complete description of gel electrophoretic condi- tions. All gels were run for 14—18 h at 65-70 V, using 8% acrylamide for analysis of OtslOO and Otsl03, and 10% acrylamide for analysis of Omy77, Ots3, Otsl07 and Otsl08. Twenty-nine lanes per gel were loaded. One out- side lane contained a one-kb ladder (Gibco BRL), three lanes contained a 20-bp ladder (Glensura Labs Inc., Del Mar, CA) evenly spaced across the gel, one lane con- tained a standard fish to determine precision of estima- tion of allele size, and 24 lanes contained an individual fish for analysis. Gels were scanned at a 1024 x 1024 pixel density with a Kodak charge coupled device (CCD) camera with low-light capability and a yellow filter. Images were analyzed by using Biolmage Whole Band soft- ware (Genomic Solutions Inc., 1995), where the size of the amplified microsatellite alleles were reported to the nearest base pair (bp) based upon the molecular size grid created with the 20-bp markers. Because some uncertainty occuiTed in estimation of allele size fi-om the 20-bp grid, we identified alleles on the basis of a bimiing procedure ( Gill et al., 1990). Peaks in the allele fi-equencies used to identify main alleles and bin widths generally corresponding to a repeat unit were set so that the main allele was located in the middle of the bin. Precision of estimation of allele size was evalu- ated with the standard fish analyzed for each locus. Data analysis Annual variation in allele fi-equencies within populations was tested with GENEPOP version 3.1 with the Mai-kov- Table 1 Primer sequences for the micnisatellite loci analyzed in the study. Locus Sequence (5-3) Source Omy77 F: CGT TCT CTA CTG AGT CAT R: GOG TCT TTA AGG CTT CAC TGC A Mon-iset all 1996) Ots3 F: CAC ACT CTT TCA GGA G R; AG A ATC ACA ATG GAA G Banltsetal. (19991 OtslOO F: TGA ACA TGA GCT GTG TGA G R: ACG GAC GTG CCA GTG AG NeLsonetal. 11998) OtslO:j F: AGG CTC TGG GTC CGT G R: TGA TAT GGT GTG ATA GCT GG Beacham et al. (1998) OtslOT F: ACA GAC CAG ACC TCA ACA R: ATA GAG ACC TGA ATC GGT A Nelson and Beacham ( 19991 OtslOS F: TCT GTT TAT CTT TCT ATT A R: AAG GAG AGA CAG AGG G Nelson and Beacham ( 1999) Chain approach by using;);'- probability values (Raymond and Rousset, 1995). The dememorization number was set at 1000, and 50 batches were run for each test with 1000 iterations/batch (Raymond and Ptousset, 1995 ). Each stock at each locus was tested for departure from Hardy- Wein- berg equilibrium by using GENEPOP. Gametic Linkage disequihbrium between loci in each population was also evaluated with GENEPOP. Tests of genetic differentiation with three pairwise comparisons among the populations wei-e also conducted with GENEPOP with the Markov- Chain approach by using _;f2 probability values. Critical sig- nificance levels for simultaneous tests were evaluated by using sequential Bonferroni adjustment (Rice, 1989). Fgj. estimates for each locus were calculated with GENEPOP, and the standai'd deviation of the estimate for an individ- ual locus was determined with FSTAT (Goudet, 1995) by jackknifing over stocks and for all loci combined by boot- strapping over loci. Estimation of variance components of stock differences and annual variation within stocks was determined with BIOSYS (SwoSbrd and Selander, 1981). Piincipal components of nine (three annual sam- ples multiplied by three stocks) composite airays of allele frequencies for six loci were calculated with the PRIN- COMP procedure in SAS (SAS, 1989). Estimation of stock composition The effectiveness of using variation at microsatellite loci for the practical assessment of stock composition in mixed-stock fisheries of Barkley Sound was evaluated from the stand points of precision of stock composi- tion data and accuracy of stock composition estimates in simulated fishery samples. Although only three stocks could contribute to the fishei-y samples, we wished to determine the sample size required to detect accurately the relatively small proportion of Henderson Lake sockeye salmon that were expected to be present in most fishery samples. In addi- tion, we wished to examine the effect of the number of loci used in the estimation of stock composi- tion. The simulated mixtures were composed of 30% Sproat Lake fish, 60% Great Central Lake fish, and 10% Henderson Lake fish because these proportions are the approximate long-term mean of the Barkley Sound fishery. Allele fi'equencies wei'e deter- mined for each locus in each stock, and the model of Foumier et al. ( 1984 ) was used to estimate stock composition by the condi- Beacham et a\ ; Microsatellite DNA variation and estimation of stock composition of Oncorhynchus neika 17 tional maximum likelihood method. Baseline genotypic frequencies for each of the three stocks were calcu- lated from the obsei^ved allele fre- quencies under the assumption of Hardy -Weinberg equilibrium. Each baseline stock was resampled with replacement in order to simulate random variation involved in the collection of the baseline samples during the estimation of stock com- position of each mixture. Hypo- thetical fishery samples of 100-300 fish with fixed stock composition were generated by randomly resam- pling with replacement the baseline stocks, and adding the appropriate number of fish from each stock to the mixture. Estimated stock compo- sition of the mixture was then deter- mined, and the whole process was repeated 100 times to estimate the mean and standard deviation of the individual stock composition estimates. Fishery samples In 1997, samples were collected fi-om three commercial gillnet fishery openings in Barkley Sound, a gillnet test fishery, a purse-seine test fishery, the recreational fish- ery, and an aboriginal fishery. The commercial gillnet fishery was conducted primarily in Barkley Sound, with gillnet mesh sizes ranging from 114 mm (4.5 inches) to 133 mm (5.25 inches). The gillnet test fishery was con- ducted farther inland at the head of Barkley Somid and at the mouth of Albemi Inlet with a gill net 110 m (60 fathoms) in length and 180 meshes deep, and having a mesh size of 114 mm. Samples from the purse-seine fishery, the recreational fishery, and the aboriginal fish- eiy were derived entirely from Albemi Inlet. The recre- ational fishery was conducted near the head of Albemi Inlet and the aboriginal fishery, conducted at the head of Albemi Inlet and in the Somass River, was the most terminal fishery. Estimated stock contributions to each sample were determined as a point estimate from all the fish in the sample, and standard deviations of the esti- mates were derived fi'om bootstrap resampling of both the baseline stocks and the mixture. Results Precision of estimation of allele size Standard deviations of the estimated allele sizes for the heterozygous standard fish analyzed at each locus Table 2 Precision of estimates of allele size (in basepairs) at each microsatellite locus for standard fish run only once per eloctrophoretic gel. n is the number of gel 1 on which allele sizes foi a standard fish were estimated. Stan dard deviation is given in paren- | theses. Locus n Allele size Range Allele size Range Ots3 15 74.1(0.35) 74-75 93.1(0.35) 93-94 Omy77 28 94.9(0.63) 94-96 110.3(0.53) 109-111 24 100.4(0.53) 100-101 116.0(0.62) 11.5-117 8 104.0(0.00) 104-104 116.1(0.64) 115-117 Otsl07 46 109.7(0.55) 109-111 117.7(0.48) 117-118 Otsl08 12 112.1(0.67) 111-113 184.6(0.51) 184-185 OtslOO 8 158.3(0.71) 157-159 184.4(1.30) 183-186 26 164.5(0.71) 163-166 181.6(0.64) 180-183 11 1.58.4 (0.50) 158-159 196.5(0.52) 196-197 otsioa 33 175.1(0.601 174-176 213.4(1.00) 211-215 ranged from 0.00 to 1.30 and tended to increase with allele size (Table 2). For both alleles at Ots3, 100% of the estimated sizes for each allele spanned a 2-bp interval. For alleles <110 bp at Omy77, 93% (56/60) of the estimated sizes of the allele were in a 2-bp interval, as were 90% of the estimated sizes of alleles between 110 and 120 bp. Estimated sizes of alleles of the standard fish that were analyzed at the other loci were all estimated within a 4-bp interval for alleles <200 bp, with 85% of the estimated sizes of the larger allele (213 bp) at Otsl03 within a 4-bp interval. Variation within stocks All six microsatellite loci examined were polymorphic for all three stocks. Obsei-ved heterozygosity of the loci examined over all stocks was as follows: Omy77 0.70 (stock range 0.61-0.80), Ots3 0.67 (0.64-0.70), OtslOO 0.75 (0.69-0.79), Otsl03 0.83 (0.81-0.86), Otsl07 0.28 (0.17-0.40), and OtslOS 0.85 (0.80-0.89). Significant departures (correction for three tests per locus, a=0.0167) from the expected Hardy-Weinberg distribution of genotypic frequencies were observed at the Omy77 locus in all stocks, owing, in the case of Sproat and Henderson lakes, to a deficiency of het- erozygotes. A similar significant heterozygote defic- iency was also detected at Otsl08 in Sproat Lake sockeye salmon. Significant annual variation (correc- tion for six tests per stock, a=0.0083) in allele fre- quencies was observed at Omy77 in Sproat Lake and Henderson Lake sockeye salmon, and at Otsl08 in Henderson Lake sockeye salmon. No significant link- age disequilibrium between any pair of loci in any stock was obsei"ved. 18 Fishery Bulletin 98(1) 1.5- # 0.5- ▲ -1.5 -0.5 0.5 1.5 • Sproat AGCL • Henderson • • -1.5 J -' PCI Rgure 2 Plot of the first two principal components incorporating variation at microsatellite loci for Great Central Lake (GCLl, Sproat Lake, and Henderson Lake sockeye salmon sampled in each of three years. Variation among stocks The three sockeye salmon stocks in Barkley Sound were genetically distinct at all six loci examined. All pairwise tests of allele frequencies among stocks were highly significant at all loci iP<0.001). At Omy77, the frequency of the Omy??^"* allele ranged from 0.005 in Henderson Lake sockeye salmon to 0.449 in Sproat Lake fish, and the frequency of Omy77^'^'' ranged from 0.216 in Sproat Lake fish to 0.546 in Henderson Lake fish (Table 3). Substantial differentiation in allelic frequencies among stocks was observed at Ots3. For example, the frequency of OtsS'^'^ in Henderson Lake sockeye salmon was 0.114, whereas in Sproat Lake fish it was 0.543. Similarly, the frequency of Ots3^^ was 0.059 in Sproat Lake fish and 0.243 in Hender- son Lake fish. At OtslOO, the frequency of OtslOOi^s ranged from 0.256 in Sproat Lake sockeye salmon to 0.504 in Henderson Lake fish (Table 3). Although the three stocks were distinct at OtslOS, the allele fre- quency variation was less marked at that locus. At OtslOT, the combined frequency of four alleles (81, 109, 113, 117) was greater than 0.95 in all stocks, but stock differentiation was nonetheless apparent. For example, the frequency of Otsl07'^' was 0.135 in Great Central Lake sockeye salmon, but < 0.010 in the other two stocks (Table 3). Variation in allelic frequencies at OtslOS was evident among stocks, with the frequency of OtslOS^" ranging from 0.000 to 0.196. Strong genetic differentiation among these three stocks was evident at all six microsatellite loci. Comparison of the relative magnitude of differen- tiation among stocks and among samples from the same stock collected in different years showed that differentiation among stocks always exceeded tempo- ral variation within stocks and was on average 12 times greater (Fig. 2; Table 4). At Ots3, the differ- ences among stocks were 235 times greater than the observed annual variability. With data combined over years, individual locus Fg-p estimates ranged from 0.013 to 0. 107, with an overall value of 0.056 (Table 4). The loci displaying the greatest differentiation among stocks were Ots3 and Omy77, whereas Otsl03 dis- played the least differentiation. Sproat Lake and Great Central Lake stocks, both in the same river drainage, were genetically the most similar (pairwise F^^ esti- mate over all loci: 0.032). The Great Central Lake and Henderson Laike stocks were more genetically distinct Beacham et al : Microsatellite DNA variation and estimation of stock composition of Oncorhynchus nerka 19 Table 3 Observed allele frequencies at six microsatellite loci for three stocks of Barkley Sound sockeye salmon. Alleles have been designated by the lower size limit of the bin. n is the number offish scored at each locus in each stock. Allele Sproat Great Central Henderson Allele Sproat Great Central Henderson Omy7 Otsl03 n 264 303 305 n 221 304 308 90 0.000 0.000 0.002 144 0.007 0.021 0.000 92 0.011 0.003 0.002 152 0.043 0.026 0.000 94 0.449 0.172 0.005 156 0.027 0.002 0.002 96 0.042 0.020 0.000 160 0.009 0.018 0.003 98 0.006 0.007 0.000 164 0.034 0.008 0.013 100 0.125 0.231 0.277 168 0.016 0.015 0.006 102 0.011 0.015 0.030 172 0.038 0.033 0.019 104 0.216 0.361 0.546 176 0.020 0.025 0.034 106 0.002 0.002 0.020 180 0.029 0.064 0.080 108 0.040 0.005 0.000 184 0.032 0.076 0.057 110 0.051 0.054 0.100 188 0.050 0.178 0.073 112 0.002 0.003 0.002 192 0.305 0.268 0.312 114 0.004 0.086 0.008 196 0.269 0.151 0.185 116 0.038 0.040 0.010 200 0.075 0,079 0.130 118 0,004 0.002 0.000 204 0.043 0,028 0.071 Ots3 208 0.002 0,008 0.011 n 219 296 303 212 0.000 0,000 0.003 74 0.112 0.127 0.086 Otsl07 78 0.000 0.000 0.003 n 269 307 310 80 0.000 0.000 0,002 81 0.006 0,135 0.000 82 0.000 0.003 0.005 101 0.000 0,002 0.000 84 0.002 0.000 0.000 105 0.000 0.000 0.005 86 0.000 0.002 0.000 109 0.084 0.062 0.053 88 0.543 0.378 0.114 113 0.866 0.762 0.913 92 0.002 0.007 0.000 117 0.030 0.036 0.029 93 0.274 0.394 0.526 121 0.015 0.003 0.000 96 0.000 0.002 0.005 OtslOS 97 0.007 0.005 0.013 n 214 199 269 99 0.059 0.073 0.243 122 0.196 0.106 0.000 103 0.000 0.005 0.000 126 0.014 0.035 0.004 105 0.000 0.005 0.003 130 0.002 0.000 0.000 OtslOO 133 0.002 0.000 0.000 n 242 321 282 137 0.000 0.128 0.002 130 0.010 0.003 0.000 141 0.002 0.008 0.002 134 0.039 0.006 0.000 145 0.007 0.010 0.000 138 0.000 0.002 0.000 149 0.121 0.038 0.035 142 0.006 0.000 0.002 153 0.056 0.098 0.007 150 0.008 0.002 0.004 156 0.136 0.095 0.178 154 0.025 0.037 0.025 160 0.086 0.146 0.229 158 0.256 0.364 0.504 164 0.189 0.163 0.158 162 0.169 0.064 0.133 168 0.042 0.070 0.048 166 0.052 0.047 0.064 172 0.035 0.038 0,043 170 0.002 0.003 0.007 177 0.068 0.050 0.216 174 0.002 0.006 0.004 182 0.016 0.010 0.039 179 0.324 0.202 0.193 187 0.019 0.005 0,035 184 0.085 0.115 0.050 192 0.005 0.000 0,004 190 0.008 0.107 0.007 197 0.002 0.000 0,000 195 0.012 0.039 0.009 200 0.000 0.002 0.000 20 Fishery Bulletin 98(1) (pairwise F^^j, estimate: 0.042), and the Sproat Lake and Henderson Lake stocks showed the greatest dif- ferentiation (pairwise Fgj. estimate; 0.091). Estimation of stock composition The three loci with the highest Fgj< estimates (Omy77, Ots3, and Otsl07) also possessed the highest ratio of Table 4 F^-j. estimates and the ratio of the variance components attri- | butable to among and within stock differentiation ( over time ) for six microsateUite loci of Barkley Sound sockeye salmon. Standard deviation of F^j. estimates is given in parentheses. Locus FsT Variance ratio Omy77 0.107(0.0941 20.3 Ots3 0.099(0.0891 235.2 OtslOO 0.027(0.013) 5.8 Otsl03 0.013(0.007) 3.9 OtslOV 0.043 (0.035) 6.3 OtslOS 0.039(0.018) 3.6 All 0.056 (0.032) 11.8 Table 5 Average estimated stock composition C/r ) of simulated mixtures of Barkley Sound sock- eye salmon based on variation at three to six microsateUite loci. True mixture percent- ages were as follows: Sproat Lake 30*7^, Great Central Lake 60'7r. and Henderson Lake 10'7( . The three loci initially used to estimate the percentages of the mixtures were Omy77, Ots3, and Otsl07. and OtslOO, Otsl08, and Otsl03 were added sequentially. Each mixture was generated 100 times with replacement, and stock compositions of the mixtures were estimated by resampling each baseline stock with replacement to obtain a new distribution of allele frequencies, with the same sample size in the new distribu- tion as in the original one. Standard deviation is given in parentheses. Mixture size (no. offish) Source and loci 100 150 200 Sproat Lake 3 4 5 6 30.7 (9.98) 30.7(8.21) 30.1(7.54) 31.2(7.23) Great Central Lake 3 4 5 6 Henderson Lake 3 4 5 6 60.5(11.74) 60.4(9.41) 60.1(8.26) 58.8(8.03) 8.8(5.62) 8.8(5.30) 9.8 (4.56) 10.0(4.33) 31.1(8.51) 30.8(6.87) 30.6(5.87) 29.0(5.27) 60.1(9.27) ,59.8(7.40) 60.0(6.63) 60.9(5.96) 8.8(4.84) 9.3 (3..53) 9.5 (3.73) 10.1(3.70) 30.8(6.81) 30.8(6.38) 30.0(4.44) 29.8 (4.42) 60.2(7.40) .59.7 (6.88) 60.1(5.11) 60.1(5.09) 9.0(3.89) 9.6(3.52) 9.9(2.76) 10.0(2.81) spatial to temporal variation (Table 4) and were there- fore selected to form the core database for the analysis of the simulated mixtures. The number of loci used in the determination of stock composition or mixture sample size had little effect upon the accuracy of the estimated stock compositions (Table 5). Precision of the estimated stock compositions increased as both the number of loci and mixture size used in the deter- mination increased. However, different options were available to obtain estimates of a desired precision. For example, higher levels of precision were obtained with four loci (Omy77, Ots3, Otsl07, OtslOO) in con- junction with a 150-fish sample size (600 units of data) than with all six loci and a 100-fish sample (600 units of data) (Table 5). The coefficient of varia- tion for the estimated proportion of the predominant Great Central Lake stock was always less than that for estimated proportions of the other two stocks in the mixture. For the 4 loci in 150-fish mixture analysis, the coefficient of variation for the estimated propor- tion of Great Central Lake fish was 12%, whereas it was 22'^ for Sproat Lake fish, and 38% for Henderson Lake fish. The simulations indicated that fewer than six microsateUite loci could be used to provide rea- sonably precise data and accurate estimates of sock- eye salmon stock composition for Barkley Sound fishery samples. Conditional maximum likeli- hood estimation can overesti- mate the relative abundance of rare stocks. For Barkley Sound, this would likely be the Hender- son Lake stock. The precision of data and accuracy of stock com- positions estimates were inves- tigated for mixture samples of 100 fish composed of 2% Hen- derson Lake (38% Sproat Lake, 60% Great Central Lake) and 5% Henderson Lake (35% Sproat Lake, 60% Great Central), where stock compositions were esti- mated by using the four mic- rosateUite loci (Omy77, Ots3, OtslOO, and Otsl07) generally used for estimation of stock compositions in the 1997 fish- ery samples. Estimated stock compositions of the simulated 100 mixtures for the 2% Hender- son Lake composition were 2.5% (SD=3.1%) Henderson Lake, 34.3% (SD=9.3%) Sproat Lake, and 60.4% (SD=9.3% ) Great Cen- tral Lake. For the 5% Hender- 300 30.7 (6.48) 30.2(4.44) 29.4(4.18) 29.8(4.00) 59.7(7.74) 60.7(5.21) 60.8(4.50) 59.9(4.31) 9.6(3.98) 9.1 (2.89) 9.8(2.52) 10.3(2.39) Beacham et al : Microsatellite DNA variation and estimation of stocl< composition of Oncorhynchus nerka 21 son Lake composition, estimated stock compositions were 5.39c (SD=3.9'7f) Henderson Lake, 34.3% (SD=8.2%) Sproat Lake, and 60A7c{SB=8.97c) Great Cen- tral Lake. No significant bias was observed when Henderson Lake sockeye salmon composed 57c or less of the mixture. Application of estimates to 1997 fisheries Although estimated stock con- tributions varied according to sampling period, sockeye salmon fi-om Great Central Lake tended to predominate in all fisheries at any week (Table 6). However, differences in stock composition estimates among fishing gears were evident. In the commer- cial gillnet fishery, Great Central Lake sockeye salmon constituted about 70% of the catch (Table 6). In the gillnet test fishery, the proportion of Great Central Lake sockeye generally varied between 55 and 75% prior to July 25th. In the purse-seine test fishery, they accounted for about 50-55% of the catch. Higher proportions of Great Central Lake sockeye salmon were observed in the selective gillnet gear than in the more nonselective purse-seine gear. For example, for the week ending 4 July, Great Central Lake sockeye were estimated to have represented 70-757f of the catch in the commer- cial gillnet fishery and in the gillnet test fishery, but only about 40% of the catch in the seine test fishery. Although the samples analyzed from the purse-seine fishery were derived fi-om more inland locations than those from the commercial and test gillnet fisheries, the differences in proportions of Great Central sockeye salmon more likely resulted from differences in gear selectivity than ft-om differences in stock distribution because fish fi-om all three stocks are generally dis- tributed throughout Barkley Sound and Albemi Inlet when present. Sockeye salmon stock ft-om Henderson Lake are the smallest salmon exploited in the fishery, and thus the most vulnerable to overfishing in the mixed-stock har- vest that takes place. Henderson Lake fish, which do not have to travel through Albemi Inlet in their spawn- ing migration, were apparently caught in fisheries throughout Albemi Inlet, although there was a high degree of uncertainty about whether they were caught in the aboriginal fishery at the extreme head of Albemi Table 6 Estimated stock compositions (7r)for sockeye salmon from three lakes sampled in gill- net test fisheries, seine test fis heries. commercial fishery openings, a native fishery, and 1 recreational fishery in Barkley Sound during 1997. Four loci (Ots3, OtslOO, Otsl07, and Omy77) were used to estimate stock composition. n is the number offish analyzed, and standard deviation of the estimates is given in parentheses. Source Week ending n Sproat Great Central Henderson Commercial 4 Jul 118 25.8 (7.3) 73.9(8.2) 0.3 (2.8) Commercial' 11 Jul 95 22.3(6.2) 71.1(7.9) 6.2 (3.8) Commercial' 18 Jul 95 29.7(6.9) 63.0 (7.8) 7.3 (4.8) Seine 27 Jun 120 35.9(7.3) 56.6 (8.0) 7.5(3.4) Seine 4 Jul 111 48.4(8.6) 42.2(10.4) 9.0(5.4) Seine 18 Jul 117 37.3(7.0) 59.6(7.8) 3.1(4.6) Gillnet 20 Jun 50 34.4(10.3) 65.6(10.4) 0,0(0.9) Gillnet 27 Jun 50 36.2(10.9) 51.0(12.9) 12.9 (6.8) Gillnet 4 Jul 50 21.2(9.2) 71.9(10.5) 6.7(5.1) Gillnet 11 Jul 50 33.1(9.7) 54.1(11.7) 12.8(7.7) Gillnet 18 Jul 50 24.5(11.8) 72.2(13.6) 3.4(6.3) Gillnet 25 Jul 50 19.9(11.5) 60.0(13.8) 20.2(9.1) Gillnet 1 Aug 50 28.2(11.6) 42.7(13.7) 29.1(10.3) Aboriginal 18 Jul 86 45.0 (8.4) 52.8(8.5) 2.2(2.3) Recreational 18 Jul 33 33.5(12.3) 54.8(13.3) 11.7 (8.4) ' Additional loci Otsl03 and OtslOS, were used in estimation of stock composition. Inlet (Table 6). Henderson Lake sockeye salmon gen- erally represent 10% or less of the catch, except for sockeye salmon sampled after 18 July in the gillnet test fishery, when the relative abundance of Hender- son Lake sockeye salmon substantially increased. By late July, Henderson Lake sockeye salmon constituted nearly 30% of the gillnet test fishery sample. Discussion DNA variation at microsatellite loci is becoming an increasingly important tool in fisheries research and management (see review by O'Connell and Wright [1997]). In salmonids, microsatellite loci are gener- ally characterized by high levels of variability and dif- ferentiation among spawning populations (Angers et al., 1995; McConnell et al., 1997; Seeb et al., 1998), even in very localized areas (Beacham and Dempson, 1998). The feasibility of applying biological markers to salmon stock identification is enhanced when they display limited annual variation. With temporal sta- bility of the discriminating characters, annual surveys of contributing populations are unncecssary once they have been adequately characterized. As for other neu- tral genetic markers (Wood et al., 1994; Beacham et al., 1996), temporal stability of allele frequencies at microsatellite loci has generally been observed in sal- monid populations (Small et al., 1998). For popula- 22 Fishery Bulletin 98(1) tions in which annual variation has been detected, the magnitude of variation has been substantially less than that among populations (Nielsen et al., 1997; Beacham and Wood, 1999). For sockeye salmon, in which the greatest geo- graphic determinant of neutral genetic differentiation is the nursery lake (Wood, 1995), the task of identify- ing the contributions of three different lake systems to a mixed-stock sample should be relatively straightfor- ward. Although significant genetic variation can occur among spawning sockeye salmon subpopulations iso- lated by time or space (or both) within a lake system, the extent of this variation is consistently much less than that observed among lakes — even those lakes within a single drainage system (Wood, 1995). Each of the three lakes is the confluence of multiple tripu- taries and may harbor genetically differentiated sub- populations of sockeye salmon. The spawning ground samples in our study were collected from locations within each lake system at which fish ft-om more than one subpopulation may have been present, and dif- ferent subpopulations may have been sampled among years. Thus, the departure of Omy77 (and OtslOS for Henderson Lake) genotypes fi-om Hardy- Weinberg equilibrium and significant annual variation observed at these loci might both have reflected subpopulation dif- ferentiation in allele fi-equencies. It is unlikely that the heterozygote deficiency observed at Omy77 in Sproat Lake and Henderson Lake sockeye salmon would be a result of a null allele because genotypic fi-equencies of other sockeye salmon stocks surveyed at this locus have been in Hardy- Weinberg equilibrium (Beacham and Wood, 1999). Nevertheless, the level of differentia- tion at Omy77 was about 20 times greater among lakes than was the temporal variation observed within lakes. For all six microsateUite loci surveyed, differences among lakes were on average 12 times greater than variation within populations, confirming the relative stability of the microsateUite loci in Barkley Sound sockeye salmon populations over the 5—8 yr sampling period. The six microsateUite loci used in the current study were also surveyed in nine sockeye salmon stocks of the Nass River drainage in northern British Columbia (Beacham and Wood, 1999). In the Nass River, the three loci displaying the greatest differen- tiation among stocks were OtslOO (Fgj^O.131), Ots3 (FsT^O.lll), and OtslOS (Fs7^0.084), whereas in the Barkley Sound stocks, the three most discriminating loci were Omy77 {Fg.,^0.W7), Ots3 (^^7^0.099), and Otsl07 (Fgj^b.043). The fact that loci differed in their relative levels of variation between the two areas is not surprising given the rapid evolution of mic- rosateUite loci and the likelihood that the regions were founded postglacially by different sockeye salmon "races" (Wood, 1995). For stock identification applica- tions, surveys of microsateUite variation in each geo- graphic region of interest will generally be necessary to determine which loci are the most effective in dif- ferentiating local populations. Effective assessment and management of sockeye salmorf production in Barkley Sound is dependent upon determination of stock composition in fishery catches. Previous evaluation has indicated that the application of microsateUite technology to stock iden- tification can provide the most reliable and cost-effec- tive results (Beacham et al., 1998), but determination of the feasibility of such technology for Barkley Sound fisheries awaited examination of the relation between the number of loci used, the sample size of the stock mixture to be analyzed, and the precision of the esti- mated stock contributions. For any stock identification application, the optimal combination of number of loci surveyed and number of fish sampled from the catch is dependent on the genetic distance among stocks, the desired precision for an individual stock estimate, and the cost of the analysis for each locus. The simulated mixtures evaluated for Barkley Sound sockeye salmon indicated that microsateUite variation could be used to provide accurate and rea- sonably precise estimates of individual stocks in the catch mixtures. They ftirther indicated that although genotjqaic fi-equencies at Omy77 and Otsl08 were not in Hardy-Weinberg equilibrium in some stocks, but assumed to be so in the stock composition estimation procedure, the violation of this assumption did not have a marked influence on the accuracy of the esti- mated stock compositions. The precision, but not accu- racy, of the estimated contributions increased with both the number of loci (from 3 to 6) and the sample size of the mixture (fi"om 100 to 300). For sample sizes of 150 fish and larger, a greater increase in preci- sion for stock contribution estimates could always be achieved by increasing the number of loci surveyed to six than by increasing the sample size to 300. How- ever, these simulations did not include estimation of the random error associated with sampling only a por- tion of the catch, and this error will always be reduced by increasing sample size. The level of precision of an estimated stock contribution increased with the con- tribution of the stock to the mixture. For estimation of the more abundant Great Central and Sproat sockeye salmon, the increase in precision afforded by additional data was approximately equivalent whether more fish (beyond 150) or more loci were analyzed (i.e. approxi- mately equaUy precise stock contribution estimates were achieved by analyzing four loci in 300 fish and six loci in 200 fish). However, estimation of the small {I07i) Hen- derson Lake contribution to the mixture was more sensi- tive to sample size and was more precise in the analysis of four loci in 200 fish than of six loci in 150 fish. Beacham et al : Microsatellite DNA variation and estimation of stock composition of Oncorhynchus nerka 23 Successfial application of microsatellite loci to esti- mation of stock composition in mixed-stock fisheries requires that loci be chosen that highlight differences among stocks to be separated and that adequate num- bers of fish in the baseline stocks be surveyed to pro- vide reliable estimates of allele fi-equencies, and thus genotypic fi-equencies used in the conditional maxi- mum likelihood analysis. Microsatellite loci can con- tain a large number of alleles, and baseline sample sizes need to be of sufficient size to ensure that alleles present in fish fi-om a stock in the mixture have also been observed in the baseline samples. Binning low- frequency similar-size alleles (Small et al., 1998) is also a strategy to consider in practical applications. Although simulated mixtures can provide insights into the expected performance of the mixture analy- sis, the stock contribution estimates for actual fishery samples can only be evaluated by corroboration with data fi-om other sources. Two supportive sources of independent information occur: time of return of the Henderson Lake stock and the typical catch composi- tion for Barkley Sound that was previously derived from parasites. In Barkley Sound, the time of return of Henderson Lake sockeye salmon has been reported to be later than that of either Sproat Lake or Great Central Lake fish (Steer et al., 1988). For example, in 1984, Henderson Lake sockeye salmon were evident, on the basis of parasite analysis, in the commercial fishery prior to 27 June but increased in relative abun- dance after that time. The current analysis indicated that Henderson Lake sockeye salmon were absent from, or at low abundance in, the 1997 commercial fishery prior to the week of 4 July. Analysis of the gillnet test fishery and purse-seine samples indicated that the proportion of Henderson Lake sockeye salmon in those catches was low until mid-July but thereafter was substantial, consistent with a later time of arrival of the Henderson stock in Barkley Sound. In a typi- cal return year, about 60% of the Barkley Sound sock- eye salmon catch is derived fi-om Great Central Lake, 30% fi-om Sproat Lake, and 10% fi-om Henderson Lake (Steer et al., 1988). Estimated stock compositions for the 1997 fishery catches are in reasonable agreement with the expected stock contributions. Results of the simulation analysis indicated that more precise, but not necessarily more accurate, estimates of the stock contributions (especially that from Henderson Lake) could have been obtained for the fishery catches if sample sizes had been larger than 50 (for the gillnet test fisheries) or approximately 100 (for the purse- seine test and commercial gillnet fisheries). Differences in estimated stock composition were obtained for the purse-seine and gillnet test fisheries in July samples, where higher proportions of Great Central Lake sockeye salmon were observed in the gillnet fishery samples. Although the fishery samples came ft-om different areas (the purse-seine samples were collected farther inland in Albemi Inlet than were the gillnet samples), the most likely explana- tion of the difference in estimated proportions of stock composition between the two gears is a difference in size selectivity. Sockeye salmon caught in purse seines in Barkley Sound are generally more variable in size and of smaller mean size than those caught in gill nets (Steer et al., 1986). Probably gill nets were more selective for Great Central Lake sockeye salmon than for Sproat Lake salmon. Thus, it is important to estimate stock con- tributions to a fishery catch based on samples collected with the type of gear employed in the fishery. Further- more, the analysis of catch samples to estimate the stock composition of fish present in an area (as opposed to those caught in an area) vriH be biased to the degree that the sampling gear nonrandomly catches the fish that are present. The results of this study and other analyses (Beacham et al., unpubl. data) indicate that for salmo- nids, different gear types sample the various stocks in a stock mixture with very different efficiencies. Differentiation among local spawning populations that are relatively stable over time provides the basis for applying biological markers to problems of salmo- nid fisheries management. This study confirmed our expectation that the level of differentiation observed at microsatellite loci among sockeye salmon of the three major lake systems draining into Barkley Sound is sufficient and stable to assess stock composition of the fishery catches. The abundance of highly polymor- phic microsatellite loci in salmonid fish, the relative ease of nonlethal sample collection, and the moderate cost per fish for laboratory analysis combine to provide a technology that will become increasingly used in the assessment and management of salmonid fisheries. Acknowledgments We would like to acknowledge all those involved in sampling collections irom sockeye salmon spawning grounds. The collection of test fishery samples was supervised by Bruce Patten and Jim Mitchell, and scale samples from the 1997 fisheries were provided to us by D. Gillespie of the Ageing Laboratory at the Pacific Biological Station. J. Candy assisted in some of the data analysis and figure preparation. Funding was provided by the Department of Fisheries and Oceans. Literature cited Angers, B., L. Bematchez, A. Angers, and L. Desgroseillers. 1995. Specific microsatellite loci for brook charr reveal strong 24 Fishery Bulletin 98(1) population subdivision on a microgeographic scale. J. Fish. Biol. 47(suppl.A):177-185. Banks, Mj\., M. S. Blouin, B. A. Baldwin, V. K. Rashbrook, H. A Fitzgerald, S. M. Blakenship, and D. Hedgecock. 1999. Isolation and inheritance of novel microsatellites in chinook salmon (Oncorhynchus tshawytscha). J. Heredity 90:281-288. Beacham, T. D., and J. B. Dempson. 1998. Population structure of Atlantic salmon from the Conne River, Newfoundland as determined from microsatellite DNA. J. Fish Biol. 52:665-676. Beacham, T. D., L. Margolis, and R. J. Nelson. 1998. A comparison of methods of stock identification for sockeye salmon Mncorhynchus /lerka) in BarkJey Sound, British Columbia. North. Pac. Anad. Fish. Comm. Bull. 1:227-239. Beacham, T. D., R, E. Withler, and T. A. Stevens. 1996. Stock identification of chinook salmon (.Oncorhynchus tshawytscha) using minisatellite variation. Can. J. Fish. Aquat. Sci. 53:380-394. Beacham, T. D., and C. C. Wood. 1999. Application of microsatellite variation to estimation of stock composition and escapement of Nass River sockeye salmon iOncorhynchus nerka). Can. J. Fish. Aquat. Sci. 56:297-310. Foumier, D. A, T. D. Beacham, B. E. Riddell, and C. A. Busack. 1984. Estimating stock composition in mixed stock fisheries using morphometric, meristic, and electrophoretic character- istics. Can. J. Fish. Aquat. Sci. 41:400-408. Genomic Solutions Inc. 1995. Biolmage Whole Band analyzer Genomic Solutions, Inc.. Ann Arbor, MI, 68 p. Gill, P., K. Sullivan, and D. J. Werrett. 1990. The analysis of hypervariable profiles: problems associ- ated with the objective determination of the probability of a match. Hum. Genet. 85:75-79. Goudet, J. 1995. FSTAT A program for IBM PC compatibles to calculate Weir and Cockerham's (1984) estimators of F-statistics (ver- sion 1.2). J. Heredity 86:485-486. Hyatt, K. D., and G. J. Steer. 1987. Barkley Sound sockeye salmon, iOncorhynchus nerka): evidence for over a century of successful stock development, fisheries management, research and enhancement efforts. In H. D. Smith, L. Margolis, and C. C. Wood (eds.) Sockeye salmon (Oncorhynchus nerka) population biology and future management, p. 435-457. Can. Spec. Publ. Fish. Aquat. Sci. 96. Hyatt, K. D.. and J. G. Stockner. 1985. Responses of sockeye salmon iOncorhynchus nerka) to fertilization of British Columbia coastal lakes. Can. J. Fish. Aquat. Sci. 42:320-331. LeBrasseur, R. J., C. D. McAllister, W. E. Barraclough, O. D. Kennedy, J. Manzer, D. Robinson, and K. Stephens. 1^78. Enhancement of sockeye salmon (Oncorhynchus nerka) by lake fertilization in Great Central Lake: Summary Report. J. Fish. Res. Board Can. 35:1580-1596. McConnell, S. K. J., D. E. Ruzzante, P. T. O'Reilly, L. Hamilton, and J. M. Wright. 1997. Microsatellite loci reveal highly significant genetic dif- ferentiation among Atlantic salmon iSalnio salar L.l stocks from the ea.st coast of Canada. Mol Kcol. 6:1075-1089. Morris, D. B., K. R. Richard, and J. M. Wright. 1996. Microsatellites from rainbow trout (Oncorhynchus mykiss) and their use for genetic study of salmonids. Can. J. Fish. Aquat. Sci. 53:120-126. Nelson, R. J., and T. D. Beacham 1999. Isolation and cross species amplification of microsatel- lite loci useful for study of Pacific salmon. Anim. Genet. 30:228-229. Nelson, -R. J., T. D. Beacham, and M. P. Small. 1998. Microsatellite analysis of the population structure of a Vancouver Island sockeye salmon (Oncorhynchus ner^a) stock complex using non-denaturing gel electrophoresis. Mol. Mar Biol. Biotech. 7:312-319. Nielsen, E. E., M. M. Hansen, and V. Loeschcke. 1997. Analysis of microsatellite from old scale samples of Atlantic salmon Salmo salar: a comparison of genetic compo- sition over 60 years. Mol. Ecol. 6:48-492. O'Connell, M., and J. M. Wright. 1997. Microsatellite in fishes. Rev. Fish Biol, and Fisheries 7:331-363. Quinn, T. P., C. C. Wood, L. Margolis, B. E. RiddeU, and K. D. Hyatt. 1987. Homing in wild sockeye salmon (Oncorhynchus nerka) as inferred from differences in parasite prevalence and alio zyme allele fi-equencies. Can. J. Fish. Aquat. Sci. 44:196- 1971. Raymond, M., and F. Rousset. 1995. GENEPOP (vers. 1.2): population genetics software for exact tests and ecumenism. Heredity 86:248-249. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223- 225. SAS Institute Inc. 1989. SAS/STAT users guide, version 6, 4th ed., vol. 1. SAS Institute, Cary, NC, 890 p. Seeb, J. E., C. Habicht, J. B. Olsen, P. Bentzen, J. B. Shaklee, and L. W. Seeb. 1998. Allozyme, mtDNA, and microsatellite variants describe structure of populations of pink and sockeye salmon in Alaska. North Pac. Anad. Fish. Comm. Bull. 1:.300-318. SmaU, M. P., T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998. Discriminating coho salmon (Oncorhynchus kisutch) populations within the Eraser River, British Columbia using microsatellite markers. Mol. Ecol. 7:141-155. Steer, G. J., N. B. F. Cousens, H. W. Stiff, and K. D. Hyatt. 1986. An analysis of gear selectivity and sources of bias in esti- mates of age and stock composition of the 1980-1984 Barkley Sound sockeye salmon (Oncorhynchus nerka) catch. Can. Tbch. Rep. Fish. Aquat. Sci. 1445, 77 p. Steer, G. J., N. B. F. Cousens, H. W. Stiff, K. D. Hyatt, and D. W. Welch. 1988. A description of the 1984 fishery, stock composition, and biological characteristics of sockeye salmon (Oncorhyn- chus nerka ) in the catch from Area 23, Barkley Sound. Can. Tech. Rep. Fish. Aquat. Sci. 1667, 78 p. Swofford, D. L., and R. B. Selander. 1981. BIOSYS-1: a FORTRAN program for the comprehen- sive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72:281-283. Wood, C. C. 1995. Life history variation and population structure in sock- eye salmon. Am. Fish. Soc. Symposium 17:195-216. Wood, C. C, B. E. Riddell, D. T. Rutherford, and R. E. Withler. 1994. Biochemical genetic survey of sockeye salmon (Oncor/jyn- chus nerka) in Canada. Can. J.Fish.Aquat. Sci.51(suppl. 1): 114-131. 25 Abstract.-From 1990 to 1996, during a large-scale tag-and-release program in the Great Australian Bight, 20,204 southern bluefin tuna (SBT), Thiinnus maccoyii. were injected with strontium chloride (SrCl,^). The objectives of the marking experiment were to examine the efficacy of SrClo as an otolith marker and to determine the periodicity of increment formation in SBT otoliths. Nine-himdred and sixty-one Sr-injected fish were recaptured and 616 otoliths were sampled from these: the high level of sampling success was attributable to a major liaison effort throughout the multinational SBT fishery. The same tag return rates for fish that were tagged and injected and for fish that were tagged only, indicated that the injection of strontium did not affect the survival rate of tagged fish. Strontium marks were detected with a Robinson detector or an energy dispersive spectrometer ( EDS ) ( or with both) linked to a scanning electron microscope; 59 of the 67 otoliths from injected fish examined had discernible marks. The intensity of the strontium mark and the dosage rates were linked; a dosage of 100 mg Sr/kg fish weight is recommended to ensure easy identi fication of the strontium mark. Using the strontium marks, we established that in SBT with 1 to 6 increments in their otoliths, one increment is laid down per year at liberty. In the 59 marked fish that were examined, there was 100'7f agreement between the ex- pected and observed number of incre- ments after marking. These results, and the data fi-om two supplementary tag returns ft-om unmarked fish recaptured after long times at liberty, provide un- ambiguous evidence that increments on the otoliths of SBT are formed annually, to at least the age of 13 years. In addition, a recent study that used bomb-radiocarbon levels to estimate ages of older SBT has provided strong evidence that annual increments are deposited in the sagittal otoliths of SBT throughout life. Direct validation of annual increments in the otoliths of juvenile southern bluefin tuna, Thunnus maccoyii, by means of a large-scale mark-recapture experiment with strontium chloride Naomi P. Clear John S. Gunn Anthony J. Rees CSIRO Marine Research GPO Box 1538 Hobart, Tasmania, 7001, Australia Email address (for N P Gear) Naomi ClearaTnanneairoau Manuscript accepted 11 January 1999. Fish. Bull. 98:25-40 (2000). Southern bluefin tuna (SBT), Thu?2- nus maccoyii Castelnau, 1872, is a large, long-lived, migratory, pelagic fish with a circumglobal distribution between 30°S and 50°S (Caton, 1991). Its only known spawning ground is in the Indian Ocean south of Java, between 7°S and 20°S (Caton, 1991). Since it was first exploited in the 1950s the stock has declined dramatically to between 16% and 25% of its initial level (SainsburyM. The fishery is currently managed by individual transferable quotas (ITQ) and the total allowable catch (TAG) is assessed each year. Vutual population analysis (VPA) has been the main method of as- sessing the condition of SBT stock since 1980 (Murphy and Majkowski, 1981). The age structue of the pop- ulation, a major input to VPA, has been estimated by converting lengths and weights to ages, using growth curves derived fi-om tagging data (Hampton, 1991; Polacheck et al.^). lb reduce the unmeasurable uncertainties that the estimates introduce into VPA assess- ments, a validated direct method for determining age was required. In 1992 we began a study to de- velop reUable techniques for deter- mining ages of SBT. Validation of as- signed ages is critical in age estimation studies (Beamish and McFarlane, 1983; Smith, 1992; Secor et al., 1995); therefore a large-scale mark-recapture experiment was initiated to provide the basis for vahdating the age estimates. From the validation study we aimed to confirm the periodicity of the zones that were counted on the hard parts collected from SBT. The overall objective of these two studies was to develop a validated length-at- age key for the entire size range of the SBT population. We present details of the mark-recapture experiment and our evidence that increments in otoliths are formed annually. 1 Sainsbury, K. 1993. What is happening to the southern bluefin stock? In W. White- law and V. Mawson (eds.), Proceedings of the inaugural southern bluefish tuna science-industry-management workshop. Port Lincoln, Australia, p. 5-19. Commonwealth Scientific and Industrial PJesearch Organi- zation (CSIRO I Marine Research, GPO Box 15,38 Hobart, 7001 Australia. 2 Polacheck, T, K. Sainsbury. and N. Klaer. 1995. Assessment of the status of the southern bluefin tuna stock using virtual population analysis— 1995. Paper SBFWS/ 95/17. First scientific meeting of the Com- mission for the Conservation of Southern Bluefin Tuna (CCSBT). Shimizu, Japan, 70 p. Commonwealth Scientific and Industrial Research Organization (CSIRO) Marine Research, GPO Box 1538, Hobart, Tasmania, 7001, Australia. 26 Fishery Bulletin 98(1) Previous attempts to estimate SBT ages directly, either did not attempt validation, or attempted it for only a few age classes. Hynd (1965) used scales to estimate ages of fish up to 80 cm fork length (FL) but did not attempt to validate his age estimates. Yukinawa (1970) counted up to eight rings on scales, using marginal increment analysis, to show that the rings form at the same time each year. Thorogood (1987) used otoliths to estimate age in fish ft-om 42 to 167 cm FL and, using marginal increment analysis, was able to show seasonal band formation in what he called ages 2 to 4. Jenkins and Davis (1990) examined microincrements in the otoliths of SBT larvae between 3.5 and 12 mm standard length (SL) collected ft-om the same cohort over consecutive days. From these microincrements, they validated daily increment formation and assigned approximate ages of 7 to 18 days to their samples. In many age determination studies of other species of tuna, tetracycline has been used in marking ex- periments to validate daily increment formation in wild and captive tunas: e.g. yellowfin tuna, Thunnus albacares in the wild (Wild and Foreman, 1980; Wild et al., 1995) and in captivity (Yamanaka'^); skipjack tuna, Katsuwonus pel amis, in the wild (Wild et al., 1995); black skipjack tuna, Euthynnus lineatus, in captivity (Wexler, 1993); and Atlantic bluefin tuna, Thunnus thynnus, in the wild (Inter-Am. Trop. Tuna Comm.-*). However, similar experiments using oxytetracycline (OTC) as a marker in SBT in the 1980s were less successful. In high proportion of OTC-injected fish, a mark failed to show up in the otoliths (Gunn^). Given this previous failure, and public health concerns over the use of tetracycline (in the USA, the Federal Drug Administration [US FDA] prohibits its use in wild fisheries), we selected strontium chloride (SrCl2) as an alternative marker. Strontium chloride is a nontoxic salt that occurs naturally in sea water. It is a component of some foods and is considered to be benign at the concentrations used as a marking agent (Sax and Lewis, 1987). Strontium is readily incorporated into the bloodstream offish and, although not used previously on scombrids. ^ Yamanaka. K. L. 1990. Age, growth and spawning of yellowfin tuna in the southern Philippines. FAO. Indo-Pacif Tuna Dev. Man. Prog. Working paper 90AVP/21. 87 p. " Inter-Am. Trop. Tuna Comni. 1982. Annual Rep. for 1981, .303 p. ^Gunn.J. S. 1992. Progress report on .strontium chloride mark- ing of SBT during 1990-92 CSIRO^JAMARC tagging programs. Paper M\VS4A\T-3. Fourth workshop of the southern bluefin tuna recruitment monitoring and tagging programs. Hobart, Australia, 9 p. Commonwealth Scientific and Indu.strial Research Organization (CSIRO) Marine Research. GPO Box 1.538 Hobart. 7001. Australia. it has been used successfiilly with other fish species to mark vertebrae (by introduction into food or in the surrounding water [Behrens et al., 1990]), and otoliths (by immersion and injection [Brothers, 1990]). Strontium is chemically and biologically similar to calciuffi. Because calcium and strontium ions have the same valency (2+) and a similar ionic radius (Ca, 0.099 nm; Sr, 0.113 nm), strontium readily substitutes for calcium during deposition of calcium carbonate. The first two objectives of this study were 1) to evaluate whether intramuscular injection of strontium chloride resulted in effective and reliable marking of otoliths, and 2) to determine if- strontium chloride injections increased mortality and, hence, affected recapture rates. If successful and benign marking was demonstrated, we planned to use the strontium chloride marks to verify the accuracy of direct aging techniques by determining the periodicity of increment formation for as many year classes of SBT as possible. Materials and methods Tagging and marking From 1990 to 1996, a total of 64,497 juvenile SBT in the Great Australian Bight were tagged and released. Of these, 20,204 tagged SBT were injected with SrCl2 (Table 1). All fish were double-tagged (in case of "tag shedding"! (Williams, 1992): strontium-injected fish were tagged with orange tags, fish that were not in- jected were tagged with yellow tags. When both orange and yellow tags were being deployed, an equal num- ber of fish from targeted schools were chosen at ran- dom for injecting or not injecting. The smallest fish caught during the tagging program was 40 cm fork length (FL); we did not tag and inject fish smaller than this size because they were considered prere- cruits, i.e. were not caught in the fishery. The lengths of orange-tagged fish ranged ft-om 41 to 120 cm in FL. The return rates of yellow-tagged and orange-tagged fish were compared by using a chi-squared test to de- termine if they were significantly different. The strontium chloride solution for injection was prepared in the laboratory. A stock solution of 1 g SrCl2.6H20/mL was made up by dissolving 1 kg of analytical grade SrCl^.eH.jO crystals in 1 liter of distilled water, resulting in a 0.21 g/mL solution of Sr2+. The solution was buffered to pH 7.0 with KOH and stored between 0°C and 4°C. For rapid injections into large numbers of fish, a 5-mL automatic vaccinator fitted with a 0.2-mm needle was used. Flexible tubing connected the vaccinator to a plastic storage container that was either worn as a Clear et a\: Validation of annual increments in otoliths of Thunnus maccoyii 27 Table 1 Numbers of southern bluefin tuna released with yellow tags and numbers of SBT injected with SrClj and released with orange tags, and a summary of yellow and orange tag returns from 1990 to 1996. The number of returned yellow and orange tags released in each year of the tagging program, and the number of returned tags as a percentage of the total number released in the year (shown in parentheses) are shown. (The years of release and recapture are from October of one year to September of the next.) NS = not significant. Fish recaptured 1990-91 1991-92 1992-93 1993-94 1994-95 1995-96 1990-96 All recapture years Difference between yellow and orange tag returns 1990-91 1991-92 1992-93 1993-94 1994-95 1995-96 1990-96 All release year.^- yellow orange tags tags «=6909 n=835 yellow orange yellow orange yellow orange yellow orange yellow orange yellow orange tags tags tags tags tags tags tags tags tags tags tags tags H=4.543 n=3595 n=5907 n=5304 «=8253 fi=82.51 n=15,683 «=2219 n=2998 n=0 n=44,293 n=20,204 183 25 — — — — — — — — — 12.65) (2.99) 180 19 84 55 — — — — — — — (2.61) (2.28) (1.8) (1.53) 111 11 116 86 56 63 — — — — — 11.61) (1.32) (2.6) (2.39) (0.95) (1.19) 62 6 102 76 130 90 50 47 — — — (0.90) (0.72) (2.2) (2.11) (2.20) (1.70) (0.61) (0.57) 29 6 43 36 177 116 171 168 67 20 — (0.42) (0.72) (0.95) (1.00) (3.00) (2.19) (2.07) (2.04) (0.43) (0.90) 2 6 6 29 17 86 89 94 25 21 (0.03) (0.13) (0.17) (0.49) (0.32) (1.04) (1.08) 0.60) (1.13) (0.70) 567 67 351 259 392 286 307 304 161 45 (8.22) (8.03) (7.68) (7.20) (6.64i (5.40i (3.72) (3.69) (1.03) (2.03) 5.02(0.83) 3.17(0.94) 1.44(0.61) 3.22(0.40) 1.68(0.39) 183 25 264 74 283 160 344 219 487 346 217 137 1,779 961 (4.01) (4.76) 2.10(0.56) NS NS NS NS NS NS back-pack or attached to the tagging cradle ( WilHams, 1992). We injected the fish about 2 cm below the dorsal midline, in line with the center of the first dorsal fin. We occasionally obsei-ved a loss of the solution from the muscle, especially in larger fish. When this happened, we noted the loss and injected the fish a second time. To minimize loss of solution we carefully expelled all the air from the vaccinator so that air bubbles were not injected into the muscle. To determine a suitable dosage rate for SBT, initial tiials with strontium were made on thi-ee nontuna species in captivity: silver trevally, Pseudocaranx dentex, sand flathead, Platycephalus bassensis, and purple wrasse, Pseudolabnis fucicola, the largest fish weighing 500 g. The dosages were based on Brothers' (1990) immersion and injection trials with strontium on trout — 100 mg/kg — and Wild and Foreman's ( 1980 ) tetracycline experiments on tuna — 0.3 mlVkg or 27.5 mg/kg of fish. Dosages between 50 and 200 mg Sr/kg resulted in clear Sr marks on otolith sections and no mortalities (CSIRO, unpubl. data^). In the field, the length of the fish was measured to the nearest centimeter, but weight could only be estimated. We therefore chose the required SrClj dose according to length (Table 2). These dosages were increased for fish longer than 90 cm when the marks on the first otoliths recovered ft-om SBT of this size showed up faintly, indicating low strontium absorption. The dosages were adjusted so that at least 80 mg of Sr was injected per kilogi'am of fish. Tag collection and otolith sampling A critical factor in the experiment was the sustained effort, over many years, to recover otoliths from recap- tured fish. An extensive campaign was conducted to explain the objectives of the marking experiment to the SBT fishing industiy and to develop a protocol for '' CSIRO (Commonwealth Scientific and Industrial Research Organi- zation) unpublished data. 1989. CSIRO Marine Research, GPO Box 1538, Hobait, Tasmania, 7001, Australia. 28 Fishery Bulletin 98(1) Table 2 Dosage s for SrCl2-injected fish, based on length and weight of fish FL = fork length. FL Dose 1990- ■92 Dose 1993-96 -^ (cm) (kg) SrCI,, (niL) Sr(mg) mg Sr/kg fish SrC 2(mL) Sr(mg) mg Sr/kg fish 40-50 1.5-3.0 2 400 130-270 2 400 130-270 51-70 3.0-7.0 3 600 86-200 3 600 86-200 71-80 7.0-10 4 800 80-114 4 800 80-114 81-90 10-15 5 1000 67-100 5 1000 67-100 91-95 15-18 6 1200 67-80 7 1400 78-93 96-100 18-21 7 1400 67-78 9 1800 86-100 >100 >21 8 1600 76 12 2400 114 collecting the samples. Posters and information notes, in Japanese and English, were distributed through- out the fishery. Rewards were offered for the return of tags and a substantial bonus for allowing scien- tists to sample otoliths fi-om orange-tagged fish. To aid in otolith recovery, kits containing large orange disks were distributed to the crew of Japanese vessels; when an orange-tagged fish was caught, the disks were at- tached to the fish to clearly identify it in the freezers. Given the high value of SBT on the Japanese Sashimi market, it was essential that otoliths could be sampled without affecting the external appearance of the fish. Using a modification of the technique described by Thorogood (1986). we removed from the skull 35^4 mm cores that contained the semicircular canals and sagittal otoliths with a holesaw attached to a drill. The points of entry for the cores were over the basioccipital plates, which are anterior to the first vertebra and immediately lateral to the junction of the skull and first vertebra; this area was exposed when the tuna were cleaned and dressed — a process which removes the gill-filaments and surrounding tissue and most of the opercular flaps. By drilling through each of the plates in the direction of the back of the opposite eye, the cores coalesced at a point beyond the sagittal otoliths and could be removed easily from the skull, leaving no external mark on the fish (Fig. 1) Sagittal otoliths were removed from the cores, cleaned in distilled water, and dried at 28°C. Age estimates An age estimate was made from the otoliths before we attempted to identify a Sr mark. Increments on the whole sagittal otoliths comprised two zones: an opaque (assumed to be fast growth) and a narrower translucent zone (assumed to be slow growth) that ap- peared dark under a dissecting microscope with re- flected lighting and a black background. Following Thorogood's (1987) method, one of each pair of sagit- tae was burned on a 400°C hot plate until it turned golden brown. The color change was greater in the translucent zones, making them more visible (Fig. 2A). Adigital image of the burnt otolith was taken (us- ing the public domain "NIH Image" program" and a video camera mounted on a Wild MSA dissecting mi- croscope) and measurements were made of the otolith length and of the distance between the primordium and the inside of the translucent zones along both the postrostral (PR) and transverse axes. We use termi- nology that is currently widely accepted (Secor et al. [1992]; KaHsh et al. [1995]; Stequert et al. [1996]). For each specimen, the reader made three independent age estimates by counting the number of increments obvious on the distal surface of the sagitta. These were made without reference to either the length of the fish or the time that the fish was at liberty after tagging and injection. Detection of strontium marks Scanning electron microscope (SEM) with a Robinson backscatter detector In the early stages of the project we used a Robinson backscatter detector coupled to a Philips 515 SEM for detecting the strontium-rich band in the otoliths (which we refer to as "the strontium mark"). The Robinson detector visualizes differences in the total atomic weight ( Z ) across a specimen . Because a strontium mark within the calcium carbonate contains a higher concentration of Sr, and hence a higher Z than surrounding uncontaminated calcium carbonate, it appears as a weak to intense bright band across the growth axis of the section (Figs. 2B and 3). The Rasband, W. 1994. NIH Image, vers. 1.54. U.S. National Insti- tutes of Health. [Available from the Internet by anonymous ftp from zippy.nimh.nih.gov or as a floppy disk from NTIS, 5285 Port Royal Rd., Springfield, VA 22161, part number PB93-504868.) Clear et al : Validation of annual increments in otoliths of Thunnus moccoyil 29 intensity of the band depends on the magnitude of the difference in Z between the two portions of the otohth. This kind of analysis requires a flat, polished surface; therefore we sectioned the sagittal oto- liths either along the postrostral axis to produce an oblique longitudinal section (LS), or along a transverse axis. The rostral axis often shows clear increments, but we did not find distinct Sr marks in this part of the otolith. The sections were ground and polished following Gunn et al.'s (1992) methods and an evaporated carbon coat (25-30 nm thick) was applied to the sections to minimize charging in the SEM. The position of the Sr mark along the axis was measured with the vernier attached to the SEM stage drives (Fig. 2B). Energy-dispersive spectroscopy (EDS) In the later stages, an EDS x-ray microanalysis system became available and we used it to confirm the presence of Sr in the bright bands, and also to detect Sr marks in unsectioned otoliths (the use of unsectioned otoliths decreased the preparation time required for SEM analysis ). The system consisted of a Link 133 eV Si(Li) detector with light element capability and a Thomson Scientific "WinEDS" PC-based analyzer attached to a Philips 515 SEM. Before x-ray analysis, whole (unsectioned) otoliths were acid-etched along the postrostral axis from the surface with 1 N and 3.5 N HCl, to expose the growth plane, then rinsed in bleach and distilled water, and dried. To minimize charging in the SEM, the otoliths were dipped in a dilute carbon DAG solution ( approximately 1:50 with dichloro- ethane) immediately before analysis. Strontium marks were detected by operating the SEM in "spot" mode and searching for a point or zone where a significant Sr signal was detected on the x-ray microanalyzer (typically, a strong peak at 1.81 keV in the spectrum, corresponding to emission of Sr La x-rays). When a strontium mark was detected, its position along the PR axis was measured and the mark photographed either with conventional SEM photography (Fig. 4) or with a rapid, low-grade video print, which also showed the features of interest. To confirm that the suspected mark was strontium-rich, two plots of the x-ray spectra from the otolith were taken: one on the strontium mark and one just before the mark. Acceptable evidence of the correct identification of a strontium mark was considered to be the presence of an enhanced Sr level in the area analyzed, together with an absence of Sr (except for background levels) immediately before the area (Fig. 5). Figure 1 The drilling technique used to extract otoliths from southern bluefin tuna destined to be sold as "whole" fish. The measurements of increments on the whole otoliths, and the strontium mark in sections or whole otoliths, were made along the same axes without reference to the other. This procedure enabled us to compare the number of increments observed after the strontium mark with the number expected, calculated from the known time at liberty after tagging. Quantitative EDS analyses for linescans were car- ried out on carbon-coated polished sections in the Phil- ips 515 SEM operated at an accelerating voltage of 20 kV, by using a focused electron beam of 0.15 pm diam- eter and analysis times of 60-200 seconds. The effec- tive area analyzed by the beam was larger than the diameter of the beam itself because the beam spreads within the specimen after entry; examination of su- perficial beam damage to specimens after analysis suggested that the area analyzed by the beam is in the order of 2 pm diameter. Elemental concentrations were calculated by reference to appropriate standards 30 Fishery Bulletin 98(1) B Figure 2 Sagitta (specimen OB 764) from a fish tagged and injected in March 1994 at 57 cm FL and recaptured in February 1997 at 111 cm FL. (A) Four increments were counted on the whole, burnt otolith and measured from the primordium (P) to the beginning of each translucent zone apparent on the distal surface along the postrostral axis at 2.2 mm, 3.6 mm, 4.4 mm, and 4.9 mm. (B) The sagitta was sec- tioned through the postrostral axis and viewed in the SEM: the micrograph shows the strontium mark which was located 3.4 mm from the primordium and the positions of the translucent zones measured on the otolith before it was sectioned. Scale bars: 1 mm (calcite and celestite for Ca and Sr, respectively), with the "WinEDS" software. Recapture rates of orange-tagged fish and recovery of otoliths ResuKs Of the 20,204 fish injected with SrCl2 between 1990 and 1996, 9614 had been recaptured and 616 sets of sagittal otoliths were recovered from these by 1 January 1996. Seventy sets of otoliths were chosen for the validation study, selected from the range of size classes in the recaptures — fish of 45 to 102 cm FL at release and 57 to 133 cm FL at recapture — and from the range of times at liberty. Age estimates were made from 67 of the 70 otoliths; three sets of otoliths were excluded from the experiment because the increments on the whole otoliths were either ambiguous or uninterpretable and the reader could not give an age estimate with confidence. There were no statistically significant differences be- tween the return rates of yellow tags (from fish not injected with SrCL) and orange tags (from fish in- jected with SrCl2) released in all years of the program (X^=2. 10, P=0.56) nor between the return rates of yellow tags and orange tags for any of the release years (Table 1). The number of otoliths recovered, as a percentage of orange tags recaptured, varied between 20%, in the first year of the tagging program, and 88% in the final year (Table 3); overall, otoliths were recovered from 65% of the orange-tagged fish that were recaptured. Detection of Sr marks In the otoliths of orange-tagged fish The otoliths removed from fish injected with stron- tium chloride typically showed a bright band in back- Clear et aL Validation of annual increments in otoliths of Thunnus maccoyn 31 B Figure 3 SEM micrographs showing strontium marks apparent as bright bands on sections taken from sister otohths of specimen OB 102. (A) An oblique longitudinal section through the postrostral axis on which the primordium (P) and postrostrum (PR) are marked. (B) A transverse section on which the primordium and ventral margin (VM) are marked. Scale bars: 1 mm Table 3 The number of tagged fish recaptured, and the number of otoliths recovered from orange-tagged fish during each year of the tagging program. (Data to 16 February 1996. The years of release and recapture are from October of one year to September of the next.) Recapture year Yellow-tagged fi recaptured sh Orange-tagged fish recaptured Otoliths recovered (otoliths recovered as a percentage of the orange-tagged fish recaptured ) 1990-91 183 25 5(20) 1991-92 264 74 49(66) 1992-93 283 160 94(59) 1993-94 344 219 107(49) 1994-95 487 346 248(72) 1995-96 238 137 120(88) Tbtal 1799 961 623(65) scattered electron images of polished sections through appropriate growth planes (e.g. the oblique longitudi- nal section along the PR axis). The visibility of the Sr mark (brightness in the backscattered electron image) was highest in fish that had been relatively small at the time of injection (e.g. 50-55 cm FL). An example is specimen OB 102 (Fig. 3), which measured 49 cm at time of release and was injected with 2 mL SrClg solution. This bright band was frequently associated with a local growth interruption immediately before the band, presumed to be a tagging check. This check was sometimes seen on the whole, etched otoliths (Fig. 4) but, although apparent in the SEM at high magni- fications, it was not discernible fi-om the other surface 32 Fishery Bulletin 98(1) B Figure 4 SEM micrographs of a carbon-coated whole etched otolith of specimen OB 127, showing the position of the strontium mark and the check associated with tagging and injection. The dotted line (B) indicates the axis along which the position of the strontium mark was measured in relation to the primordium and the postrostrum. Scale bars: 1 mm (A and B), 0.1 mm(C). features on the whole otolith when increments were counted. We could not identify Sr marks on vertebral sections from fish injected with SrCl2. The presence of strontium in the bright bands was demonstrated by EDS spectra, which showed a strong peak of strontium Lor x-rays when, the electron beam was directed to the Sr mark, in contrast with very low (background) levels in the regions of the otolith pre- ceding the mark (Fig. 5). The relative levels of stron- tium and calcium in the bright band were further con- firmed by inspection of the difference spectrum ob- tained by subtracting a spectrum acquired in the area Clear et al.: Validation of annual increments in otoliths of Thunnus maccoyii 33 4095 Ca 3 O ^ O before Sr band (background Sr level) O 5110 B 5110 Hgure 5 Examples of EDS spectra from a sectioned SET otolith showing peaks due to background levels of Sr (A) and enhanced Sr levels associated with the strontium mark (B). Spectra similar to that shown in Figure 6B were used to positively identify the location of strontium marks in sections smd etched whole otoliths (see text for further details). preceding the mark (e.g. Fig. 5A) from one acquired on the bright band (Fig 5B), which demonstrated that strontium levels were enhanced and calcium was re- duced in the bright band; however, no increase in chlo- rine was apparent. A quantitative EDS linescan across a bright band in one specimen, SBT OB 96, sectioned in oblique LS along the PR axis and analyzed along the direction of maximum observed growth, revealed Isackgroimd" levels of 0.1% to 0.25% Sr by weight up to 5 microns before the band and a measured peak of 7.1% Sr on the band, falling to 3.5% (50% of peak level) 6 pm after the start of the band, and 0.7% (10% of peak level) approximately 15 pm after the start of the band. There is some indication of continuing sHghtly elevated Sr levels out to aroimd 50 pm beyond the band, although visibility of these levels is at the Umits of the EDS technique (Fig. 6). Accompanying the measured maximum 7.1% increase in Sr level in the bright band is a fall in measured Ca concentration from 39% to 40% before the band to a minimum of 35.5% on the band — a decrease of 3.5-4.5% in absolute value or 10% relative value. Within the limits of accuracy of the EDS technique, this decrease in calcium concentration supports the theory that Ca atoms are being replaced by Sr atoms in the atomic structure on a 1:1 basis, each Sr atom being approximately twice as heavy as a Ca atom. Calculation of the increase in mean atomic number of the specimen resulting from a 7% increase in Sr and a 3.5% decrease in Ca gives a value of approximately 104 for the Sr-enriched zone. This value compares with 100 for the tmaltered CaCOg — a difference resolvable with backscattered electron imaging on the SEM on a suitably polished and coated specimen. The extent of the visible bright band in this specimen (OB 96) coincided with measured Sr levels in the range of 5—7%; thus it is possible that elevated Sr concentrations in the range of 0.5—5% may not be detectable by backscattered imaging although they should still be detectable by EDS. The EDS system is also essential for testing the identity of weak bright bands in sectioned specimens when it is not clear from 34 Fishery Bulletin 98(1) B Element wt % 40 ^'^^- i — i- =5 Calcium -Cal -Sr% k lirtl ^-^ I — ^ -i Strontium \_y Distance from leading edge of Sr band (microns) Figure 6 Details of the strontium mark on polished LS of specimen OB 96. Backscattered SEM images showing overall location of mark (A) and detail of region analyzed for Sr and Ca levels (B). Dashed line indicates transect followed for x-ray microanalysis (C): Measured variation in Sr and Ca levels along transect shown in B; points A-D indicate representative positions along the transect to enable comparison of sections B and C. the backscattered imaging which band, if any, is a strontium mark. Reliabilrty of marking through injection of strontium Of the 67 otoliths from which an age was estimated, strontiimi marks were detected with the Robinson detector in 19 of the 20 sectioned otoliths (95% detection rate) and with EDS in 40 out of 47 whole otoliths (85% detection rate). Strontium marks were as visible in the obhque longitudinal section (postrostral axis) as in the transverse section (Fig. 3). However, as increments are more widely spaced along the postrostral axis, we generally used the postrostral axis on the whole otolith for measuring the Clear et a\: Validation of annual increments in otoliths of Thunnus maccoyii 35 Number of samples 2 Number of 1 translucent zones after Sr mark Age at recapture (estimated from increment counts on wfiole otoliths) Figure 7 The number of strontium-marked otoliths used to validate age estimates that were made by counting mcrements on whole otoliths. The number of translucent zones observed after the stron- tium mark is shown for each age class. For all otoliths analyzed, the number of translucent zones counted after the Sr mark equalled the number that were expected from the period at liberty after fish were tagged and injected with SrCl.,. position of translucent zones and for locating the stron- tium mark, either on a section (on the Robinson detector), or from the whole, etched otolith (with EDS). Early in the experiment we found that the Sr marks in fish tagged and injected when they were 90 cm PL or larger were consistently fainter than in smaller fish. The concentration of Sr in the Sr marks of large fish was also significantly lower than in smaller fish caused, possibly, by loss of some of the solution during injection. To overcome this, dosages for large fish were increased in 1993 (Table 2), after which the bands were markedly more intense and easier to detect. There was no apparent correlation between the time at liberty (i.e. time between Sr injection and recap- ture) and the intensity of the Sr marks; the longest time at liberty for a fish from which otoliths were analyzed for Sr was 1638 days. There was also no correlation between the intensity of the Sr marks and the delay between otolith recovery and analysis. Unlike tetracycline, which is photosensitive, the strontium mark did not fade after exposure to light. Validation of annual increment formation from Sr marks The 59 fish in which Sr marks were located ranged from 45 to 102 cm FL at the time of release, which corresponds to estimated ages of 1 to 4 (Gunn et al.'^). Times at liberty ranged fi-om 8 to 1638 days. The oldest recaptured fish was estimated to be 6-i- years old; it had been at liberty for 1242 days (over 3 years). Gunn, J., N. Clear, T Carter, A. Rees, C. Stanley J. Kalish, and J. Johnstone. 1995. Age and growth of southern bluefin tuna. Paper SBFWS/95/8. First scientific meeting of the Commission for the Conservation of Southern Bluefin Tuna (CCSBT), Shimizu, Japan, 37 p. Commonwealth Scientific and Industnal Research Organization (CSIRO) Marine Research, GPO Box 1.538 Hobart, 7001 Australia. 36 Fishery Bulletin 98(1 ) We counted six translucent zones on the otolith and the Sr mark occurred between the third and fourth zones (Fig. 7). For all otoliths examined, there was agreement between the number of increments observed after the strontium mark and the number of increments expected, calculated from time at liberty. Thus, the annual periodicity in formation of increments 2 to 6 was validated for the otoliths analyzed. Because we were unable to tag young-of-the-year fish, in which the first translucent zone on the sagitta had yet to form, we could not determine when this translucent zone is laid down and when the formation of the first increment is completed. However, studies of daily microincrements (Itoh and Tsuji, 1996; Rees et al.^) have calculated that the approximate size at age 1 is 50 cm. We found otoliths of 50-cm fish had one increment. Of the otolith increments counted, the first translu- cent zone was typically the most difficult to measure. The beginning of the first translucent zone occurred between 2.2 and 3.2 mm fi-om the primordium along the postrostral axis, the most commonly used axis for analysis. Rees et al.^ found microincrements in this region to be narrower than those deposited earlier, in- dicating a period of slow growth of the fish. Additional validation of annual increment formation from tagged fish at liberty for extended periods During the course of our experiment, two fish tagged by CSIRO in the 1980s were recaptured and their otoliths sampled. From lengths at first release of 45 cm and 82 cm FL, the fish had grown to 163 cm and 162 cm after being at liberty for 9 years, 7 months, and 10 years, 8 months, respectively. From the age-length key developed by Gunn et al.** we calculated that the 45-cm fish tagged in 1983 was one year old when tagged, whereas the 82-cm fish tagged in 1984 was two years old. The ages at recapture of these two fish were estimated fi-om transverse sections through the primordium of the sagittal otoliths. Eleven increments (opaque and translucent zones) were counted on the otoliths from the fish released as a one-year-old and caught 9.58 years later; 13 increments were counted in the fish released as a two-year-old and recaptured 10.75 years later. ' Rees,A.J.,J.S.Gunn,andN. P.Clear. 1996. Age determination of juvenile southern bluefin tuna, Thitnnus maccoyii, based on scanning electron microscopy of otolith microincrements. In J. Gunn, N. Clear, T. Carter, J. Farley, A. Rees. and C. Stanley, Appendix 1 : The direct estimation of age in .southern bluefin tuna. Second scientific meeting of the Commissionfor Con.servation of Southern Bluefin Tuna (CCSBT). Hobart, Australia. 26 August-,'5 September 1996, 22 p. Commonwealth Scientific and Industrial Re.search Organisation (C^SIRO) Marine Research, GPO Box 1,5.38, Hobart, Tasmania, 7001 Australia. Discussion Validation This study demonstrated that, in the sagittae of SBT, the second through sixth increments, are deposited annually. This validation is independent of when the marked fish were tagged or recaptured. Because daily age estimates have been used to demonstrate that the first major increment in the sagitta forms in the first year of life (Rees et al.^), the armual formation of translucent zones appears to hold for the first six increments in SBT sagittae — corresponding to fish up to approximately 133 cm fork length. The close agreement between increment counts on otoliths and the sum of age-at-tagging and time-at- liberty for two fish tagged in the 1980s and recaptured in the 1990s indicated that increment formation continues to be annual in fish up to at least 13 years old. Further evidence that increments in SBT sagittae are formed annually throughout life has been provided by a recent comparison of increment counts with age estimates derived fi-om levels of bomb-radiocarbon in the early growth zones of sagittae (Kalish et al. 1996). This study reports close agreement between the two methods of estimating age for fish up to 34 years old. Three sources of data — those fi-om our marking ex- periment, the increment counts for two fish at liberty for over a decade, and the bomb radiocarbon data — provide strong evidence that seasonal changes in growth are expressed as clearly identifiable annual increments in the sagittal otoliths of SBT. These increments can be used to estimate the age of individual fish at any point in their lifespan. Prior to our studies, Yukinawa (1970, using scales) and Thorogood (1987, using otohths) used marginal increment analyses to demonstrate the annual check or translucent band deposition in fish they considered to be between 2 and 4 years old. Their results differ from ours only in the identity of year classes; their two- to four- year-olds correspond to our one- to three-year-olds. The difference in scale readings derives from Hynd's (1965) observation of two "checks" on the scales of new recruits (approximately 50 cm FL) to the Western Australian fishery. Interpretation of otolith microincrements (Itoh and Tsuji, 1996; Rees et al.^) indicates that these fish are only one year old. Unequivocal validation of these estimates is not possible at this stage because samples from prerecruits were not available to either Hynd or Yukinawa and we were not able to tag and mark prerecruit fish. In a number of other Thiinnus species however, 50-60 cm fish were found to be aroimd one year old (Uchiyama and Struhsaker, 1981; Wild, 1986; Foreman, 1996) and our counts of otolith microincrements and the data based on their inter- Clear et a\. Validation of annual increments in otoliths of Thunnus maccoyii 37 pretation are consistent with this age. Therefore, we beheve that the interpretation of Itoh and Tsuji (1996) and Rees et al.^, that 50-cm-FL fish are one-year-olds, is most Ukely correct. The identification offish that we considered to be one- year-olds as two-year-olds was made by Thorogood in his 1987 study. However, we have found no evidence of two increments in the otoliths of new recruits. The early zones on all axes of otohth growth are difficult to read on some otoliths, and the increments in these areas are less distinct than those deposited later. In some fish a poorly defined "band" is also present very close to the primordium (within 2 mm along the postrostral axis). Although these two factors may confuse an inexperienced reader, Thorogood makes no mention of difficulty in reading the first increment. An alternative explanation for Thorogood's interpretations may be that his readings were influenced by the findings of Hynd (1965) and Yukinawa (1970) that were based on scales, because their estimates of age at recruitment were entrenched within the dogma of SBT population dynamics current in the 1970s and 1980s. It has been hypothesized that more than one trans- lucent zone forms per year in the otoUths of mature Atlantic bluefin tuna, Thunnus thynnus (Berry et al., 1977; Lee et al., 1983). In females, one translucent zone may correspond to a winter slow-growth period, the other to a spawning period (Lee et al., 1983). In the two large, tagged fish examined in our study, only one opaque and one translucent zone were deposited per year throughout hfe. The outer increments (i.e. those assumed to be deposited after sexual maturation) were consistent in both their width and optical density and were visually equivalent to the increments described by Lee et al. ( 1983), comprising a wide opaque region and a narrow translucent area that, on a black background, appears dark under reflected light. Occasionally, there appeared to be two translucent zones closer together than normal and, if these bands coalesced at the margin, they were counted as part of the same increment. These may be equivalent to the bands described by Berry et al. ( 1977 ) who hypothesized that a pair of these paired bands represented an annual increment. The close agreement between otolith increment counts and bomb- radiocarbon age estimates for mature SBT up to 34 years old (Kahsh et al., 1996) supports our hypothesis that one increment, comprising one translucent and one opaque zone, continues to form per year, as does the consistency of the width and optical density of increments deposited after sexual maturation in the otohths aged by Kalish et al. (1996). In summary, there is no significant evidence to suggest that mature female SBT deposit two translucent zones per year. In this regard our findings are similar to those of Hurley and lies (1983) and Hurlbut and Clay (1988) for T. thynnus; they found, albeit in the absence of direct validation, that a single translucent zone is laid down per year in medium- and giant-size classes. The use of strontium chloride to mark otoliths of large fish This study has shown that intramuscular injection of strontium chloride leaves a distinct mark on the oto- liths of SBT that is clearly visible as an SEM back- scatter image in the Robinson detector. In the 20 oto- lith sections from Sr-injected fish that we examined, 95% had detectable marks. On this basis, we conclude that the compound is an efficient marker. Success of OTC as a marker at this rate of detection (95%) leads to high mortalities (McFarlane and Beamish, 1987). The high detection rates and lack of evidence of mor- tality for SrClg are not surprising. This mineral occurs naturally in sea water, the mean concentration being 3.8-8.2 ppm (Carriker et al., 1991) or 0.09 mM/kg (Bruland, 1983), and both Sr and CI are major con- stituents of the otoliths of SBT (Gunn^). When SrCU is injected into the muscle it is taken up into the blood stream and incorporated in the endolymph and then the otolith, substituting for Ca within the CaCOg por- tion of the aragonite. The combined weight fraction of Ca and Sr ( approximately 42% ) within the otolith does not change as a result of the injection. However, the Ca:Sr ratio changes fi-om 250:1 before injection to as low as 5:1 during the period over which the Sr spike induced by the injection is metabolized. At a dis- tance of 6 pm beyond the injection mark, the Sr lev- els have dropped to about 50% and at 15 pm to 10% of peak values (Fig. 6). These distances correspond to time periods in the order of 2 and 5 days, respectively, based on median growth rates of around 3.0 pm/day estimated along this axis (Rees et al.^). Detecting Sr marks on whole otoliths by using EDS was possible because the growth plane of tuna otoliths lies near the surface of the distal face. In otoliths of young fish, etching will expose the growth plane, so that sectioning is not required. Although this method of detection was slightly less successful (85%), it had two advantages. First, the preparation time was around half that required to prepare sections suitable for the Robinson detector. Second, the age estimate and measurements of increments were made along the postrostral axis on whole otoliths from the smaller fish (up to six years old), and the method of locating the strontium mark by EDS meant that the position of the strontium mark was measured along the axis in the same plane. With the Robinson detector, the same axis was measured but in cross section. In fish older than about 6 years, the increments de- posited on the margin can be unclear on whole oto- 38 Fishery Bulletin 98(1 liths; therefore transverse sections are used to deter- mine ages of older fish (Gunn et al.^). In the future, as strontium-marked otoliths are returned from older fish that have been at liberty for longer periods, we will locate the strontium mark in transverse sections with the SEM and increase the number of increments that have been validated. The recapture rates of orange-tagged and injected SBT were not significantly different ft-om the recapture rates of yellow-tagged SBT; therefore the Sr injections apparently did not affect survival rate. Although the dosages of Sr varied between 65 and 250 mg/kg of fish, there is no evidence to suggest that higher doses increased mortality because Sr-injected fish with the highest doses were among those recaptured. A direct relation between the dosage and the intensity of the mark had been found in trials with three other species in 1990-9 1 ( CSIRO, unpublished datai^ ). The increased dose for large SBT resulted in much more distinct marks on their otoliths, which have a larger surface area than that in smaller fish. The less distinct marks could also be attributable in some large fish to a loss of strontium solution fi-om the muscle after injection. Although more solution was injected if a loss was noticed, there may have been further loss of solution afler the fish was returned to the water, resulting in a less effective dose. Thus, as a general guideline, we recommend a dose of 100 mg Sr/kg for marking otoliths in SBT. We note, however, that tissue area around the injection should be observed to ensure that there is no loss of injected solution fi-om the muscle tissue. The problem of detecting indistinct marks that re- sult from low dosage levels are eliminated by using SrCl2 as a marker. Unlike fluorescent marking, where it is very difficult to evaluate faint or ambiguous marks objectively (particularly if they are close to the outside edge of the otolith), it is possible to evaluate Sr marks objectively by x-ray analysis. Because the concentra- tions of the Ca and Sr on the Sr mark are high, very simple energy dispersive spectroscopy systems, which are available in many SEM facilities, can be used. Al- though not a trivial procedure, x-ray analysis requires preparation methods similar to those used for examin- ing fluorescent markers and can usually be contracted to facilities at a low cost. Given the ofi;en substantial investment in tagging progi'ams, and the common combination of low recapture rates and even lower oto- lith sampling rates, every sample is extremely valu- able in a marking experiment. The safety net of chem- ical analysis is thus very advantageous. Comparison of strontium and fluorescent markers At the beginning of this project we chose strontium chloride over the more commonly used fluorescent markers because previous work on SBT with oxytet- racycline had been unsuccessful. Although immersion in high concentrations of strontium or feeding with strontium-laced food (or both) had been used success- fully for marking hard parts of larvae and juveniles of hatcheiy-reared salmon (Behrens Yamada and Mul- ligan, 1982; 1990), salmonids and a variety of tropi- cal fish species (Brothers, 1990) and squid (Hurley et al., 1985), strontium had not previously been used to mark otoliths of large fish. On the basis of his experi- ments. Brothers ( 1990 ) concluded that, for mass mark- ing of fish in captivity, detection of strontium marks was expensive and involved more difficult preparation than did fluorescent markers and other marking tech- niques such as thermal inducement (Volk et al., 1990). Brothers' (1990) comment on expense is certainly pertinent but the expense of analyzing marked oto- liths is often a small fraction of the cost of a marking experiment, particularly one where large numbers of fish have been tagged, injected, and released. Perhaps most important in the cost equation should be the rate of success of detecting marks in the otoliths of marked fish rather than the comparative cost of analysis. In otoliths from large tuna whose time at liberty has been long, the strontium marks are covered by a large amount of otolith material deposited after the time of injection. Sectioning is necessary for either marker; thus preparation times in these cases are much the same. For smaller tuna and those at liberty for short periods, fluorescent markers can be detected in the whole otolith, whereas detection of strontium with- out an EDS system would require sectioning, which would increase preparation time. The equipment for fluorescent markers is cheaper and comprises a light microscope equipped with an ultraviolet illumination source and filters to match the wave length of the fluo- rescence emitted from the marker when excited by the light source (see Wild and Foreman, 1980). For stron- tium, an SEM equipped with a Robinson detector is the minimum requirement; an EDS system is a use- ful extra. Although an SEM is a common apparatus in large research laboratories, hourly charges to the user can be high, although we have found that, with well prepared specimens, as many as four otoliths can be examined and analyzed per hour with an SEM. Apart from preparation time and costs, strontium marking for age validation has clear advantages over fluorescent marking. One benefit of a technique that requires both a light microscope and an SEM is that measurements of increments and strontium marks are independent: the strontium cannot be detected in whole otoliths under the light microscope and the annual increments cannot be observed in the SEM. Allergic reactions by humans to compounds such as oxytetracycline have led the U.S. FDA to ban their use Clear et a\: Validation of annual increments in otoliths of Thunnus maccoyii 39 in commercial fisheries. Strontium chloride, on the other hand, is regarded as safe for human consumption because it is a salt with a low order of toxicity (Sax and Lewis, 1987). It is even used in toothpaste by some manufacturers (e.g. "Sensodyne"). Strontium chloride, unlike fluorescent markers such as oxytetracycline, is not photosensitive. Neither the marking solution nor the marked otoliths need to be stored in the dark, and the mark does not fade with exposure to light or with time. In our study, strontium marks were as evident in fish that had been at liberty for long periods as in fish recaptured soon after release. In summary, strontium chloride injection has proved to be a very successful way to mark the otoliths of southern bluefin tuna: 959f of those marked and re- captured in this study had detectable Sr marks in sec- tioned otoliths. This high "success rate," the harmless nature of SrCl, to both fish and humans, the capacity of EDS to positively identify the strontium mark, the insensitivity of the strontium mark to light, and the longevity of the strontium mark indicate that it should be seriously considered by those interested in large- scale marking experiments on commercial fishes. Acknowledgments The study was part of a tag-recapture project run by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the Japan Marine Fishery Resource Research Centre (JAMARC). The efforts of K. Williams, D. Waddington, W. Whitelaw, T. Carter, and the Australian Fisheries Management Authority observers ensured the collection of adequate numbers of marked otoliths. We acknowledge the valuable contribution of W. Hearn, C. Proctor, and K. Williams, who initiated and implemented the 1990-91 study to assess the feasibility of injecting strontium chloride to mark juvenile SBT which was partly financed by an Environmental Research Grant from the Australian Government Department of Primary Industries and Energy. Our thanks also go to Sandy Morison, Craig Proctor, and other anonymous reviewers who gave their time to assess this manuscript and to Vivienne Mawson for her editorial comments. Literature cited Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fisher- ies biology. Trans. Am. Fish. Soc. 112 (6):735-743. Behrens Yamada, S., and T. J. Mulligan. 1982. Strontium marking of hatchery reared coho salmon, Oncorhynchus kisutch Walbaum, identification of adults. J. Fish Biol. 20:5-9. 1990. Screening of elements for the chemical marking of hatchery salmon. Am. Fish. Soc. Symposium Ti.'iSO-SGl. Berry, F. H. , D. W. Lee, and A. R. Bertolino. 1977. Progress in Atlantic bluefin tuna ageing attempts. Coll. Vol. Sci. Pap. Int. Comm. Conserv. Atlantic Tunas. 6121:30.5-317. Brothers, E. B. 1990. Otolith marking. Am. Fish. Soc. Symposium 7:183- 202. Bruland, K. W. 1983. Trace elements in sea-water. In J. P. Riley and R. Chester (eds). Chemical Oceanography, vol. 8. Academic Press. London. Carriker, M. R., C. P. Swann, R. S. Prezant, and C. L. Counts, m. 1991. Chemical elements in the aragonitic and calcific micro- structural groups of shell of the oyster Crassostrea virginica: a proton probe study. Mar. Biol. 109:287-297. Caton, A. E. 1991. Review of aspects of southern bluefin tuna biology, pop- ulation and fisheries. In R. B. Deriso and W. H. BayliflF (eds.). World meeting on stock assessment of bluefin tunas: strengths and weaknesses. Inter-Am. Trop. Tuna Comm. Special Report 7:181-350. Foreman, T. 1996. Estimates of age and growth, and an assessment of ageing techiques, for northern bluefin tuna, Thunnus thyn- nus, in the Pacific Ocean. Inter. Am. Trop. Tuna Comm., Bull. 21(21:75-123. Gunn, J. S., I. R. Harrowfield, C. P. Proctor, and R. E. Thresher. 1992. Electron probe microanalysis of fish otoliths — evalu- ation of techniques for studying age and stock discrimina- tion. J. Exp. Mar Biol. Ecoi. 158:1-36. Hampton, J. 1991. Estimation of southern bluefin tuna Thunnus mac- coyii gi'owth parameters from tagging data, using von Ber- talanffy models incorporating individual variation. Fish. Bull. 89(4):577-590. Hurlbut, T., and D. Clay. 1988. A review of age and growth of Canadian giant bluefin as estimated from otoliths. Coll. Vol. Sci. Pap. Int. Comm. Conserv. Atlantic Tunas. 28:192-195. Hurley, G. V., P. H. Odense, R. K. O'Dor, and E. G. Dawe. 1985. Strontium labelling for verifying daily growth incre- ments in the statolith of the short-finned squid (Illex illece- brosus). Can. J. Fish. Aquat. Sci. 42:380-383. Hurley, P. C. F, and T. D. lies. 1983. Age and growth estimation of Atlantic bluefin tuna, Thunnus thynnus. using otoliths. In E. D. Prince and L. M. Pulos (eds.). Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, billfishes and sharks, p. 71-75. U.S. Dep. Commer., NOAATech. Rep. NMFS 8. Hynd, J. S. 1965. Southern bluefin tuna populations in south-west Aus- tralia. Aust. -J. Man Freshwater Res. 16:25-32. Itoh, T., and S. Tsuji. 1996. Age and growth of juvenile southern bluefin tuna Thun- nus maccoyii based on otolith microstructure. Fish. Sci. 62(6):892-896. Jenkins, G. P., and T. L. O. Davis. 1990. Age, gi-owth rate, and gi'owth trajectory determined from otolith microstructure of southern bluefin tuna Thun- nus maccoyii larvae. Mar. Ecol. Prog. Sen 63:93-104. 40 Fishery Bulletin 98(1) Kalish, J. M., R. J. Beamish , E. B. Brothers, J. M. Casselman, R. I. C. C. Francis, H. Mosegaard, J. Panfih, E. D. Prince, R. E. Thresher, C. A. Wilson and P. J. Wright. 1995. Glossary. In D. H. Secor. J. M. Dean and S. E. Campana ( eds. ). Recent developments in fish otolith research, p. 723-729. Belle W. Baruch Library in Marine Science, No. 19, Univ. South Carolina Press, Columbia, SC. Kalish, J. M., J. M. Johnston. J. S. Gunn, and N. P. Clear. 1996. Use of the bomb radiocarbon chronometer to determine age of southern bluefin tuna Thunnus maccoyii. Mar Ecol. Prog. Ser 143:1-8. Lee, D. W, E. D. Prince, and M. E. Crow. 1983. Interpretation of growth bands on vertebrae and oto- liths of Atlantic bluefin tuna. Thunnus thynnus. In E. D. Prince and L. M. Pulos (eds.), Proceedings of the interna- tional workshop on age determination of oceanic pelagic fishes: tunas, billfishes and sharks, p. 61-69. U.S. Dep. Commer. NOAATech. Rep. NMFS 8. McFarlane, G. A., and R. J. Beamish. 1987. Selection of dosages of oxytetracycline for age valida- tion studies. Can. J. Fish. Aquat. Sci. 44:905-909. Murphy, G. M., and J. M^'kowski. 1981. State of the southern bluefin tuna population: fully exploited. Australian Fisheries 40(11 ):20-29. Sax, N. I., and R. J. Lewis Sr. 1987. Hazardous chemicals desk reference. Van Nostrand Rienhold, New York, NY. 1084 p. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstructural examination. In D. K. Stevenson and S. E. Campana (eds. ). Otolith microstructure examination and analysis, p. 19-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Secor, D. H., J. M. Dean, and S. E. Campana. 1995. Fish otoliths: faithful biological and environmental chronometers? In D. H. Secor, J. M. Dean and S. E. Campana (eds.). Recent developments in otolith research, p. xxv-xxvii. Belle W. Baruch Library in Marine Science 19. Univ. South Carolina Press, Columbia, SC. Smith, D. C. 1992. Introduction. In D. C. Smith (ed.). Age determination and growth in fish and other aquatic animals. Aust. J. Mar Freshwater Res. 43:vii-viii. Steqert, B., J. Panfili and J. M. Dean. 1996. Age and growth of yellowfin tuna, Thunnus albacares, from the western Indian Ocean, based on otolith microstruc- ture. Fish. Bull. 94:124-134. Thorogood, J. 1986. New technique for sampling otoliths of sashimi-grade scombrid fishes. Trans. Am. Fish. Soc. 115:913-914. 1987. Age and growth rate determination of southern blue- fin tuna, Thunnus maccoyii. using otolith banding. J. Fish Biol. 30:7-14. Uchiyama, J. H., and P. StruhsEiker. 1981. Age and growth of skipjack tuna, Katsuwonus pelamis, and yellowfin tuna, Thunnus albacares. as indicated by daily growth increments of sagittae. Fish Bull. 79(1):151-162. Volk, E. C, S. L. Schroder, and K. L. Fresh. 1990. Inducement of unique otolith banding patterns as a practical means to mass-mark juvenile Pacific salmon. Am. Fish. Soc. Symposium 7:203-215. Wexler, J. B. 1993. Validation of daily growth increments and estimation of growth rates of larval and early-juvenile black skipjack, Euthynnus lineatus. using otoliths. Inter-Am. Trop. Tuna Comm. Bull. 20(7):399^40. Wild, A. 1986. Growth of yellowfin tuna. Thunnus albacares. in the eastern Pacific Ocean based on otolith increments. Inter. Am. Trop. Tuna Comm. Bull. 18(6):423^82. Wild, A., and T. J. Foreman. 1980. The relationship between otolith increments and time for yellowfin and skipjack tuna marked with tetracycline. Inter-Am. Trop. Tuna Comm. Bull. 17(7):507-560. Wild, A., J. B. Wexler, and T. J. Foreman. 1995. Extended studies of increment deposition rates in oto- liths of yellowfin and skipjack tunas. Bull. Mar. Sci. 57(2): 555-562. Williams, K. 1992. The tagging technique. Australian Fisheries 51(6): 1.5-17. Yukinawa, M. 1970. Age and growth of southern bluefin tuna Thunnus mac- coyii (Castelnau) by use of scale. Bull. Far Seas Fish. Res. Lab. 3:229-257. 41 Abstract.— Porbeagle sharks, Lamna nasus. are caught in large numbers as bycatch in tuna longline fisheries in the southwest Pacific Ocean. Information on reproduction, embryonic development, and size and sex composition was col- lected by scientific obser\'ers fi-om New Zealand and Australian waters, and sup- plemented with data fi-om other sources. Most sharks were juveniles less than 150 cm fork length (FL), and length-fre- quency distributions showed 3-5 modal peaks that we interpret as age classes. Juveniles grow linearly and rapidly ( 16- 20 cm per year), reaching 110-125 cm FL in three years. Females mature at around 165-180 cm. Litter size is usually four embryos and parturition probably peaks in June-July (winter). This finding con- trasts with data for North Atlantic porbeagles which give birth in spring- summer Embryos grow about 7 cm per month, and are bom at 58-67 cm FL. The gestation period appears to be about 8-9 months, but there is considerable vari- ability in embryo length at any one time, suggesting an extended mating period. Embryos are nourished by oophagy, and develop a grossly distended abdomen as their "yolk stomach" fills with ova. Small embryos have fang-like functional teeth that tear open egg capsules to release the contained ova. The fangs are shed at 34-38 cm FL. The weight of yolk in the stomach peaks at 30-42 cm FL, and accounts for up to Sl'i of total body weight. Waste products of yolk diges- tion accumulate steadily in the spiral valve throughout gestation, and the liver reaches its maximum size in near-term embryos as excess energy from yolk digestion is stored for postnatal use. Reproduction, embryonic development, and growth of the porbeagle shark, Lamna nasus, in the southwest Pacific Ocean Malcolm P. Francis National Institute of Water and Atmosphenc Research PO Box 14-901 Wellington, New Zealand E mail address m franciSiS'niwa.ai.nz John D. Stevens CSIRO Manne Research G PO Box 1538 Hobart, Tasmania, Australia Manuscript accepted 26 June 1999. Fish. Bull. 98:41-63 (2000). The porbeagle, Lamna nasus (Bon- naterre, 1788) is a pelagic mackerel shark (family Lamnidae) that inhab- its cool, temperate oceans. It occurs in the North Atlantic Ocean and in a circumglobal band in the southern Pacific, Atlantic and Indian Oceans (Compagno, 1984; Last and Stevens, 1994; Yatsu, 1995). It is absent from the North Pacific, where it is re- placed by its closest relative, the salmon shark (Lamna ditropis). Lamnid sharks produce a small number of large, live young that are nourished by oophagy (Gilmore, 1993), In this unusual form of em- bryonic development, the pregnant female ovulates an enormous num- ber of ova which are consumed by the embryos in the uteri. The embryos develop grossly swollen abdomens as they store large quantities of yolk for later growth. Oophagy was first described in porbeagles (Swenander, 1906, 1907; Shann, 1911, 1923) and salmon sharks (Lohberger, 1910), but has only recently been confirmed in shortfin and longfin makos (Isu- rus oxyrinchus and /, paucus) and white sharks iCarcharodon carchar- ias) (Gilmore, 1983; Stevens, 1983; Francis, 1996; Uchida et al., 1996), The unusual bloated appearance of porbeagle embryos has led to a num- ber of reports in the literature (Big- elow and Schroeder, 1948; Graham, 1956; Templeman, 1963), but the ab- sence of a series of embryos at dif- ferent stages of gestation has ham- pered attempts to understand their development. Litters usually consist of four embryos (Templeman, 1963; Gauld, 1989), which are thought to be bom at about 60-80 cm total length (TL) (Shann, 1923; Compagno, 1984; Last and Stevens, 1994). Female size at maturity is often cited as 152 cm TL (Bigelow and Schroeder, 1948; Compagno, 1984; Last and Stevens, 1994), apparently based on two preg- nant females reported to have been "about five feet long" (Shann, 1911). However, no other mature females under 2 m TL have been reported, leading some authors to regard the length at maturity as 2-2.5 m TL (Aasen, 1963; Pratt and Casey 1990). The length of the gestation period is unknown; estimates, however, range from eight months to two years (Shann, 1923; Aasen, 1963; Gauld, 1989 ). The timing of parturition is var- iously stated as spring, summer, or autumn in the North Atlantic (Bige- 42 Fishery Bulletin 98(1) low and Schroeder, 1948; Aasen, 1963; Gauld, 1989). Thus, despite the early discovery of oophagy in por- beagles, little is known about their reproduction. Most parameter estimates are imprecise, and several are speculative or conflicting. Few pregnant females have been reported from the Southern Hemisphere, and few details have been pro- vided for any of them. Graham (1939, 1956) reported one caught at Otago Heads, New Zealand, in 1933. It had three embryos that were approaching full term and weighed 3.4-4.3 kg each. Graham (1956) also examined several other pregnant females but he re- ported few details. Duhamel and Ozouf-Costaz (1982) found four small embryos in a female caught in 1981 near Kerguelen Island in the southern Indian Ocean (51°S, 70°E). Growth curves are available for northwest Atlantic porbeagles, based on modal analysis of length-fre- quency distributions, and back-calculation of length- at-age from bands on a vertebra (Aasen, 1963). They suggest that growth is relatively fast, at least in the first few years, and that longevity is 20-30 years. No growth information is available for the Southern Hemisphere. Porbeagles have been exploited for their flesh for many decades, and have proven to be vulnerable to overfishing. A target longline fishery in the northwest Atlantic in the 1960s lasted only six years before col- lapsing (Anderson, 1990; Pratt and Casey, 1990). In the Southern Hemisphere, porbeagles have not been targeted, but they are frequently taken as bycatch in tuna fisheries, especially the pelagic driftnet fishery for albacore (Thunnus alalunga) during 1982-91 in the South Pacific (Murray, 1994; Yatsu, 1995), and the longline fishery for southern bluefin tuna (Thunnus maccoyii) and bigeye tuna (Thunnus obesus) in the southern Indian and Pacific Oceans (Stevens et al., 1983; Francis et al., 1999). In the New Zealand long- line fishery, porbeagles are the second most commonly caught shark after the blue shark (Prionace glauca) (Francis et al., 1999). The collapse of the northwest Atlantic fishery in the 1960s provides ample justification for a cautious ap- proach to managing porbeagles. In view of recent in- creased landings in the North Atlantic (O'Boyle et al.. 1996), and the size and scope of the tuna longline fish- ery in the southern oceans, there is an urgent need for improved information on reproduction, growth, and stock productivity as a basis for effective management. Much of the Southern Hemisphere longline fishei-y oc- curs in international waters, making monitoring and management difficult. Recently, the New Zealand and Australian governments implemented scientific ob- server programs to monitor catches of foreign and do- mestic longline vessels in their respective Exclusive Economic Zones (EEZs). These programs provided an opportunity to collect information on the reproduction and growth of porbeagles. In this paper, we describe the geographical distribution and length composition of sharks taken by longline vessels in the southwest Pacific, estimate the growth rate of embryos and juve- niles, and describe embryonic development and ooph- agy. We also estimate the length of the gestation pe- riod, the timing of parturition, and the size at birth, and compare these with estimates for North Atlantic porbeagles. Materials and methods Data sources Most of our data and specimens were collected by sci- entific observers aboard Japanese and domestic tuna longline vessels operating in the New Zealand and Australian EEZs. In New Zealand, fishing and ob- server effort was concentrated in two regions: 1 ) north- east New Zealand (north and east coasts of North Is- land and the Kermadec Islands), and 2) southwest New Zealand (east and west coasts of South Island) (Fig. 1). In Australia, most effort was around Tasma- nia (Fig. 2). New Zealand obsei'vers began recording the quantity of bycatch in 1987, measuring and sexing porbeagles in 1990, and examining females for em- bryos in 1992. In Australia, the respective years were 1988, 1990 and 1991. The primary task of observers was to monitor the target tuna species (mainly south- ern bluefin and bigeye tuna). Porbeagles were counted on most longline sets but were not always measured or examined for embryos. Therefore our data repre- sent a subsample of the catch taken by observed long- liners. The opportunistic nature of this collection pro- cess, and the low catch rate of pregnant females, ne- cessitated the accumulation of specimens and data over a lengthy period. Embryos collected by observers were supplemented by specimens and data from other sources, and the lit- erature, including three litters from Heard and Ker- guelen Islands in the southern Indian Ocean (Table 1). The four embryos from the Kerguelen female were de- posited in the Museum National d'Histoire Naturelle (MNHN 1981-1432-1981-1435) (Duhamel and Ozouf- Costaz, 1982), and were photogi-aphed and remea- sured for us by Duhamel.' A 185-cm-fork length (FL) female from Macquarie Island (Fig. 1) was the only intact pregnant female we examined. ' Duhamel. G. 1997. Museum National d'Histoire Naturelle ( MNHN), 75231 Paris cede.x 05, France. Personal commun. Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 43 25 S 30 35 40 45 50 55 160 rp,— r 165 170 175 E — 1 1 r- 180 175 W 170 25 S o Observed on longlines Pregnant females ;Macquane Island 30 35 40 45 50 55 160 165 170 175 E 180 175 W 170 Figure 1 Map of the New Zealand region showing start positions of tuna longUne sets from which porbeagles were recorded, and capture locations of pregnant females (n=35). The 250-m isobath and Exclusive Economic Zone are also shown. Size and growth Porbeagles were measured in one or more of three ways: precaudal length (PL; snout to the precaudal pit), FL (snout to the fork in the tail), and TL (snout to the tip of the tail). TL can be measured in two differ- ent ways — ^with the tail in the natural position (TL^^j) (Bigelow and Schroeder, 1948), or with the tail flexed down so that the upper caudal lobe lies parallel to the long axis of the body ( TLjj^j, ) ( Compagno, 1984 ). Observ- ers probably measured TL^^, on postnatal porbeagles because TLj^^j^ is difficult to measure in species with a relatively rigid caudal fin. TL^^^j measurements in em- bryos are not strictly comparable with TL^^.^^ measure- ments in postnatal porbeagles because of the curved and folded nature of the caudal fin in embryos. Most observers measured FL; therefore we adopted that as our standard. Regression equations relating FL to TL and PL are given in the "Results" section. Literature reports of TL were converted to FL before comparison with our data. Hereafler, FL is reported unless otherwise stated. Porbeagles were also sexed, weighed whole, and sometimes weighed after process- ing. Data were inspected for outliers on bivariate plots of PL, FL, TL, whole weight, and processed weight. Obvious errors were corrected if possible, and deleted if not. Before 1993, some New Zealand observers con- fused porbeagles and shortfin makos. We therefore re- stricted our New Zealand analyses of length, weight and location to data collected from 1993 onwards. Initial inspection of the length-frequency data re- vealed modes that might correspond with juvenile age 44 Fishery Bulletin 98(1) 100 E 110 120 130 140 150 160 10S 20 30 40 50 o Observed on longlines Pregnant females 10S 20 30 40 50 100 E 110 120 140 150 160 130 Figure 2 Map of the Australian region showing start positions of tuna longline sets from which por- beagles were recorded, and captui-e locations of pregnant females l7!=5.). The Exclusive Eco- nomic Zone is also shown. classes. To assist modal discrimination, we limited our length-frequency analyses to the period April-July, during which Kl'''( of New Zealand and 86'^^^ of Austra- lian length measurements were taken. The MIX com- puter program (MacDonald and Pitcher, 1979; Mac- Donald, 1987; MacDonald and Green, 1988) was ap- plied separately to the New Zealand and Australian length-frequency data for combined sexes to decompose the distributions into their component age classes. The program estimates the mean length, and the standard deviation of the lengths, for each age class, and the proportion of the sample in each age class. Length-fire- quency data were grouped into 3-cm class intei'vals, and truncated at 162 cm (New Zealand) and 150 cm (Australia) before analysis because large sharks were poorly represented in the samples. For each data set, we fitted a MDC model with three age classes and then progressively added extra age classes until there was no significant improvement in the )[- goodness of fit (Mac- Donald and Green, 1988). Occasionally, partially con- strained fits ( alternately fixing the standard deviations and proportions of one or two of the older age classes ) were necessary for successfiil convergence (MacDonald and Green, 1988). This had no effect on the estimates of mean length, which were never constrained. Embryos and ova Embryos were placed in plastic bags and ft-ozen; some- times only partial litters were retained. Rarely, the uteri were removed intact and frozen with the em- bryos still inside. In the laboratory, embryos were thawed, sexed, and their jaws were examined for func- tional teeth. The embryos were then weighed, and measured (usually PL, FL, TL„.,,, and TL^^,^). FL was estimated from TL for three embryos without FL mea- surements (see "Results" section for regression equa- tions). The liver and the contents of the stomach and intestine were weighed separately. Stomach con- tents were expressed as a percentage of total weight. Liver weight, and the weight of the intestinal contents were expressed as percentages of yolk-free embryonic weight to avoid distortions caused by the large varia- tion in the stomach contents. The uteri and right ovary- of the Macquarie Island female were examined. The diameters of a subsample of ovarian ova were measured using an image analy- sis system attached to a binocular microscope, and an - The left ovary of lamnid sharks is vestigial iPratt, 1988). Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 45 estimate was made of the total number present by counting ova in six weighed subsamples. The four MNHN embryos from Kerguelen Island were measured for TL, FL, and weight in December 1997. Because there was a mean shrinkage of 4.4% from the original TL measurements (Duhamel and Ozouf-Costaz, 1982), we applied an equivalent shrink- age correction to the 1997 FL measurements. Fresh embryo weights were not reported by Duhamel and Ozouf-Costaz (1982), and we have not used the 1997 weights because they probably underestimate the orig- inal weights due to dissolution and leaching of lipids from the yolk in the stomach. Similarly, measure- ments from the Museum of New Zealand (NMNZ) em- bryos are not included here because of likely shrink- age and weight loss following preservation. North Atlantic embi-yo lengths and dates of cap- ture were obtained from the literature for comparison with Southern Hemisphere data (Swenander, 1906, 1907; Shann, 1911, 1923; Nordgard, 1926; Bigelow and Schroeder, 1948; Templeman, 1963; Gauld, 1989; Moss^; Newton'*). For some litters, only one or two embryos were measured. Data were used only if they specified the month of capture, and some measure- ments that were known or thought to have been made on preserved specimens were corrected for shrinkage. Sea surface temperature was recorded at about hourly intervals during hauling of each longline in New Zealand. The number of sharks caught per 1000 hooks (CPUE) was determined for each set and plotted against the mean of the hourly SSTs. There was no ap- parent trend in CPUE between 9.85°C (the minimum set temperature) and 19.5°C (mean CPUE=1.82, max- imum=44.8, n = 1292 sets). Between 19.5 and 23.0°C, mean CPUE was lower (mean=0.54, maximum - 5.0, M=105), and above 23.0°C no porbeagles were caught Most pregnant females were caught in the cooler southern waters of New Zealand and Australia (Figs. 1 and 2 ), and some were taken from the subantarctic Auckland, Macquarie, Heard, and Kerguelen Islands ( 50— 54°S ). However, two were also caught in northeast New Zealand. For longline-caught females, SST was 10.2-17.2°C (mean 12.9°C, /i=32 ), and bottom depth at the capture locality was 600^300 m (mean=2104 m, n = \\). The two Heard Island pregnant females were taken by bottom trawl at depths of 248 and 259 m and bottom temperatures of 2.9 and 2.5°C, respectively. The Auckland Islands female was caught by midwater trawl at 160-164 m and a temperature of 11.9°C. Por- beagles have also been caught by bottom trawl near Macquarie Island at temperatures of 1°C (Williams^). Results Geographical distribution Porbeagles have a wide latitudinal distribution. In the New Zealand region, they range from the Kermadec Islands (30°30'S) to Macquarie Island (53°52'S) (Fig. 1). In the Australian FEZ, they range from near the Tropic of Capricorn in southern Queensland (23°44'S) to south of Tasmania (45°44'S) (Fig. 2). The large num- ber of capture records from northeast and southwest New Zealand, and around Tasmania, reflect concentra- tion of fishing effort, and not necessarily high shark densities. Porbeagles also occur near Heard Island (51-52°S), and Kerguelen Island (5rS) in the southern Indian Ocean (Duhamel and Ozouf-Costaz, 1982). Porbeagles were caught off southern Queensland (Fig. 2, north of 31°S) only in winter (June-Septem- ber), when water temperatures were lowest. Sea sur- face temperature (SST) at the time of capture of six sharks off Queensland in July-August 1997 was 21.3-21.6°C, about 4°C lower than normal. 3 Moss. S. A. 1995. University of Massachusetts, North Dart- mouth, MA 02747, USA. Personal commun. ■• Newton. A. 1996. The Marine Laboratory, P.O. Box 101. Aber- deen, Scotland. Personal commun. Length, weight, and growth The relationships between PL and FL (both in cm) for New Zealand porbeagles were as follows: PL = -1.366 + 0.907 FL FL - 1.990 + 1.098 PL (n=866, r2=0.995, range 61-223 cm FL, 54-208 cm PL The relationships between TL and FL (both in cm) for Australian porbeagles were as follows: rL = 4.165 -I- 1.098 FL FL = -0.567 -I- 0.881 TL (?i=173, /•2=0.967, range 63-180 cm FL, 71-202 cm TL) Length-weight data were available for 641 New Zea- land porbeagles (330 males, 309 females, and 2 un- sexed) over the range 61-228 cm FL and 3-153 kg weight. However 96.7% of the sample was less than 150 cm FL; therefore the results represent only juve- niles. There was no evidence from the raw data, or the residuals from a log-log regression, of a difference between the sexes. The regression equation for com- bined sexes was as follows: ^ Williams. R. 1997. Australian Antarctic Division. Tasmania, Aus- tralia. Personal commun. 46 Fishery Bulletin 98(1) 40 - 30 20 10 40 c 30 - 2 20 -I 10 80 60 40 20 ^fl Males {n I Females I Total {n = (n = 608) 50 75 100 125 150 175 200 225 250 Fork length (cm) Figure 3A Porbeagle length-frequency distributions by sex for April-July: (A) southwest New Zealand; (Bl northeast New Zealand; (Cl Australia. The bottom panel in each series includes some unsexed sharks, n = sample size. Triangles indicate mean lengths of the age classes identi- fied by MIX modal decomposition. hog^giweight) ^ -5.050 + 3.128 Logjt, (FL), (n= 641, r-=0.956) where weight is expre.ssed in kg and FL in cm. In New Zealand, length ranges were 64-228 cm for males and 61-206 cm for females (Fig. 3, A and B). Most were shorter than 150 cm. The size distribu- tions were similar for males and females up to 150 cm, and the sex ratio (both subregions combined) was not significantly different from one (M:F=0.93:1; ;t:^=1.99, P>0.1). However, males outnumbered females above 150 cm by 3.18:1 (;t:''^= 16.28, P<0.01). In southwest New Zealand, there was a strong modal peak at 79-93 cm for both sexes, and for males there were also clear peaks at 100-115 cm and 118-133 cm (Fig. 3A). The best MIX fit to the combined .sexes data consisted of three age classes whose mean lengths are shown in Fig. 3A. Sample sizes were small in north- east New Zealand; therefore no MIX model was ap- plied. A strong mode was present at 67-88 cm (Fig. 3B), and indistinct modes were present for both sexes at about the same position as the second and third modes in southwest New Zealand. In Australia, length ranges were 61-204 cm for males and 58-208 cm for females (Fig. 3C). Most were shorter than 150 cm, with a strong mode at 76-94 cm. The size distributions of males and fe- males were similar. The sex ratio of sharks smaller than 150 cm did not differ significantly from one ( 1.07:1; /2=i.47, p>0.1), but males outnumbered fe- males above 150 cm by 2.71:1 (;f'^=11.08,P<0.01). The best MIX fit to the combined sexes data consisted of five age classes whose mean lengths are shown in Figure 3C. Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 47 Males (n = 230) Females ;n= 133) 75 100 125 150 175 200 225 250 Fork length (cm) Figure 3B The mean lengths of the modes for southwest New Zealand and Australia were plotted against age, which was calculated from a theoretical birth date of 1 June (see below) and mean sampling dates of 5 May and 16 Jime, respectively (Fig. 4). Thus the ages classes were sampled near their respective birth dates. We interpret the five Australian modes as representing sharks that were recently bom, and 1^ years old, and the three southwest New Zealand modes as those representing sharks that were 1-3 years old. The first length mode in the northeast New Zealand distribu- tion was substantially shorter than the first southwest New Zealand mode and the second Australian mode, and we are uncertain about assigning an age to it. At age 1 year, southwest New Zealand and Australian por- beagles were similar in length, but for older ages New Zealand sharks were slightly larger Growth in both re- gions was linear over the range of the data (Fig. 4): Australia: FL = 65.4 + 16.1 (Age) {r^=0.997) Female length at maturity and reproductive development Maturity status was not recorded; therefore we cannot estimated length at maturity. However, 37 pregnant females ranged from 167^-199 cm (mean=185 cm), suggesting that female maturity is reached around 165-180 cm. The Macquarie Island female had four embryos 21.5-23.2 cm long. The right ovary was of the "inter- nal" type, which is typical of lamnid sharks (Pratt, 1988). It weighed 2.75 kg (2.359^ of total weight) and was undergoing active oogenesis. The entire ovary was packed with ova; other than a thin external envelope it had no macroscopically visible ovarian tissue. There was a single large efferent pore in the ovary, from which ova are shed (Stevens, 1983; Gilmore, 1993). southwest New Zealand: FL = 66.5 -I- 19.8 (.Age) (/•2=1.000) « FL of 167 cm was calculated from a PL of 151 cm. 48 Fishery Bulletin 98(1) 80 c 60 150 100 Males (n = 658) Females (n = 592) Total (n= 1327) 125 150 175 Fork length (cm) Figure 3C 200 225 250 The mean ovum count from six weighed subsamples was 72.2 ova per gram (SE=2.6), producing an esti- mated total number of 198,000 ova. Ova diameters were mostly 1.5-3.3 mm; a group of larger ova had di- ameters of 3.4-4.7 mm (Fig. 5). In addition to the em- bryos, the right uterus contained three egg capsules, and in the left uterus a 22.0-cm embryo had an egg capsule lodged in its mouth. Three of the capsules were empty and the fourth contained four ova (Fig. 6). Uterus width, estimated from a photograph contain- ing a ruler, was about 10 cm. In other females with near-term embryos, uterus width was about 20 cm. The anterior quarter of the uteiois had many longitu- dinal folds or pleats, and the rest was covered with small papillae and had a velvety texture. Litter size, embryonic growth, and gestation Data were obtained from 43 litters and 138 embryos (Table 1). All but four of the 40 litters for which litter size was known contained four embryos, two in each uterus. The exceptions were two litters reported by Graham ( 1939, 1956) with two and three embiyos re- spectively, one reported by Hanchet " with three em- bryos, and a New Zealand longline-caught litter with two midterm embryos. Mean litter size was 3.85. Of 132 embryos that were sexed, 73 were males and 59 were females, producing a sex ratio not significantly different from one (;^-=1.48, P>0.1). Regi-ession equations relating different embryo length measurements (in cm) were: PL = -0.125 -I- 0.885 FL TL^.^^=: 0.180 -I- 1.162 FL FL = -0.085 -I- 0.859 TL,,^, TL,,,.^ = 0.836 -f 1.170 FL FL = -0.644 + 0.853 TL Ilex (n=97,r-=0.999) (n=96, r"=0.998) (/!=96, r-=0.998) (,z=87,r-=0.998) (n=87, /•2=0.998) A log-log regi-ession of yolk-free embryo weight (kg) against FL ( cm ) gave 'Hanchet.S. 1996. National InstituteofWaterandAtmospheric [{(■search (NIWA). P. O. Box 893. Nelson, New Zealand. Personal commun. Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 49 160 E 140 - • Australia, length modes o SW New Zealand, length modes □ Aasen (1963) length modes 2 3 4 5 6 Age (years) Figure 4 Porbeagle mean lengths-at-age determined by MIX analysis of the south- west New Zealand and Australian data in Figure 3, A and C, with fitted linear regressions. Also shown are Aasen's ( 1963 ) northwest Atlantic length- frequency modes, which were plotted assuming a 9 April theoretical birth date ( 1.7 months earlier than the Southern Hemisphere) and a mean sam- pling date of 9 August. Table 1 Sources, numbers, collection periods and collection localities of Southern Hemisphere porbeagle embryos. NZ = New Zealand; NMNZ = Museum of New Zealand. Additional pregnant females reported by Graham (1956) were not included because of lack of data. Number of Number of Collection Collection Source litters embryos period locality Tuna longline 27 87 1992-98 New Zealand Tuna longline 5 17 1991-95 Tasmania, Australia Bottom trawl 2 8 1997 Heard Island Bottom trawl 1 4 1995 Macquarie Island Midwater trawl 1 4 1996 Auckland Island, NZ NMNZ P22122 1 3 1987 Auckland Island, NZ Hanchet' 2 2 1982-83 Otago, NZ Berquist^ 1 4 1998 Otago, NZ Graham (1939, 1956) 2 5 1933 Otago, NZ Duhamel and Ozouf-Costaz ( 1982) 1 4 1981 Kerguelen Island Total 43 138 ^ See Footnote 7 in tlie main text. - Berquist, R. 1997. Department of Zoology, University of Otago, P.O. Box 56, Dunedin, New Zealand, Personal commun Logjo (weight) = ^.719 + 2.916 Logjo (FL) (« = 100, r2=0.974) The Kerguelen Island embryos reported by Duhamel and Ozouf-Costaz ( 1982 ) were the smallest in our sam- ple, with estimated FLs of 9.6-10.4 cm. Our largest em- bryos were in litters of 62.0-67.0 cm, 64.2-65.6 cm, and 64.3-66.6 cm. Typically, all embryos in a litter were sim- ilar in length, but two litters each contained one unusu- ally small embryo (runt) (Table 2). Variation in length within a litter increased with mean length, but the per- centage range in length was relatively constant. 50 Fishery Bulletin 98(1) Table 2 Fork lengths and weights of stomach contents, intestinal contents, and livers of embryos from two litters that contained one unusually small embrvo. Litter 1 Litter 2 Embryo fork lengths (cm) Length range ( cm ) Length range as percentage of mean length (%) Smallest embr\'o Stomach contents (kg) Intestine contents (kg) Liver (kg) Other three embryos Stomach contents (kg) Intestine contents (kg) Liver (kg) 22.5, 27.7, 29.0, 30.8 51.4, 55.3, 55.3, 63 8.3 12.4 30.2 22.0 0.01.5 0.077 0.007 0.069 0.003 0.056 0.117-0.440 0.18.3-0.340 0.114-0.296 0.092-0.130 0.006-0.017 0.151-0.364 All embryos were collected between 11 March and 16 July, reflecting the seasonality of the longline fisheries. The mean length of Southern Hemisphere embryos in a litter in- creased significantly (P<0.01 ) with sampling date (Fig. 7): Mean FL i cm) 9.36 + 7.48 month, {«=39,r2=0.27) where "month" is defined as the number of months elapsed since 1 January. There was considerable unexplained variability. For ex- ample, in April, mean embryo length varied between 20.3 and 63.5 cm. Inspection of the data by year of collection and sampling lo- cation showed that the variability was not caused by interannual or spatial differences. A regression equation fitted to North Atlan- tic embryo data (Fig. 7) was Mean FL (cm) 23.51 + 6.78 month {n=29, 7-2=0.70) Diameter (mm) Figure 5 Frequency distribution of ovarian ova diameters from a pregnant por- beagle from Macquarie Island, n = sample size. The North Atlantic regression explained a much higher proportion of the variation but displayed con- siderable length variability in early gestation. A homo- geneity of slopes test showed that the regression slopes for the two hemispheres were not significantly differ- ent (P=0.76). The pooled data had a regression slope (=embryonic growth rate) of 7.1 cm per month, and the regression intercepts differed significantly (analysis of covariance, P=0.004) by 12.0 cm, which is equivalent to a temporal displacement of about 1.7 months. Small, postnatal juveniles 58-68 cm long (72=53), were obsei-ved on Australian longlines between 22 May and 9 September, with a mean capture date of 15 July. Large embryos (up to 67 cm) were observed be- tween mid-April and mid-June (Fig. 7). In the South- em Hemisphere, the large variation in embryo length at any one time, and the long period over which small juveniles, assumed to be newborn, were collected, in- dicate that the parturition period is lengthy. Parturi- tion probably peaks in June-July (winter) but may Francis and Stevens: Reprodurtion, embryonic development, and growth of Lamna nasus 51 Figure 6 Four empty egg capsules found in the uteri and in the mouth of an embryo, from a pregnant porbeagle from Macquarie Island. Also shown are several loose ova found in one of the egg capsules. The largest egg case was 75 mm long. extend from Ap:il to September. For aging pur- poses, we defined the theoretical birth date as 1 June. Based on the lengths of the smallest juve- nile and the largest embryo, the length at birth is 58—67 cm. If a growth rate of 7.48 cm per month is maintained by Southern Hemisphere porbeagles throughout gestation, the gestation period is about 8—9 months. However, the un- explained variability in Fig. 7 compromises our ability to accurately estimate the gestation period. Embryonic development Porbeagle embryos develop the distended yolk stomach that is characteristic of all oophagous lamnid sharks (Fig. 8). In the Kerguelen em- bryos (9.6-10.4 cm), such distension was al- ready apparent. The caudal fin was notably curved, with the upper lobe much longer than the lower lobe, and there were no external gill filaments. At 19.8-20.7 cm, the bulging yolk stomach was the most noticeable feature, along with a marked lateral expansion of the head and branchial region (Fig. 8A). The body lacked pig- mentation (except for the eyes), and appeared pink because of the presence of blood vessels under the 70 - B ^° ^ O 40 - £ 30 - o - 20- (0 0) 2 10- New Zealand Tasmania -^ Heard Island □ Kerguelen Is NW Atlantic • NE Atlantic /^ y y / ° o f • Monthl four So and th( r-'=0.70 Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Month Figure 7 y variation in the mean lengths of porbeagle embryos from jthem Hemisphere locations (white symbols, n=39, r^=Q.21), i northwest and northeast Atlantic (black symbols, n=29, . skin. At 34.2 cm, the yolk stomach had become enor- mously distended, and measured 22.6 cm long by 15.9 cm high; the branchial and throat regions remained 52 Fishery Bulletin 98(1) 03 CO d CO n 00 CO 3 u at Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 53 B iX Figure 9 Anterior view of the heads of (A) 9.6-cm embryo, and (B) 26.4 cm female embryo showing the functional fangs. (Photo A. by G. Duhamel.) swollen, and the upper body and pectoral fins had be- come pigmented (Fig. 8B). At 40.3 cm, pigmentation was essentially the same as in postnatal porbeagles, the swelling of the head had disappeared, and the yolk stomach had begun to shrink (Fig. 8C). At 58.0 cm the juvenile body form had been attained, apart from an enlarged abdomen (Fig. 8D). Other embryos around this size and larger had a more streamlined shape, with little noticeable abdominal distension. Distension of the abdomen during early develop- ment causes the subdermal muscle layers to split along the ventral midline, extending anteriorly as far as the fifth gill slits. The expanding stomach protrudes between the muscle layers and stretches the abdomi- nal skin. Later, the stomach shrinks back inside the muscle layers, and the stretched skin returns to its original shape. A distinct "scar" remains in the ventral midline in the area between the origins of the pectoral fins and the fifth gill slits, marking the anteriormost point of the split muscle layers. Small embryos had large, erect, tubular, recurved "fangs" in both jaws (Fig. 9, A and B). These teeth, which were quite unlike those found in postnatal por- beagles, were clearly functional. In the Kerguelen em- bryos (9.6-10.4 cm), the tooth formula was (1-t- 1/1+1), and the lower teeth were massive in relation to mouth size (Fig. 9A). Larger embryos (19.8-38.3 cm) had more functional upper teeth (3+3/1+1) (Fig. 9B). Ad- ditional minute teeth were visible under a light micro- scope, but they appeared vestigial and nonfunctional and were not included in the tooth formula. Replace- ment fangs were present behind the functional series, but they were irregularly spaced and usually located between the functional teeth. Oval scars on the gum of both jaws external to functional fangs indicated that fangs are progressively shed and replaced. The largest 54 Fishery Bulletin 98(1) O) 3 - 11) § 2 1 - Variation embryos. embryo with fangs was 38.3 cm, and the smallest embryo without fangs was 33.9 cm. In the range of overlap, there were 12 embryos with fangs and 8 without fangs. Therefore, the fangs are shed between 34 and 38 cm. Most of the larger embryos without fangs possessed several series of develop- ing, nonerect, nonfunctional teeth shaped like those found in postnatal porbeagles, except that they lacked lateral cusps. One near-term litter had three embryos with nonfunctional upper teeth but partially erect lower teeth, whereas the fourth had nonfunctional teeth in both jaws. The stomach contents consisted wholly or mostly of viscous, amorphous, light yel- low yolk. In many embryos there were also discrete masses of clear, white or greyish gelatinous material, probably the remains of empty egg capsules, embed- ded in the yolk. This gelatinous material usually represented less than 10% of the stomach contents and has been included in the yolk weights reported below. Occa- sionally we found shed fangs in the stom- achs, but the thick glutinous nature of the yolk made them difficult to find, and it was impossible to assess their frequency or abundance. Embryonic total weight increased rapidly between 20 and 35 cm, changed little between 35 and 50 cm, then increased again during the rest of the gestation period (Fig. 10). Embryonic weight minus yolk weight increased steadily throughout gestation. The weights of five fi-ee-living juveniles shorter than 70 cm were similar to the yoLk-fi-ee weights of the largest embryos. The weight of yolk in the stomach increased steadily between 20 and 30 cm, before increasing rapidly to peak at 30^2 cm (Fig. 11). Yolk weight at 30-42 cm varied from 0.39 kg to 1.82 kg, representing 26.7-80.8% of total body weight. Absolute and per- centage yolk weight both generally declined at lengths greater than 42 cm. Three embryos longer than 60 cm still had around 1 kg of yolk in their stomachs, but it represented a low proportion of their total weight (ca. 17-22%^). All embryos longer than 50 cm had yolk or gelatinous material in their stomachs, suggesting that the yolk may not be completely digested before birth. The spiral valve of the intestine contained a thick, gritty, greenish-brown sludge that is the waste prod- A Embryo total weigtit • Embryo weigtit minus yolk O Juvenile total weigtit ^ A A^ AA A ^ «i^ A A A If* A .V 10 20 30 40 Fork length (cm) 50 60 70 Figure 10 in total weight and yolkfree weight with length for porbeagle Also shown are total weights of five small juveniles. * Not shown in the percent yolk weight panel of Fig. 11 because embr>'o total weight.s were not mea.sured accurately. uct of yolk digestion. The smallest embryos dissected ( 19.8-20.7 cm) contained small amounts of waste, and the quantity of waste increased steadily throughout gestation (Fig. 11). Intestinal contents composed the greatest percentage of yolk-fi-ee weight at 30-50 cm. Newborn porbeagles had smaller quantities of intes- tinal waste than large embryos, suggesting that the waste is retained until after birth. Liver weight increased exponentially with FL, show- ing the most rapid increase above 55 cm (Fig. 11). Relative liver weight also increased to a maximum in the longest embryos. Postnatal porbeagles usually had smaller livers, both in absolute and relative terms, than those of the largest embryos, indicating that en- ergy reserves stored in the liver are consumed by the young after birth. The runts in two litters had low stomach and intes- tine contents, and liver weights (Table 2). The 22.5-cm runt in litter 1 had numerous short (1-3 mm) lacera- tions on its distended abdomen, and a few elsewhere on its body, presumably inflicted by the teeth of its larger sibling. However none of the gashes had pen- etrated the body cavity, and they did not appear to be life-threatening. Five pairs of uteri containing embryos were ob- tained. The orientation of the embryos could be deter- mined for only five of the 10 uteri; these five uteri con- tained embryos 38.2-62.6 cm long. In all cases, the two embryos were facing in opposite directions. Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 55 2.0 1.5 1.0 0.5 0,15 - 0-05 - 00 Yolk weight V • Embryos o Juveniles 20 30 40 Intestine contents weight 50 60 • •• ^ i /*^ vi*^ Percent yolk weight ••• ...V 80 60 40 - 20 20 30 40 50 60 • Percent intestine contents weight 20 30 40 50 60 12 Fork length (cm) Figure 11 Variation in the weight of yolk, intestine contents, and liver with length (left panels), and the percentage of the total weight contributed by yolk, and percentages of yolkfree weight contributed by intestinal contents and liver (right panels) for porbeagle embryos. Comparable data for five small juveniles are also shown Discussion Geographical distribution Porbeagles occur throughout the New Zealand EEZ and the southern half of the Australian EEZ. On the east coast of Australia, they reach subtropical waters ( to 23°44'S), but only during winter. At the other extreme, porbeagles are found in subantarctic waters in the In- dian and southwest Pacific Oceans, reaching almost 54°S. Our results are consistent with observations of porbeagles between about 28 and 58°S across the entire South Pacific between New Zealand and Chile (Yatsu, 1995). Porbeagles are caught only north of 30°S in win- ter-spring (August-November), and they penetrate far- ther south during summer and autumn (Yatsu, 1995). Sea surface temperature at locations where porbea- gles were caught was 9.9-22.6°C, but catch rates were low above 19.5°C. This is similar to SSTs reported pre- viously for porbeagles in the South Pacific by Stevens et al. ( 19831 (7.&-22.8°C, with most captures less than 16.7°C) and Yatsu (1995) (5-20°C with most captures less than 15 °C). Temperatures at the actual depth of capture would be similar to these because most long- lines fished at 50-100 m depth, which is shallower than the depth of the mixed ocean layer in autumn- winter. Bottom temperatures for trawl-caught porbea- gles were as low as 1-3°C, which is consistent with temperatures of 1.7°C (Svetlov, 1978) and 3.1-t.l°C (Templeman, 1963) reported elsewhere. Thus the tem- perature range inhabited by porbeagles in the South- em Hemisphere is probably about 1-23°C, with abun- 56 Fishery Bulletin 98(1) dance declining above about 19°C. The preferred tem- perature in the North Atlantic is less than 18°C (Aasen, 1963). Porbeagles can maintain their body temperature up to 11°C above that of the surrounding water (Carey et al., 1985). Among the small group of endothermic sharks, they are exceeded in this capacity only by salmon sharks (Carey et al., 1985). The apparent pref- erence for higher latitudes by large (Yatsu, 1995) and pregnant porbeagles may indicate an increased ability to thermoregulate in larger sharks. High body tem- perature is probably necessary for lamnid sharks to function as active predators of fast-moving prey in cold water (Goldman, 1997). In the North Atlantic, porbeagle abundance varies seasonally and spatially (Aasen, 1961, 1963; Temple- man, 1963; Mejuto and Garces, 1984; Mejuto, 1985; Gauld, 1989), and there are indications of seasonal variability in their vertical distribution ( Bigelow and Schroeder, 1948; Aasen, 1961, 1963). Limited tagging results show that they are capable of movements up to 2370 km (Aasen, 1962; Stevens, 1976, 1990). In combi- nation with evidence of seasonal latitudinal migration in the South Pacific (Yatsu, 1995), this range of move- ment suggests that Southern Hemisphere porbeagles may exhibit complex seasonal, spatial, and length-re- lated distribution patterns. Our data were collected mainly during April-July, and therefore provide little information on seasonality. Length, weight, and growth Several length-weight relationships have been pub- lished for the North Atlantic (Aasen, 1961; Mejuto and Garces, 1984; Gauld, 1989; Stevens, 1990; Ellis and Shackley 1995; Kohler et al., 1995). All were based on small samples, except those of Gauld (1989), who found a significant difference between males and fe- males above 180 cm. Our New Zealand sample com- prised mainly sharks less than 150 cm, and compared with the length-weight relationships of Mejuto and Garces (1984), Gauld (1989) and Kohler et al. (1995), our length-weight relationship rises too steeply be- yond 150 cm. Our relationship should therefore not be extrapolated beyond 150 cm. The largest porbeagles in our samples were 228 cm (male) and 208 cm (female). The maximum length reached in the North Atlantic is not clear. Maxima of 253 cm (288 cm TL) and 278 cm (317 cm TL) for Scot- tish males and females, respectively, appear to be the largest reliable measurements (Gauld, 1989). Lengths of 294-325 cm ("estimated ... 11 feet", "ca 11-12 feet", or "340-370" cm TL) reported by McKenzie ( 1959) and Templeman (1963) were obviously not measured ac- curately and may have been overestimated. A TL of 12 feet equals 365.8 cm TL, which likely forms the basis for the maximum length of 365 cm TL reported by Pratt and Casey (1990). If Southern Hemisphere porbeagles grow as large as those in the North Atlan- tic, our samples do not include the larger size classes. Tuna longlines catch shortfin makos that exceed 300 cm FL (senior author, pers. obs.); therefore, they should be capable of retaining large porbeagles. The latter evi- dently inhabit latitudinal or depth ranges outside our sampling area, which is inhabited mainly by juveniles. Males and females were equally represented at lengths up to 150 cm. At greater lengths, males sig- nificantly outnumbered females by about 3:1 in both New Zealand and Australia. Skewed sex ratios have been reported fi-equently in the North Atlantic, in fa- vor of males (Mejuto and Garces, 1984; Mejuto, 1985; Ellis and Shackley 1995), females (Gauld, 1989), or either sex depending on the length range or sample (Aasen, 1963; O'Boyleetal., 1996). Aasen (1963) found that the overall sex ratio in large samples was close to 1:1. These results indicate that juveniles do not segre- gate by sex, but that larger sharks do. MIX analysis discriminated 3 and 5 length modes respectively in southwest New Zealand and Australia, which we interpret as age classes. In northeast New Zealand, the mean sampling date was 25 June, over seven weeks after the mean sampling date for south- west New Zealand; therefore the first mode in the for- mer could represent new-bom sharks. Alternatively, it may represent slow-gi'owing one-year-olds. The latter interpretation is supported by the similarity of the po- sitions of modes 2 and 3 in both northeast and south- west New Zealand. Juveniles gi'ow linearly and rapidly, reaching 110- 125 cm FL in three years. They may grow slightly faster in southwest New Zealand (20 cm/year) than in Australia ( 16 cm/year), but the standard errors were large for New Zealand 2- and 3-year-olds, and the com- parison could be biased by incorrect determination of the number of modes in the length-fi-equency data. The modal lengths for Australian juveniles agree well with the first four modes for northwest Atlantic porbeagles presented by Aasen (1963) (Fig. 4), providing that our modes represent age classes. Aasen (1963) presented two growth curves — one based on length-fi-equency modes, and the other on back-calculated lengths-at-age ft-om a single vertebra fi-om a 226-cm female. ^ His two growth curves were practically identical. Our growth estimates also agi'ee with length-incre- ment data for five tagged northeast Atlantic porbea- ^ In a preliminary report, Aasen (1961) presented vertebral age data for 50 porbeagles. However the mean lengths-at-age differed sub.stantially from those presented later, and Aasen ( 1963) stated that his earlier results "were not very accurate." Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 57 gles (Stevens, 1976. 1990). Their lengths at tagging and recapture were only estimated; therefore growth increments were approximate. The sharks were rela- tively small when released (84-105 cm), and were at liberty for 0.75-13 years. Annual growth increments were about 9-32 cm, with a mean of 20.4 cm. Despite good agreement among all sources of porbeagle growth data, a vertebral-based growth curve from adequate sample sizes and covering the full size range is still required for both hemispheres. Longevity is unknown, but Aasen (1963) aged his 226 cm female as 19-i- years and suggested that they may live around 30 years. This conclusion needs con- firmation because longevity is often used to estimate the natural mortality rate, which is an important pa- rameter in population models. Length at maturity and reproductive development The lengths of pregnant females (167-199 cm, mean 185 cm ) suggest that females mature at about 165-180 cm. This estimate is consistent with reports of a 191 -cm mature female from the South Atlantic (Svet- lov, 1978), and immature females of 138, 139, and 150 cm from around New Zealand (Stevens et al., 1983; Dufiyi°). The length at maturity of North Atlantic females is controversial. Shann (1911) reported two pregnant females of "about five feet long" (152 cm TL, or 133 cm FL). In both cases, the length estimate was probably third-hand, and it is unlikely that the females were measured. That estimate of length at maturity, which we believe to be unreliable and too low, has permeated the literature ( Bigelow and Schro- eder, 1948; Compagno, 1984; Last and Stevens, 1994). Other authors have reported a wide maturity range of 175-220 cm FL (Aasen, 1961, 1963; Pratt and Casey, 1990). Templeman ( 1963) and Moss^ reported females of 191 and 203 cm to be immature. The smallest ma- ture females reported by Templeman ( 1963 ) and Gauld (1989) were 203 and 196 cm respectively, and Pratt (1993) examined one of 227 cm. Aasen (1961) showed that uterine width increased rapidly in females longer than 197 cm. These results suggest that North Atlan- tic females mature at about 195-205 cm (218-229 cm TL ), which is higher than the range determined for the Southern Hemisphere. A similar between-hemisphere difference in length at maturity has been found for shortfin makos (MoUet et al.^^). 1° Duffy, C. 1998. Department ofConservation, Private Bag 3072. Hamilton, New Zealand. Personal commun. " MoUet, H. R, G. Cliff, H. L. Pratt, and J. D. Stevens. 1998. Reproductive biology of female shortfin mako Isurus oxyrinchus Rafinesque 1809. H. F. Mollet, Monterey Bay Aquarium, Mon- terey, California 93940, Unpubl. manuscript. There was no information on length at maturity in males in our data. On the basis of changes in clasper length and calcification. North Atlantic males appar- ently mature at a smaller size than do females, in the range 131-175 cm (150-200 cm TL) (Aasen, 1961; El- lis and Shackley, 1995). The "internal" type ovary from our Macquarie Is- land pregnant female conformed with the morphol- ogy found in all lamnid and alopiid sharks examined so far (Pratt, 1988). It weighed 2.75 kg (2.35% of total weight). The ovary of Swenander's (1906, 1907) 246-cm North Atlantic pregnant female measured 41 by 28 cm, and weighed 6.3 kg, or 3.6% of estimated total weight. Mean embryo lengths in the two females were 22.1 cm and about 25 cm respectively. The em- bryo stomach contents peak at 30-42 cm, suggesting that ovarian size and ovulation may peak when em- bryos are about 25-30 cm long. In the shortfin mako, the ovary of an actively ovulating female with early- term embryos weighed about 5% of her total weight, whereas the ovaries of females carrying near-term embryos were spent and weighed as little as 0.2-0.3% of total weight (Mollet et al.''). In sandtiger sharks (Carcharias taurus), relative ovarian weight peaks at 6—7% of body weight, and then declines during the second half of gestation (Gilmore et al., 1983; Gilmore, 1993; Mollet et al.''). A low relative ovary weight was also reported in a longfin mako with near-term em- bryos (Gilmore, 1983). The size-frequency distribution of ova from the Mac- quarie Island female indicated that they are ovulated at around 4—5 mm. Ova diameters measured in two pregnant North Atlantic females after preservation in 10% formalin were mainly in the range 2.3-4.3 mm, with the largest measuring 6.0 mm (Moss^). Swenander (1906, 1907) reported ova diameters be- tween 1 and 5-6 mm in a North Atlantic pregnant fe- male, and encapsulated 4—5 mm ova in the uteri of an- other female. Maximum ova diameters in other ooph- agous sharks range between 4 and 10 mm (Springer, 1948; Bass et al., 1975; Otake and Mizue, 1981; Gilm- ore, 1983; Gilmore et al, 1983; Stevens, 1983; Uchida et al., 1996; Chen et al., 1997). Empty and near-empty egg capsules were found in the uteri of the Macquarie Island female, and in the mouth of one of its embryos. Moss'^ also found empty egg capsules in the mouths and gill slits of embryos measuring 33.8 and 36.2 cm. Apparently, embryos are capable of rupturing egg capsules and swallowing the contents, although the occasional presence of ge- latinous material resembling egg capsules in embryo stomachs (Swenander, 1907; this study) suggests that whole or empty capsules are sometimes swallowed. Swenander (1906, 1907) found over 40 egg capsules, each measuring about 80 by 15 mm and containing 58 Fishery Bulletin 98(1) 21-28 individual ova, in the uteri of a female that con- tained four embryos about 4.3^.7 cm long. Duhamel and Ozouf-Costaz (1982) found 102 nonfertile eggs in the uteri of their Kerguelen Island female, which car- ried embryos of 9.6-10.4 mm. No photographs were taken of the eggs, and it is unclear whether they were egg capsules or individual ova (DuhameP). The emerging picture for oophagous sharks is that large numbers of nutritive egg capsules accumulate in the uteri during the early stages of gestation, and they are rapidly consumed as the embryos grow large enough to puncture and eventually swallow them ( Gr- uber and Compagno, 1981; Otake and Mizue, 1981; Gilmore, 1983; Gilmore et al., 1983; Mollet et al."). Ovulation peaks during midgestation, but full egg cap- sules are rarely found in the uteri, probably because they are eaten soon after entering the uteri. During later gestation, ovulation ceases and embryos metabo- lise their accumulated stomach contents for energy, growth, and storage in the liver. Litter size, embryonic growth, and gestation In 36 out of 40 of our pregnant females, litter size was four, with a mean of 3.85 embryos. Litters of two or three were occasionally recorded. In the North Atlantic, Shann (1911, 1923) stated that litters com- monly consisted of two embryos (range 1-4), but his data almost certainly included several partial litters (see Table I in Shann, 1923). Templeman ( 1963) found three, four and four embryos in his three litters, and Gauld ( 1989) found four to be the most common num- ber of embryos in a litter, with a mean of 3.7 (n = 12).^- One litter of five has also been reported ( Bigelow and Schroeder, 1948). Our sample of Southern Hemisphere litters is the largest yet assembled, and Gauld's ( 1 989 ) is the largest fi-om the North Atlantic. Both samples had very similar mean numbers of embryos. We con- clude that litter size is usually four, but smaller lit- ters are occasionally found; litters larger than four are extremely rare. Some litters with fewer than four em- bryos were probably incomplete. Abortion of embryos during capture is common among nonlamnid sharks, but it is difficult to imagine midterm embryos with grossly distended abdomens being aborted, even when the mother is compressed in a trawl net. One of our New Zealand longline litters containing two midterm embryos, 37.5 and 38.3 cm long, was presumably com- plete. Abortion of near-term embryos is quite possible. '^ Gilmore's Table 1 contains a number of errors. The reference to Aasen (19661 in the Lamna nasus section should presumably be Aasen (19631; the two Swenandcr il907) litters contained four embryos each rather than 2; and the '219-cm TL shark with four embryos attributed to Nakaya ( 19711 was actually a male, and no embryos were mentioned by Nakaya. , Linear regressions fitted to the length-month data for both hemispheres suggest that embryos have a rapid growth rate of about 7 cm per month, but there was much unexplained variability (Fig. 7). The esti- mated growth rate is almost twice that of shortfin mako embryos (3.7 cm per month) (Mollet et al.^'). The length of the gestation period appears to be about 8-9 months in both hemispheres. During July-Sep- tember, Aasen ( 1963) found no embryos in the north- west Atlantic despite examining hundreds of mature females. He argued that the gestation period was 8 months, and that the females he examined were un- dergoing a short rest period between pregnancies. Our interpretation agrees with that of Aasen ( 1963). A contrary hypothesis involving a gestation period of 1-2 years has been advanced by Shann ( 1923 ) and Gauld (1989). They argued that the high variability in embryo length and the apparent presence of two cohorts of embryos were inconsistent with a gestation period of less than one year. We cannot rule out their hypothesis, and we are conscious that our Southern Hemisphere data are limited in seasonal scope and that pooling data across locations and years is not de- sirable. However, we believe the data from both hemi- spheres are most consistent with a gestation period of less than one year. The implied rapid embryonic growth rate is not unreasonable given the abundant embryonic food supply and the relatively high growth rates of postnatal juveniles discussed above. The high variability in the embryo length data might be explained by an extended mating period. In the northwest Atlantic, Aasen (1963) found males with seminal vesicles that were filling at the end of August, indicating that mating would begin in Sep- tember. Pratt ( 1993) reported a mature female caught in October with moderate amounts of spermatozoa in the oviducal gland and with ft-esh vaginal abrasions. A mature male with haematose claspers was caught on the same longline, providing strong evidence of mat- ing in October. Gauld ( 1989) found females with fresh bite marks, thought to be inflicted during mating, on females near the Shetland Islands in December-Jan- uary. These observations suggest that mating lasts from September to January in the North Atlantic. If our hjTDothesis of rapid embryonic growth and high intracohort variability is correct, parturition probably peaks in June— July (winter) in the Southern Hemisphere and possibly extends ft-om April to Sep- tember. Parturition in the Northern Hemisphere may peak around two months earlier (spring-summer). The presence of distinct length modes in juvenile length-fre- quency distributions from New Zealand, Australia, and the northwest Atlantic (Aasen, 1963 ) confirms that par- turition is restricted to part of the year, rather than oc- curring year-round. Svetlov ( 1978) reported the capture Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 59 of a female in the South Atlantic in March and stated: "The shark had recently spawned, to judge by external characters (an inflamed area of the body around the anus) and the state of the ovaries." He did not elaborate further on the ovaries, so it is difficult to assess his re- port. If the female was postpartum, the timing is one month earlier than our suggested parturition period for Southern Hemisphere porbeagles. Another possibility is that the inflammation around the cloaca may have been the result of recent mating. If the gestation and rest periods combined last one year, and females reproduce every year, annual fecun- dity is 3.85 young per female. If there is a resting pe- riod of just over one year between pregnancies, an- nual fecundity would be half that amount. It is there- fore important to determine whether Aasen's (1963) suggestion of a one-year cycle is valid for both hemi- spheres. The gestation period in shortfin makos is thought to be 18 months, with a reproductive cycle of three years (Mollet et al.^M. Interestingly, the greater average fecundity of makos would result in an annual fecundity of about four per female (Mollet et al.^M, which is similar to the reproductive output of porbea- gles assuming a one-year cycle. For Southern Hemisphere porbeagles, the length at birth was estimated to be 58-67 cm (68-79 cm TL), based on the lengths of the largest embryos and the shortest postnatal juveniles. In the North Atlantic, the largest reported embryos were 60-64 cm (Pen- nant, in Shann, 1911) and 65 cm (Gauld, 1989). Big- elow and Schroeder (1948) reported a 55.7-cm speci- men (66.0 cm TL; USNM 47528), but it was an em- bryo rather than a postnatal juvenile (Williams^'^). The embryo would have been about 58 cm before pres- ervation. Postnatal porbeagles of 66 and 70 cm were reported by Imms and Day respectively (in Shann, 1911). These observations indicate that porbeagles are bom at about the same length in both hemispheres. Embryonic development We assembled a comprehensive series of porbeagle embryos ranging from early gestation (9.6 cm long) to full-term, enabling us to describe the main mor- phological changes that occur during gestation. Pre- vious studies that described and illustrated embryos were based on only a few embryos, most of which were midterm (Swenander, 1906, 1907; Shann, 1911; Nordgard, 1926; Bigelow and Schroeder, 1948; Tem- pleman, 1963). The following review of embryonic de- velopment is derived from our observations, supple- mented by literature reports. 13 Williams, J. T. 1997. National Museum of Natural History. Washington DC 20560-0159. Personal commun. Embryos 4.3^.7 cm long have external gills and well-developed branchial regions (Swenander, 1907). They have nearly absorbed their yolk sacs and have a large spiral valve, but there is no yolk in the digestive system. Swenander ( 1907) stated that "these embryos are too small to be able to swallow entire egg capsules and their teeth are not sufficiently developed to tear open the egg capsules." They have not begun to feed at this stage, despite the large number of egg capsules present in the uteri, but precocial teeth have already formed. At 10 cm, the lower jaw contains two rela- tively massive fangs (Fig. 9A) that appear capable of tearing open egg capsules. The upper teeth are much less developed and there is only one functional tooth on each side of each jaw. Lower jaw teeth also de- velop earlier than upper jaw teeth in sandtiger sharks (Hamlett, 1983). The abdomen is swollen and the em- bryos have presumably begun feeding (the stomachs were not dissected). Large numbers of egg capsules are present in the uteri. The external gills have been resorbed. By 15 cm, the abdomen is distended, and the head and branchial region are gelatinous and grossly enlarged (Bigelow and Schroeder, 1948). Between 20 and 42 cm, development is dominated by the consumption of large numbers of egg capsules, leading to a great increase in the relative size and dis- tension of the yolk stomach (Swenander, 1907; Shann, 1911, 1923; Bigelow and Schroeder, 1948; Templeman, 1963; Moss''; Fig. 8). Large fangs are present in both jaws (Swenander, 1907; Templeman, 1963; Moss'^; Fig. 9) and are used to open the egg capsules before re- moval of the contents; how this is accomplished is un- known. At 30-42 cm, yolk accounts for up to 81% of total weight (Fig. 11; Templeman, 1963). The stomach yolk in midterm embryos of shortfin makos peaks at about 60-70% of total weight (Mollet et al." ). Relative and absolute amounts of yolk in porbeagles and other oophagous sharks decline during the rest of gestation (Fig. 11; Molletetal.il). Porbeagle embryos shed their fangs at about 34—38 cm. Embiyonic fangs (the "emb" teeth of Gilmore, 1993) have also been reported in salmon sharks, common and bigeye thresher sharks (Alopias vulpinus and A. su- perciliosus), shortfin makos and sandtiger sharks (Loh- berger, 1910; Bass et al., 1975; Gruber and Compagno, 1981; Gilmore et al., 1983; Hamlett, 1983; Gilmore, 1993; Chen et al., 1997). They probably occur in all oophagous species, and in at least some species, they are shed part-way through gestation. In bigeye thresher sharks, fangs appear at about 11 cm TL and disappear at about 60 cm TL (Chen et al., 1997). In sandtiger sharks, they appear at 4-5 cm TL, and are lost some time before birth (Gilmore et al., 1983; Hamlett, 1983). We suspect that female porbeagles cease ovulation at about the time the embryonic fangs are lost, as 60 Fishery Bulletin 98(1) has been reported for shortfin makos and sandtigers (Gilmore et al., 1983; Mollet et al."). The embryos then rely on the yolk stored in their stomachs to provide the energy needed for growth and respira- tion during the rest of the gestation period. However it is also possible that females continue ovulating, and that the toothless embryos feed by swallowing whole egg capsules or by squashing them in their mouths. Whole egg capsules have been reported from the stomachs of near-term embryos of the bigeye thresher shark (Gilmore, 1983; Moreno and Moron, 1992), but not from near-term lamnid sharks. Clari- fication of this point requires examination of the ova- ries of near-term females to assess their ovulatory state. Above 35 cm, the waste products of yoLk digestion continue to accumulate in the intestine. The greenish coloured waste is characteristic of oophagous sharks (Swenander, 1907; Lohberger, 1910; Shann, 1923; Springer, 1948; Uchida et al., 1996). The gritty na- ture of the intestinal contents was also mentioned by Swenander ( 1907) and has been reported to consist of "crystal-like pieces" in white shark embryos (Uchida et al., 1996). The composition of this material is un- known. The liver grows most rapidly in the second half of gestation as energy reserves are transferred to it for storage. An increase in the relative weight of the liver in larger embryos has also been observed in shortfin makos (Mollet et al.'^). During the second half of gestation, several series of "post-natal" teeth develop, but they are folded back in the jaws and are nonfunctional. In white sharks, some of these teeth are shed and swallowed by the em- bryos (Francis, 1996; Uchida et al., 1996). The teeth probably become erect near or soon after birth, as has been found in near- term white shark embryos (Fran- cis, 1996; Uchida et al., 1996). Typically, embryos in a porbeagle litter are of sim- ilar size, but occasionally a large size range is en- countered. Gauld ( 1989) found one litter with embryos ranging ft-om 55.6 to 65.0 cm, and Shann (1923) re- ported a litter with a range of 38.1-50.9 cm. The runts in our two litters had small quantities of stomach and intestinal contents, and small livers, but were other- wise developing normally. This suggests that sibling competition may occur when a dominant embryo with its snout nuzzled into the anterior end of the uterus consumes most of the egg capsules as they pass into the uterus, leaving few for its sibling. However, all four embryos are usually adequately nourished, and the two embryos in each uterus are usually oriented in opposite directions. This suggests that the direc- tion of orientation within the uterus may be a problem only if the mother is unable to produce enough egg capsules to satisfy both embryos. - At birth, embryos may still have yolk in their stom- achs. Near- term white shark embryos have been re- ported with either empty (apart from some ingested teeth and denticles) or yolk-filled stomachs (Francis, 1996; Uchida et al., 1996). Near-term shortfin mako embryos "and new-bom sandtiger sharks may also have small amounts of yolk in their stomachs (Cade- nat, 1956; Bass et al., 1975; Gilmore et al., 1983; Mol- let et al.'M. Along with the energy stored in the liver, this yolk supplies the nutritional needs of the embryos until they learn to feed. However, the livers of por- beagle embryos never exceeded 10^ of the yolk-free embryo weight, compared with 13.5-18.6'^ in near- term white shark embryos (Francis, 1996; Uchida et al, 1996). Hubbs ( 1923) reported a 9.1 kg (20 lb) embryo col- lected in late August in Maine, USA. The weight is clearly too large to be a porbeagle embryo because they are not known to exceed 5 kg ( Fig. 10 ). Moss-^ sug- gests that the embryo may have been ft-om a sandtiger shark. The presence of an "umbilical scar" or "yolk sac scar" in postnatal oophagous sharks has puzzled many sci- entists who were aware that the embryos have no pla- cental connection to their mothers ( Gilmore, 1983; Ste- vens, 1983; Klimley, 1985; Cliff et al, 1990, 1996; Pratt and Casey, 1990; Moreno and Moron. 1992; Francis, 1996; Uchida et al., 1996). Our observations show that distension of the stomach stretches the abdominal skin and separates subdermal muscle layers as far forward as the fifth gill slits. As yolk is consumed the stomach shrinks and the muscle layers return to their original position, leaving a scar in the pectoral-gill re- gion. The scar is sometimes faint or absent. In all lamnid sharks, embryos are nourished by oophagy. Contrary to earlier suggestions, there is no evidence that lamnid embryos indulge in uterine cannibalism (adelphophagy), an extreme extension of oophagy that has been confirmed only in the sandtiger shark (Gilmore, 1993). All lamnids, and most ooph- agous sharks, produce litters larger than two (one per uterus) (Francis, 1996), providing strong circum- stantial evidence that adelphophagy does not occur in those species (Gilmore, 1993). One porbeagle embryo had nonlethal abdominal lacerations, probably result- ing from incidental damage inflicted by its larger sib- ling while searching for egg capsules, which are about the same size as the smaller embryo's abdomen. This searching behavior could provide a mechanism for the development of adelphophagy from oophagy. Unresolved questions Two puzzling features of the reproduction of porbea- gles demand further investigation. The first concerns Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 61 the gross abdominal distension during midgestation, a phenomenon that must create problems for the mother in accommodating them. It would seem energetically and hydrodynamically more efficient for a pregnant female to match her ovulation rate to the immediate growth and energy needs of the embryo, rather than to provide an over-supply of food during a short time period. We speculate that the answer lies in the avail- ability of food resources. Porbeagles feed mainly on small to medium pelagic fishes and cephalopods, but also eat larger demersal teleosts and elasmobranchs (Bigelow and Schroeder, 1948; Graham, 1956; Aasen, 1961; Templeman, 1963; Stevens et al., 1983; Gauld, 1989; Yatsu, 1995). Oophagy may be an adaptation that allows pregnant porbeagles ( and other oophagous species ) to maximize their use of food resources that are abundant only during a short season. The lack of a six-month phase shift between the reproductive cycles of Northern and Southern Hemi- sphere porbeagles is surprising, and suggests that water temperature and day length have little influ- ence on reproduction. This may be due to porbeagles having a highly developed endothermic ability (Carey et al., 1985) which buffers body temperature against seasonal fluctuations in temperature. But why should the timing of reproduction be so similar in the two hemispheres? The shortfin mako is also endothermic, although not to such a high degree as porbeagles (Carey et al., 1985), and its reproductive cycles are six months out of phase in the two hemispheres (Mol- let et al.i^). In the northwest Atlantic, porbeagle par- turition coincides with the aiTival of migratory stocks of Atlantic mackerel, capelin and 0+ Atlantic herring (Moss^). Linking parturition with the period of peak abundance of the common prey species in each hemi- sphere would provide new-bom young with their best chance of rapid growth and survival. Unfortunately, neither of these hypotheses can be tested, because there is no information on the existence or timing of abundance cycles of porbeagle prey in the Southern Hemisphere. This study has clarified several important aspects of the reproductive biology of porbeagles, including the length of the gestation period, mean fecundity, length at birth, and the timing of parturition. Growth rates have been estimated for embryos and juveniles and are consistent with other studies. However con- siderable imprecision and uncertainty remain in all of these estimates, especially the lengths of the ges- tation period and reproductive cycle, and therefore the annual fecundity. Such information is crucial to the determination of stock productivity in porbea- gles; therefore better estimates are required before effective stock assessment and management can be achieved. Acknowledgments We are indebted to the scientific observers who fre- quently operated under arduous conditions to collect data and specimens. Their enthusiasm and dedication made this study possible. Stephanie Kalish and Lynda Griggs provided database support, and arranged the delivery of specimens. The New Zealand Ministry of Fisheries allowed us access to the scientific observer database. Data, photographs and embryos were kindly provided by Guy Duhamel, Sandy Moss, Andrew New- ton, Stuart Hanchet, Rachel Berquist, Dick Williams, Jeff Williams, and Clinton Duffy. Sabine Wintner, Henry MoUet, and Markus Leppa translated the im- portant papers by Lohberger and Swenander. Henry also stimulated us with much discussion of lamnoid embryonic development. Three anonymous referees made helpful comments on the manuscript. Literature cited Aasen, O. 1961. Some observations on the biology of the porbeagle shark (Lamna nasus L.). ICES C. M. 1961. Near Northern Sea Comm. 109. 7 p. 1962. En interessant gjenfangst av merket habrann. Fis- kets Gang 14:221. 1963. Length and growth of the porbeagle {Lamna nasus. Bonnaterre) in the North West Atlantic. Fisk. Skrift. Ser. Havund. 13(6):20-37. 1966. Brugde,Cetor/!i>H/smaj:;m!/s(Gunnerusl, 1765. Saer- trykk av Fishets Gang 49:909-920. Anderson, E. D. 1990. Fisheries models as applied to elasmobranch fisheries. NOAATech. Rep. NMFS 90:473-484. Bass, A. J., J. D. D' Aubrey, and N. Kistnasamy. 1975. Sharks of the east coast of southern Africa. IV. The fami- lies Odontaspididae, Scapanorhynchidae, Isuridae, Cetorhini- dae, Alopiidae, Orectolobidae and Rhiniodontidae. Oceanogr. Res. Inst. Invest. Rep. 39, 102 p. Bigelow, H. B., and W. C. Schroeder. 1948. Fishes of the Western North Atlantic. Part 1. Lancelets, cyclostomes, sharks. Mem. Sears Found. Mar. Res. 1:59-546. Cadenat, J. 1956. Notes d'ichtyologie Quest Africaine. XTV. Remarques biologiques sur le requin-sable Carcharias fOdontaspis) taurus Rafinesque 1810. Bull. I'lnst. Fran. d'Afr. Noire 18:1249-1256. Carey, F. G., J. G. Casey, H. L. Pratt, D. Urquhart, and J. E. McCosker. 1985. Temperature, heat production and heat exchange in lamnid sharks. Mem. South. Calif Acad. Sci. 9:92-108. Chen, C.-T., K.-M. Liu, and Y.-C. Chang. 1997. Reproductive biology of the bigeye thresher shark, Alo- pias supercilwsus (Lowe, 1839) (Chondrichthyes: Alopiidae), in the northwestern Pacific. Ichth. Res. 44:227-235. Cliff, G., S. F. J. Dudley, and B. Davis. 1990. Sharks caught in the protective gill nets off Natal. South Africa. 3. The shortfin mako shark Isiirus oxyrinchus (Rafinesque). S. Air. J. Mar. Sci. 9:115-126. 62 Fishery Bulletin 98(1) Cliff, G., S. F. J. Dudley, and M. R. Jury. 1996. Catches of white sharks in KwaZulu-Natal, South Africa and environmental influences. In A. P. Klimley and D. G. Ainley ( eds ). Great white sharks: the biology ofCarcharodon carcharias. p. 351-362. Academic Press, San Diego, CA. Compagno, L. J. V. 1984. Sharks of the world. An annotated and illustrated cata- logue of shark species known to date. FAO Fish. Synopsis 125, vol. 4. part 1, 249 p. Duhamel, G., and C. Ozouf-Costaz. 1982. Presence de Lamna nasus (Bonnaterre, 1788) aux lies Kerguelen. Cybium 6:15-18. EUis, J. R., and S. E. Shackley. 1995. Notes on porbeagle sharks, Lamna nasus. from the Bristol Channel. J. Fish Biol. 46:368-370. Francis, M. P. 1996. Observations on a pregnant white shark with a review of reproductive biology. In A. P. Klimley and D. G. Ainley (edst. Great white sharks: the biology o{ Carcharodon car- charias, p. 157-172. Academic Press, San Diego, CA. Francis, M. P., L. H. Griggs, S. J. Baird, T. E. Murray, and H. A. Dean. 1999. Fish bycatch in New Zealand tuna longline fisheries. NIWATech. Rep. 55, 70 p. Gauld, J. A. 1989. Records of porbeagles landed in Scotland, with obser- vations on the biology, distribution and exploitation of the species. Scot. Fish. Res. Rep. 45, 15 p. Gilmore, R. G. 1983. Observations on the embryos of the longfm mako, Isurus paucus, and the bigeye thresher, Alopias supercilio- sus. Copeia 1983:375-382. 1993. Reproductive biology of lamnoid sharks. Env. Biol. Fish. 38:95-114. Gilmore, R. G., J. W. Dodrill, and P. A. Linley. 1983. Reproduction and embryonic development of the sand tiger shark, Odontaspis taurus (Rafinesque). Fish. Bull. 81:201-225. Goldman, K. J. 1997. Regulation of body temperature in the white shark, Carcharodon carcharias. J. Comp. Phys. B 167:423-429. Graham, D. H. 1939. Breeding habits of the fishes of Otago Harbour and adjacent seas. Trans. Proc. Roy. Soc. N. Z. 69:.361-372. 1956. A treasury of New Zealand fishes, 2"'' ed. Reed, Wel- lington, 424 p. Gruber, S. H., and L. J. V. Compagno. 1981. Taxonomic status and biology of the bigeye thresher, Alopias superciliosus. Fish. Bull. 79:617-640. Hamlett, W. C. 1983. Matemal-fetal relations in elasmobranch fishes. Ph.D. diss., Clemson University, Clemson, SC. 228 p. Hubbs, C. L. 1923. Notes on a small collection of fishes from Monhegan Island, Maine. Copeia 1923:101-103. Klimley, A. P. 1985. The areal distribution and autecology of the white shark, Carcharodon carcharias. off the west coast of North America. Mem. South. Calif Acad. Sci. 9:15^0. Kohler, N. E., J. G. Casey, and P. A. Turner. 1995. Length-weight relationships for 13 species of sharks from the western North Atlantic. Fish. Bull. 93:412-418. Last, P. R., and J. D. Stevens. 1994. Sharks and rays of Au.stralia. CSIRO, Hobart, 513 p. Lohberger, J. 1910. Uber zwei riesige embryontn von Lamna. Sond. Abhand. Math. -Phys. Klasse Bayer. Akad. Wissen. FV suppl. bd. 2 abhand., 45 p. MacDonald, P. D. M. 1987. Analysis of length-frequency distributions. In R. C. Summerfelt and G. E. Hall (eds.). The age and growth offish, p. 371-384. Iowa State Univ. Press, Ames, LA. MacDonald, P. D. M., and P. E. J. Green. 1988. User's guide to program MIX: an interactive program for fitting mi.\tures of distributions. Release 2.3. Ichthus data systems, Hamilton, Ontario, Canada. MacDonald, P. D. M., and T. J. Pitcher. 1979. Age-groups from size-frequency data: a versatile and efficient method of analyzing distribution mixtures. J. Fish. Res. Board Can. 36:987-1001. McKenzie, R. A. 1959. Marine and freshwater fishes of the Miramichi River and Estuary. New Brunswick. J. Fish. Res. Bd Can. 16:807-833. Mejuto, J. 1985. Associated catches of sharks, Prionace glauca. Isurus o.xyrinchus, and Lamna nasus. with NW and N Spanish swordfish fishery, in 1984. ICES C. M. 1985 H:42, 16 p. Mejuto, J., and A. G. Garces. 1984. Shortfin mako, Isurus oxyrinchus. and porbeagle, Lamna nasus, associated with longline swordfish fishery in NW and N Spain. ICES C. M. 1984 G:72, 9 p. Moreno, J. A., and J. Moron. 1992. Reproductive biology of the bigeye thresher shark, Alo- pias superciliosus (Lowe, 1839). Aust. J Mar. Freshwater Res. 43:77-86. Murray, T. 1994. Areview of the biology and fisheries for albacore, Thun- nus alalunga. in the South Pacific Ocean. FAO Fish. Tech. Pap. 336/2:188-206. Nakaya, K. 1971. Descriptive notes on a porbeagle, Lamna nasus. fiT)m Argentine waters, compared with the North Pacific salmon shark, Lamna ditropis. Bull. Fac. Fish. Hokk. Univ 21:269-279. Nordgard, O. 1926. TrondhjemsBiologiskeStasjon 1900-1925. DetKong. Norske Viden. Selsk. Skrift. 1925:37-39. O'Boyle, R. N., G. M. Fowler, P. C. F. Hurley, M. A. Showell, W. T. Stobo, and C. Jones. 1996. Observations on porbeagle shark {Lamna nasus) in the North Atlantic. DFO Atl. Fish. Res. Doc. 96/24, 29 p. Otake, T., and K. Mizue. 1981. Direct evidence for oophagy in thresher shark, Alopias pelagicus. Jpn. J. Ichthyol 28:171-172. Pratt, H. L. 1988. Elasmobranch gonad structure: a description and survey. Copeia 1988:719-729. 1993. The storage of spermatozoa in the oviducal glands of western North Atlantic sharks. Env Biol. Fish. 38:139-149. Pratt, H. L., and J. G. Casey. 1990. Shark reproductive strategies as a limiting factor in directed fisheries, with a review of Holden's method of esti- mating growth parameters. NOAA Tech. Rep. NMFS 90: 97-109. Shann, E. W. 1911. Adescription of the advanced embryonic stage of Lamna cornubica. Ann. Rep. Fish. Bd Scotland 28(3):73-79. 1923. The embryonic development of the porbeagle shark, Lamna cornubica. Proc. Zool. Soc. Lond. 11:161-171. Springer, S. 1948. Oviphagous embryos of the sand shark, Carcharias taurus. Copeia 1948:153-157. Francis and Stevens: Reproduction, embryonic development, and growth of Lamna nasus 63 Stevens, J. D. 1976. First results of shark tagging in the North-east Atlan- tic. 1972-1975. J. Mar. Biol. Assoc. U.K. 56:929-937. 1983. Observations on reproduction in the shortfin mako hunts oxynnchus. Copeia 1983:126-130. 1990. Further results from a tagging study of pelagic sharks in the North-east Atlantic. J. Mar. Biol. Assoc. U.K 70:707-720. Stevens, J. D., M. C. Dunning, and S. Machida. 1983. Occurrences of the porbeagle shark, Lamna nasus, in the Tasman Sea. Jpn. J. Ichthyol. 30:301-307. Svetlov, M. F. 1978. The porbeagle. Lamna nasus. in Antarctic waters. J. Ichthyol. 18:850-851. Swenander, G. 1906. Bidrag till kannedomen om TrondhjemsQordens fiskar. Det Kong. Norske Viden. Selsk. Skrift. 1905(9):1-112. 1907. Uber die emahrung des embryos der Lamna corn u bica . Zool. Stud. Toll. TuUberg, Upp. 1907:283-288. Templeman, W. 1963. Distribution of sharks in the Canadian Atlantic (with special reference to Newfoundland waters). Bull. Fish. Res. Board Can. 140, 77 p. Uchida, S., M. Toda, K. Teshima, and K. Yano. 1996. Pregnant white sharks and full-term embryos from Japan. In A. P. Klimley and D. G. Ainley (eds.), Great white sharks: the biology of Carcharodon carcharias, p. 139-155. Academic Press, San Diego, CA. Yatsu, A. 1995. Zoogeography of the epipelagic fishes in the South Pacific Ocean and the Pacific sector of the Subantarctic, with special reference to the ecological role of slender tuna, AUothunnus fallal. Bull. Nat. Res. Inst. Far Seas Fish. 32, 145 p. 64 Abstract.— The retrospective assign- ment of collections of larval swordfish, Xiphias gladiiis, taken from 1973 to 1980, to water tj^pes and area of the Gulf Stream front, as well as three sets of contemporary collections taken in 1984, 1988, and 1997, indicated that larvae were collected most frequently within the western Gulf Stream frontal zone. Larval swordfish accumulate by local- ized hydrodynamic convergence, rather than localized spawning, and thus these rare, surface-oriented larvae are found more frequently within the frontal zone. Lengths of larval swordfish taken from curatorial collections, from contemporary collections, and from published records from the Caribbean Sea, the Gulf of Mexico, and the western North Atlan- tic, as well as assumptions about growth rates and Gulf Stream transport, indi- cated that swordfish may spawn as far north as Cape Hatteras. Distribution of larval swordfish, Xiphias gladius, and probable spawning ofjp the southeastern United States John Jeffrey Govoni National Marine Fisheries Service, NOAA Southeast Fisheries Science Center Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516-9722 E-mail address Jeff Govonifflnoaa gov Bruce W. Stender Oleg Pashuk Marine Resources Research Institute South Carolina Wildlife and Manne Resources Department Charleston, South Carolina 29412 Manuscript accepted 22 June 1999. Fish. Bull. 98:64-74(2000). Swordfish, Xfp/z/os^/arf/f/s, is a cos- mopolitan and highly migratory spe- cies that spawns year-round (Grail et al., 1983). Larvae of this fish do not account for high numbers of the ichthyoplankton or ichthyoneuston, and as a consequence, data that de- scribe their spatial distribution are sparse. Grail et al. ( 1983) concluded from available data that, in the west- ern North Atlantic, small larvae (< 10 mm) occur most fi-equently in the eastern Caribbean and in the Straits of Yucatan and Florida in Novem- ber, and that larger larvae (>10 mm) occur most often there, as well as in the Gulf Stream north to Cape Hatteras from January to March. Both length groups occur primarily in surface water over depths deeper than 200 m (Markle, 1974). Tibbo and Lauzier (1969) first speculated that larvae may be associated with horizontal temperature and salinity gradients. The collection of larvae along the apparent edges of the Gulf Loop Current in the Gulf of Mexico (Richards and Potthoff, 1980) and the Gulf Stream in the Atlantic (Pot- thoff and Kelley, 1982; Post et al., 1997 ) supports the notion that larvae occur in greater abundance within fi-ontal zones. Yet, the rarity of larvae has hindered a clear understanding of such coarse and fine scale (Haury et al., 1978) spatial distribution. The scarcity of larval swordfish has obscured an understanding of their spawning pattern as well. The mesoscale pattern of larval distribu- tion (Grail et al, 1983) implies that swordfish spawn in the Caribbean and the Straits of Yucatan and Flor- ida, and that their larvae are carried northward by the Gulf Stream. Occa- sional small larvae taken in the Atlan- tic imply that swordfish may spawn as far north as Cape Hatteras (Markle, 1974). Large, and presumably older larvae in any location may be the re- sult either of local spawning and sub- sequent retention or of transport ft"om a distant spawning locale. Small lar- vae at a specific locale must be the ex- clusive result of local spawning. Hydrated oocytes within the ova- ries of adult females indicate that swordfish spawn south of the Sar- gasso Sea, in the northern Carib- bean Sea, and the Straits of Florida (Arocha and Lee, 1995), although Squires ( 1962 ) has suggested that Govoni et al.: Distribution of larval Xiphias gladius off the southeastern United States 65 spawning may occur as far north as Cape Hatteras. Planktonic eggs have not been identified in the west- em North Atlantic. Effective fisheries management requires knowledge of both the spatial distribution of larvae and the dis- tribution of spawning adults. Contemporary stock sta- tus ascertained by virtual population analysis can be calibrated with larval abundance estimates, even for species with rare larvae (e.g. McGowan and Richards, 1989; Scott et al., 1993). Because this calibration de- pends on accurate estimates of larval abundance, the spatial and temporal distribution of larvae must be known. Knowledge of spawning distribution would be the first step toward protection of spawning habitat and, perhaps, the restriction of fishing within spawn- ing seasons and locales. Here we examine the coarse- and fine-scale distri- bution, and the lengths of swordfish larvae off the southeastern United States. Our focus is on the influ- ence of the Gulf Stream in shaping the spatial distri- bution of larvae and in the determination of probable spawning locales. Methods We examined published records of larval swordfish (Arata, 1954; Arnold, 1955; Tibbo and Lauzier, 1969; Markle, 1974; Post et al, 1997), data from the Ma- rine Resources Monitoring, Assessment and Predic- tion Program (MARMAP), data and specimens from the Southeast Area Monitoring and Assessment Pro- gram (SEAMAP), and new data from three surveys conducted between the Florida Straits and Cape Hat- teras in 1984, 1988, and 1997 (CF8406, CH8807, and CH9703J. Spatial distribution of species For spatial distribution, we examined exclusively neuston collections (the upper 0.5 m of water), be- cause swordfish are surface-oriented as are larvae. Al- though some swordfish larvae have been collected in nets that fished principally below the surface (Grail et al., 1983), most have been collected at the surface (Taning, 1955; Yabe et al., 1959; Gorbunova, 1969; Ni- shikawa and Ueyanagi, 1974). All swordfish larvae collected by MARMAP, SEAMAP, and CH8807 were collected in the neuston, none in accompanying, sub- surface ichthyoplankton collections. Small larvae were occasionally taken from below the surface in CH9703, but these nets fished obliquely fi-om 20 m and larvae were likely captured when nets were near the surface. Collections of larvae were classified to water mass (or type) — shelf water (including Georgia water, Car- olina Capes water, and occasionally Virginia coastal water [Pietrafesa, 1989]) or Gulf stream water — by applying measured hydrographic characteristics to the classifications of Matthews and Pashuk ( 1986) for MARMAP and CF8406 collections, and Pietrafesa et al. (1985) for CH8807 and CH9703 collections. Fron- tal zone water is a mixture of these water masses (Hitchcock et al., 1994). South of Cape Hatteras the Gulf Stream courses north-northeastward in juxta- position with shelf water. Classically, the definition of the Gulf Stream front is a dynamic one: the Gulf Stream front is the point where the pressure gradient between Sargasso Sea water and slope water (north of Cape Hatteras) or shelf water (south of this Cape) is zero ( Stommel, 1966). Practically, observed horizon- tally compressed surface isotherms and isohalines, ac- companied by sharp discontinuities in sea-surface tex- ture and color, define the Gulf Stream frontal zone (Olson et al., 1994). In our study, we used this obser- vational definition to assign the surface position and to classify the water of the Gulf Stream frontal zone. Surface Gulf Stream water at its western fi-ont be- tween the Florida Straits and Cape Hatteras has char- acteristic temperatures that range from 2^ to 24°C in wdnter and from 27° to 29°C in summer, salinities that range from 35.7 to 36.4 psu and vary little sea- sonally, dissolved oxygen values that range from 4.5 to 5.0 mL/L, and nitrate values of 1.0 pM/L (Atkinson, 1985; Schmitz et al., 1993; Hitchcock et al., 1994; Xie and Pietrafesa, 1995). Shelf water is cooler and less sahne (Pietrafesa et al., 1985). The course of the Gulf Stream along the continental shelf break is unstable; it meanders onshore and offshore and projects intru- sions, filaments, and eddies onto the shelf (Pietrafesa, 1989; Lee et al., 1991 ). These processes complicate the po- sition and distort the configuration of the frontal zone. Retrospective examination of 1163 collections taken from the Straits of Florida to Cape Hatteras and off- shore from the 9 m to 2000 m isobath (Fig.l) from 1973 to 1980 (MARMAP) afforded the determination of coarse-scale distribution. These collections were taken with neuston nets of two different dimensions and meshes: a 1.0 x 0.5 m net with 505-pm mesh and a 2.0 X 1.0 m net with 947-pm mesh. Both meshes col- lect swordfish larvae, which are reported to be at least 4 mm at hatching (Sanzo, 1910; Yasuda et al., 1978). Neuston nets were towed for 10 min at 5.6 km/h. Col- lections were made in all four seasons and at all times of day and night. Larvae were preserved in 5% forma- lin solution. The probable location of the Gulf Stream front for these collections was determined from 1) advanced, very high resolution, infra-red radiometer (AVHRR) satellite images of sea-surface temperature (SST) taken fi-om 1976 to 1980; 2) expendable bathy- thermograph (XBT) profiles taken from 1973 to 1980; 66 Fishery Bulletin 98(1 3) sea-surface salinity, dissolved oxygen, and nitrate concentrations taken along with neuston collections (Mathews and Pashuk, 1986); 4 1 records of the Na- tional Oceanographic Data Center; and 5) records of U.S. Coast Guard visual observations from aircraft. The width of the frontal zone, and consequently the area occupied by frontal zone water, was determined as three times the standard deviation of the mean course of the front, i.e. ±10 nautical miles ( 18.5 km) of the estimated location of the frontal axis (Olson et al., 1983). Forty-seven collections taken from 22 to 24 June 1984 with a 1.0 X 0.5 m neuston net wdth 909-]am mesh \ \. Figure 1 Nominal station.s occupied by the Marine Resources Monitoring Assessment and Prediction Program from 1973 to 1980; each station was occupied at least once, some more than once. towed for 10 min at 5.6 km/h in the Gulf Stream and within its western frontal zone between Cape Ca- naveral and Cape Hatteras (CF8406) afforded an ex- amination of coarse scale differences in occurrence of swordfish larvae within the frontal zone and in the body of the Gulf Stream. Six collections were taken during night, the remainder during the day. Larvae were preserved in 5% formalin. A sea-surface temper- ature plot, generated from composite AVHRR images of SST compiled on 20 June 1984 (Fig. 2), in conjunc- tion vrith continuous temperature and salinity values from a hull-mounted thermosalinometer, was used to determine the position of the Gulf Stream and its frontal zone and to classify these collections to water mass. One hundred and fifty-six collections ( CH8807 ), taken from 13 to 24 September 1988 with a 2.0 x 1.0 m neuston net with 947-pm mesh along six cross-shelf and cross-slope tran- sects (encompassed by 31°30.4N/080°16.5W, 30°45.0N/078°411.5W, 32 24°30.0N/076°19.5W, and 33"16.0N/077'54.0W), afforded examina- tion of coarse- and fine-scale distribution of larval swordfish within and about the Gulf Stream frontal zone (Fig. 3). Neuston nets were towed for 10 min at 5.6 km/h. Ten sta- tions were occupied along each transect (60 stations); 96 additional neuston collections were taken within the Gulf Stream or within its frontal zone. The 60 collections along tran- sects were 18.5 km apart, whereas 94 of the additional 96 collections within the frontal zone or body of the Gulf Stream were clus- tered in groups of four to ten, with collections about 1 km apart. Surface salinity, two-dimen- sional sections of isotherms derived from XBT casts taken at each station along transects, and AVHRR images of SST were used to clas- sify collections. Six neuston collections taken on 31 May 1997 (CH9703) with a 2.0 x 1.0 m neuston net with 947-pm mesh within the Gulf Stream frontal zone in an area encompassed by 33°52.94'N/076°23.89'W,33=52.30'N/076°24.77'W, and 33°52.92N/076°22.27W afforded exami- nation of fine-scale distribution of lan'al sword- fish within the Gulf Stream frontal zone (Fig. 4). Nets were towed for 10 min at 5.6 km/h, twice in the Gulf Stream frontal zone, and twice each on the shelf and Gulf Stream sides of the front. Larvae were preserved in 959^ ethanol. Sea-surface temperature from a hull- mounted thermister in conjunction with XBT profiles and AVHRR images of SST were used to classify collections. I Govoni et a\ Distribution of larval Xiphias gladius off thie soutfieastern United States 67 Figure 2 The position of the Gulf Stream (diagonal line pattern) illustrated from a 3-d average of sea-surface temperatures compiled on 20 June 1984 with data from infrared sensors on a Global Stationary Environmen- tal Satellite operated by the National Weather Service (closed circles denote approximate locations of multiple neuston collections where swordfish larvae were present; open circles, locations where swordfish larvae were absent; fractions are the number of positive collections over the total number of collections). Length of larvae The overall distribution of the length of swordfish lar- vae in the Caribbean Sea, Gulf of Mexico, and off the southeastern Atlantic coast of the United States pro- vided a view of the occurrence of the smallest larvae. Because most published have records reported either standard length (SL) or total length (TL), both mea- sures are reported in our study. We employed no con- version, although the length of the caudal finfold or fin ranges from 3% to 12% of TL for larvae from 6 to 192 mm (calculated from Arata [1954]). We employed no correction for the shrinkage of larvae due to death or preservation, because preservation fluid was not routinely reported in the literature and some speci- mens were measured alive. Although the length from the anterior eye orbit to the tip of the notochord is per- haps a better measure of length in swordfish larvae, because the rostrum is often broken (Potthoff and Kel- ley, 1982), this measure was not uniformly available in published records. Standard or total lengths (ante- rior tip of the upper jaw to the posterior tip of the noto- chord or hypural plate (SL) or caudal fin (TL)) of sword- fish larvae and locations of collection were taken ft-om Arata ( 1954), Arnold ( 1955), Tibbo and Lauzier (1969), Markle (1974), Post et al. (1997) and MARMAP. Stan- dard length was measured on larvae from SEAMAP, CF8408, CH8807, and CH9703 (ft-om neuston as well as plankton collections taken within the Gulf Stream fi-ontal zone with a MOCNESS (Wiebe, et al., 1976) system with 505 m mesh nets), and on a larva fi"om a neuston collection taken in a Sargassum raft at 33°38.7'N/076°02.7'W in 1991. Specimens were consid- ered larvae if they were <190 mm SL or TL, because 68 Fishery Bulletin 98(1) ^■^J! pr"- A W Cape ^ W to \ i8)-^l 28 i8i L \ 13 I % Cape .) 60 ,'»i ■.51 ■t:"^.s50 • 31 \ (4, vlbj. \ "t Figure 3 Advanced very high icsolulion infrared radiometer images of sea-surface temperatures ofl' the south- eastern Atlantic coast of the United States; five day composites from National Oceanic and Atmospheric Administration, National Environmental Satellite, Data and Information Ser\ice glohal orbiting satel- lites: (Ai 11 to 15 September; (Bl 16 to 20 September; (C) 21 to 25 Septemberl9H8. Darker to lighter shades denote coolei' to warmer water; squares denote loca- tions of neuston collections; numbers denote .station numbers; and parenthetical numbers within insets denote the number of overlapping stations. swordfish retain lai-val characters (lower jaw at least half as long as the rostral cartilage; preorbital, supraor- bital, posttemporal, and preopercular spines; enlarged and spinous dorsal and ventral scales; and a continu- ous and long dorsal fin ) until they are at least 188 mm SL (Ai-ataTl954; Pothoff and Kelley, 1982; McGowan, 1988). Approximate age of larvae The age of swordfish larvae at length was approximated from published accounts of the age and lengths of labora- tory-reared larvae, and fi'om counts of apparent growth increments on otoliths excised fi:'om three specimens col- lected by CH9703. Larvae fi-om the Mediterranean hatch at a length fi'om 4.0 to 4.5 mm TL (measured aUve) after 3 d of incubation and deplete their yolk and oil globule at a length of about 5 mm TL after 5 to 7 d at 22.5-25.2°C (Sanzo, 1910; Yasuda et al., 1978). Results Spatial distribution The 1163 MARMAP collections yielded 55 swordfish larvae in 35 collections. Lai'vae were collected in all seasons and at all times of the day. Of the 35 collec- tions that produced lai'vae, 21 were taken in the day, 5 at night, and 9 at dawn or dusk. As many as nine lar- vae were taken in a single collection. Between Cape Ca- naveral and Cape Hatteras, larvae were collected more frequently within the fi'ontal zone of the Gulf Stream, than they were in shelf or Gulf Stream waters. Where the Gulf Stream jets through the Florida Straits (25° to 28°N latitude), larvae occuiTed across the narrow body of the Gulf Stream. If these Florida Straits collec- tions are excluded from consideration, along with col- lections from single seasonal or annual surveys that did not produce any larvae, 21 of these 27 collections that yielded larvae were within the probable area of the Gulf Stream frontal zone, i.e. 18.5 km of the assigned position of the Gulf Stream surface frontal axis; one was in shelf water, and five were in Gulf Stream water (Fig. 5; Table 1 1. The probability of observing the presence or absence of larvae (calculated as fi-equency of occuirence), among the two water masses and the mixed fi-ontal zone, with at least as much association was 0.000002'X (Fisher's exact test, two-tailed distribution ). In June 1984 (CF8406), the frontal zone was de- fined by the 25^ to 27 C sin-face isotherms. North of Florida, the western Gulf Stream front was smooth with no evidence of instabilities in the form of large intrusions (Fig. 2). Twelve of 47 collections of CF8406 yielded 16 swordfish lai'vae. Of these, two collections Govoni et al : Distribution of larval Xiphias g/adius off the soutfieastern United States 69 I "^•U <* Figure 4 Advanced very high resolution infrared radiometer images of sea-surface temperatures off the southeastern Atlantic coast of the United States on 28 May 1997 from NOAA polar orbiting satellite (darker to lighter shades denote cooler to warmer water; circles denote locations of hydrographic stations occu- pied on 3 June 1997). yielded more than one larva, i.e. two larvae in each. Ten collections with larvae present were within the frontal zone, whereas two collections were in the body of the Gulf Stream (Table 1). The probability of observing the presence or absence of larvae with a frequency at least this extreme was 0.176% (two-tailed test). In September 1988 (CH8807), the Gulf Stream fron- tal zone was defined by 27° to 29°C surface isotherms. With the defining criterion of horizontally compressed surface isotherms within this temperature range, the width of this frontal zone ranged from 18 to 93 km. This area, however, encompassed a westward intru- sion of the Gulf Stream that developed over the tran- sect grid (Fig. 3) in the wake of the Charleston Bump, a topographic rise at the continental shelf break that typically forces an eastward deflection of the Gulf Stream (Pietrafesa et al., 1985). Isolated surface pools of 27° to 29°C water inshore of the Gulf Stream front proper (Fig. 6, B and F) indicated this intrusion. This convolution of the Gulf Stream front proper mani- fested three fronts across the shelf but these were considered a composite single front for analysis. Shear zones (determined by rapid increases in the ship's set), drift lines of Sargassum, and discontinuities in sea-surface texture, were embedded within the fron- tal zone and evidenced probable convergence of sur- face water (Stommel, 1966; Olson et al., 1994). The 156 collections of CH8807 yielded 12 swordfish larvae. One collection yielded three larvae; another two. Lar- vae were collected exclusively within the frontal zone (Table 1), one at the tip of the intrusion (station 58; Fig. 6F). The probability of observing the presence or absence of larvae at least this extreme was 0.020%. In May 1997 (CH9703), the Gulf Stream frontal zone was defined by 27° to 29°C surface isotherms. The frontal zone was moving rapidly eastward. On 23 May, 8 d before the collection of swordfish larvae within the frontal zone, an elongate filament of a Gulf Stream intrusion lay inshore of the collection area (Fig. 41. The isolated pool of 25°C water at the sur- face and eastward of the frontal zone (Fig. 7) indicated that this intrusion was extant on 31 May 1997 when collections of CH9703 were taken. The six neuston col- lections of CH9703 yielded nine larvae. One collection taken along the frontal axis and one collection taken on the Gulf Stream side of the axis yielded two larvae each; the other collection taken along the axis yielded three. Larvae were present in two collections along the frontal axis, two collections 1 km to the Gulf Stream side of the axis, and one of two collections taken 0.5 km 70 Fishery Bulletin 98(1) ^ u c Figure 5 Position of the Gulf Stream front and locations of neuston collections of surveys when swordfish larvae were collected from 19V3 to 1980 (solid lines denote assigned axis of the Gulf Stream front, small circles locations of neuston collections; large circles locations of swordfish larvae: (A) March 19V3; (Bl May 19V3; (C) July V3; (D) November 19V3;(E) August and September 1974; (F) August and September 19V5; (G) January and February 19V6; I H ) August and September 19V6; (I ) Septem- ber 19V8;(JiApnl 1979. Table 1 Contingency tables of the absence and presence of larval swordfish, Xiphiafi gladius, in Gulf Stream frontal ?one, the Gulf Stream and shelf water in the southeastern Atlantic | bight offjhe United States. Frontal zone Gulf Stream Shelf Frequency water water water Total 1973-80 Absent 158 25 175 358 Present 21 5 1 27 Total 179 30 176 385 1984 Absent 21 14 35 Present 10 2 12 Total 31 16 47 1988 Absent 77 46 24 14 V Present 9 9 Total 86 46 24 156 1997 Absent 0' 0^' 1-J 1 Present 2' 2-' 1' 5 Total 2' 22 2^ 6 ' Taken along the frontal axis. - Taken along offshore side of the axis. ' Taken on inshore side of the axis. - to the shoreward side of the frontal axis. Larvae were evenly distributed within the frontal zone (Table 1). Length of larvae Swordfish lai-vae ranged from 2.8 mm SL to 110 mm TL. The two largest larvae, >100 mm TL, were col- lected in the Atlantic and in the Caribbean. The small- est and youngest larvae, those <5 mm SL (Fig. 8), were collected in the eastern Gulf of Mexico in the vicinity of the Gulf Loop Current between 24° and 28'N (here, 39 of 152 larvae were <5 mm), and off the southeast coast from Georgia north to Cape Hat- teras between 30° and 35°N latitude (here, 15 of 62 larvae were <5 mm SL). Larval lengths were not cor- related with latitude (Pearson product-moment corre- lation coefficient=-0.1360). Approximate age of larvae The smallest lai-va measured ( after preservation in 95*^ ethanol), taken in the Gulf of Mexico from SEAMAP, was appreciably smaller than the length at hatching for larvae (measured alive, 27 h after fertilization) from the Mediterranean (Yasuda et al., 1978). One larva 2.8 mm SL from the eastern Gulf of Mexico, and GovonI et al.: Distribution of larval Xiphias gtadius off the southeastern United States 71 one 3.8 mm SL (4.3 mm TL) taken off North Carolina (CH9703), had recognizable yolk and oil globule rem- nants. These lengths were smaller than the reported length of larvae at the completion of yolk and oil glob- ule absorption, about 5 mm TL for larvae (measured alive, 65 h after fertilization ) from the Mediterranean (Yasuda et al., 1978). Counts of increments on sagittae from larvae 4, 5, and 6 mm SL, taken in CH9703, were 4, 3, and 6 (increments on swordfish otoliths have not been vali- dated as daily intervals [Price et al., 1991] ). Discussion North of the Florida Straits, larval swordfish were collected most frequently within the western frontal zone of the Gulf Stream. This observation corroborates the speculation that larvae are associated with water within sharp horizontal gradients of temperature and salinity (Tibbo and Lauzier, 1969). Convergence of sur- face water is a possible mechanism for their accumu- lation within the Gulf Stream front. The Gulf Stream front south of Cape Hatteras is cyclonically sheared with shelf water that directly opposes Gulf Stream water (Pietrafesa et al., 1985). The retrograde, hydro- graphic discontinuity between Gulf Stream and shelf water and their hydrodynamic opposition results in convergence of surface water within the frontal zone (Garvine, 1980; Olson et al., 1994). Convergence of surface water has been implicated in the accumula- tion of adult fishes with depth-keeping abilities (Ol- son and Backus, 1985). Positively buoyant or surface- seeking larval fishes will be entrained in converging water and will be advected toward the front where they will accumulate as they resist downwelling along the frontal axis (Govoni and Grimes, 1992). Sword- fish larvae are unquestionably surface-seeking larvae. Convergence of surface water within oceanic frontal zones should produce accumulations of larvae on spa- tial scales of 10 to 100 km (Olson et al., 1994). At a coarse scale, larvae were more abundant within the frontal zone; no fine-scale pattern was evident within the frontal zone. Adaptive sampling (Lo et al., 1997) may be a more efficient means of estimating larval swordfish abundance than simple random sampling, because of the rarity and the spatial aggregation of larvae. Small swordfish larvae were collected most often in the eastern Gulf of Mexico and off the east coast of the United States as far north as Cape Lookout, North Carolina. Swordfish apparently spawn in the eastern Gulf, but the present observations corroborate the speculation of spawning off the Carolinas (Squires, 1962; Markle, 1974) as well. Off the Carolinas, larvae Station number 200 200L Figure 6 Sections of isotherms along transects occupied in September 1988: (A) 13 September; (B) 14 September; (C) 1.5 Septem- ber; (D) 16 September; (E) 17 September; (Fl 18 September (bars above station numbers indicate the Gulf Stream fron- tal zone). 5 mm SL or less occurred in 25° and 26°C water. Lar- vae that were 4 to 5 mm SL had 3 and 4 apparent growth increments on their sagittae. In water from 22° to 25°C, larvae that were 5 mm TL would be ap- proximately 6 d old (Yasuda et al., 1978). Given an egg incubation period of 3 d at 24°C (Yasuda et al., 1978) and an additional 3 or 4 d for posthatch growth, along with a average axial velocity of the Gulf Stream of 1.5 72 Fishery Bulletin 98(1) a. a Station number 36 37 38 40 42 44 46 200 Figure 7 Section of isotherms along a transect occupied on 3 June 1997, Bar above station numbers indicates the Gulf Stream frontal zone and the location of larval swordfish collections. m/s (Olson et al.. 1994), larvae that were 4 to 5 mm SL in the Atlantic could have been transported from as far away as 900 km. A similar trajectory was pro- jected for small larvae of bluefin tuna, Tim units tliyn- nus (McGowan and Richards, 1989). Larvae that were <5 mm in length, collected off North Carolina could have been spawned in the Florida Straits if they re- mained in the core of the Gulf Stream. Current ve- locities within the western Gulf Stream frontal zone, where larvae most frequently reside, are less than axial velocities (Lillibridge et al., 1990; Song et al., 1995; Limouzy-Paris et al., 1997). Further, departures from a smooth, along slope. Gulf Stream trajectory, in the form of meanders, intrusions, and filaments along the western Gulf Stream frontal zone are fre- quent (Pietrafesa, 1989). We collected swordfish lar- vae frequently within these Gulf Stream anomalies (Figs. 4-7). Water within these features veers and re- verses direction (Lee et al., 1991), the result being that the northward translocation of swordfish larvae within the frontal zone is checked and their north- ward transport delayed. The possibility of spawning between Cape Canaveral and Cape Hatteras is real, but not certain. Acknowledgments We thank J. A. Hare and L. R. Settle (NOAA) for the collection of some of the swordfish specimens reported herein, J. A. Hare for manipulation of satellite im- ages, E. H. Laban for the extraction of otoliths, and C. W. Lewis, R. L. Ferguson, and J. D. Christensen for preparation of figures. J. A. Hare, J. V. Merriner, and GovonI el al.: Distribution of larval Xiphias gladius off the southeastern United States 73 Figure 8 Locations of larval swordfish collection in the Caribbean Sea, Gulf of Mexico, and off the southeast coast of the United States north to Cape Hatteras (crosses depict larvae 2.8 through 5 mm standard (SL) or total length (TL); circles, larvae 5.1 through 110 mm SL or TL; within the box in the Gulf of Mexico, coincidental crosses represent 39 larvae <5 mm and 113 >5 mm SL; within the Atlantic, coincidental crosses represent 15 larvae <5 mm and coincidental circles 47 >5mm SL). D. S. Vaughan provided reviews of the manuscipt. J. M. Leiby loaned SEAMAP specimens and data. JJG thanks W. J. Richards for first suggesting where to find swordfish larvae and the late F. G. Carey who supported initial work. Literature cited Arata. G. F. 1954. A contribution to the life history of the swordfish, Xiphias gladius Linnaeus, from the south Atlantic coast of the United States and the Gulf of Mexico. Bull. Mar Sci. Gulf Caribb. 4:183-243. Arnold, E. L. 1955. Notes on the capture of young sailfish and swordfish in the Gulf of Mexico. Copeia ( 1955)2:150-151. Arocha, F., and D. W. Lee. 1995. The spawning of swordfish from the northwest Atlan- tic Int. Comm. Conserv. Atl. Tuna, Collect. Vol Sci. Pap. 44:179-183. Atkinson, L. P. 1985. Hydrography and nutrients of the southeastern U.S. con- tinental shelf. In L. P. Atkinson, D. W. Menzel, and K. A. Bush (eds. ), Oceanography of the southeastern U.S. continental shelf, p. 77-92. American Geophysical Union, Washington. Garvine, R. W. 1980. The circulation dynamics and thermodynamics of upper density fronts. J. Phys. Oceanogr. 10:2058-2081. Gorbunova, N. N. 1969. Breeding grounds and food of the larvae of the sword- fish .Yip/iiasg/arf;(/s Linneus (Pisces, Xiphiidae). Prob, Ich- thyol. 9:375-387. Govoni, J. J., and C. B. Grimes. 1992. The surface accumulation of larval fishes by hydro- dynamic convergence within the Mississippi River plume front. Cont. Shelf Res. 12:1265-1276. Grail, C, D. P. de Sylva, and E. D. Houde. 1983. Distribution, relative abundance, and seasonality of swordfish larvae. Trans. Am. Fish. Soc. 112:235-246. Haury, L. R., J. A. McGowan, and P. H. Weibe. 1978. Patterns and processes in the time-space scales of plankton distributions. In J. H. Steele (ed.). Spatial pattern in plankton communities, p. 277-327. Plenum, New York. Hitchcock, G. L., T. Rossby, J. L. Lillibridge, E. J. Lessard, E. R. Levine, D. N. Connors, K. Y. Borsheim, and M. Mork. 1994. Signatures of stirring and mixing near the Gulf Stream front. J. Mar. Res. 52:797-836. Lee, T. N., J. A. Voder, and L. P. Atkinson. 1991. Gulf Stream frontal eddy influence on productivity 74 Fishery Bulletin 98(1) of the southeast U S. continental shelf. J. Geophys. Res. 96:22191-22205. Lillibridge, J.L., G. Hitchcock, T. Rossby, E. Lessard, M. Mork, and L. Golmen. 1990. Entrainment and mixing of shelfslope waters in the near-surface Gulf Stream. J. Geophys. Res. 95:13065- 13087. Limouzy-Paris, C. B., H. C. Graber, D. L. Jones, A, W. Ropke, and W. J. Richards. 1997. Translocation of larval coral reef fishes via sub-meso- scale spin-off eddies from the Florida Current. Bull. Mar Sci. 60:966-983. Lo, N. C. H., D. Griffith, and J. R. Hunter. 1997. Using restricted adaptive cluster sampling to estimate Pacific hake larval abundance. Cahf. Coop. Oceanic Fish. Invest. Rep. 38:103-113. Markle, G. E. 1974. Distribution of larval swordfish in the Northwest Atlan- tic Ocean. U.S. Dep. Commer, NOAA Tech. Rept NMFS- SSRF 675:252-260. Mathews, T. D., and O. Pashuk. 1986. Summer and winter hydrography of the U.S. south Atlantic bight ( 1973-1979). J. Coast. Res. 2:311-336. McGowan, C. 1988. Differential development of the rostrum and mandible of the swordfish (Xiphias gladuis) during ontogeny and its possible functional significance. Can. J. Zool. 66:496-503. McGowan, M. F., and W. J. Richards. 1989. Bluefin tuna. Thunnus thynnus, larvae in the Gulf Stream off the southeastern United States: satellite and shipboard observations of their environment. Fish. Bull. 87:61,5-631. Nishikawa, Y., and S. Ueyanagi. 1974. The distribution of the lar%'ae of swordfish, Xip/irasg/arf- ius. in the Indian and Pacific oceans. U.S. Dep. Commer., NOAA Tech. Rep.NMFS-SSRF 675:261-264, Olson, D. B., O. T. Brown, and S. R. Emmerson. 1983. Gulf Stream frontal statistics from Florida Straits to Cape Hatteras derived from satellite and historical data. J. Geophys. Res. 88:4569-4577. Olson, D. B., and R. H. Backus. 1985. The concentrating of organisms at fronts: a cold-water fish and a warm-core Gulf stream ring. .J. Mar Res. 43:113-137. Olson, D. B., G. L. Hitchcock, A. J. Mariano, C. J. Ashjian, G. Peng, R. W. Nero, and G. P. Podesta. 1994. Life on the edge: marine life and fronts. Oceanogra- phy 7:52-60. Pietrafesa, L. P. 1989. The Gulf Stream and wind events on the Carolina Capes shelf National Undersea Research Program. U.S. Dep. Commer, NOAA-NURP Rep. 89-2:89-100. Pietrafesa, L. P., G. S. Janowitz, and P. A. Wittman. 1985. Physical oceanographic processes in the Carolina Capes. In L. P. Atkinson, D.W. Menzel, and K. A. Bush (eds. ), Oceanography of the southeastern U.S. continental shelf, p. 23-32. American Geophysical Union, Washington. Potthoff, T., and S. Kelley. 1982. Development and structure of the vertebral column, fins and fin supports, branchio.stegal rays and squamation in the sv/OTA^xshXiphias gladms. Fish. Bull. 80:161- 186. Post, J. T., J. E. Serafy, J. S. Ault, T. R. Capo, and D. P. de Sylva. 1997. Field and laboratory observations on larval Atlantic sailfish (Istiophorus platypterus) and swordfish (Xiphias gla- dius). Bull. Mar. Sci. 60:1026-1034. Prince, E. D., D. W. Lee, J. R. Zweifel, and E. B. Brothers. 1991. Estimating age and growth of young Atlantic blue marlin Makaira nigricans otolith microstructure. Fish. Bull. 89:441-459. Richards, W. J., and T. Potthoff. 1980. Larval distributions of scombrids (other than bluefin tuna) and swordfish in the Gulf of Mexico in the spring of 1977 and 1978. Int. Comm. Conserv. Atl. Tuna. Coll. Vol. Sci. Pap. 9: 680-694.. Sanzo, L. 1910. Uovo e larva di pesce-spada (X(p/!(as^/arf;us L.). Riv. Mens. Pesca Idrobiol. 12:206- 209. Schmitz, W. J., J. R. Luyten, and R. W. Schmitt. 1993. On the Florida Current T/S envelope. Bull. Mar. Sci. 53:1048-1065. Scott, G. P., S. L. Turner, C. B. Grimes, W. J. Richards, and E. B. Brothers. 1993. Indices oflarval blue-fin tuna, Thunnus thynnus, abun- dance in the Gulf of Mexico; modelling variability in growth, mortality, and gear selectivity. Bull. Mar. Sci. 53:912-929. Song, T., T. Rossby, and E. Carter. 1995. Lagrangian studies of fluid exchange between the Gulf Stream and surrounding waters. J. Phys. Oceanogr. 25:46-63. Squires, J. L. 1962. Marlin and swordfish in oceanic waters of the western North Atlantic. Copeia (1962) 1:216-219. Stommel, H. 1966. The Gulf Stream. Univ. California Press, Berkeley, CA, 248 p. Taning, A. 1955. On the breeding areas of the swordfish iXiphtas). Deep-Sea Res. .3i.suppl.):4.38-450. Tibbo, S. N., and L. M. Lauzier. 1969. Larval swordfish fXiphms gladius) from three locali- ities in the western Atlantic. J, Fish, Res. Board Can. 26:3248-3251. Wiebe, P.H., K.H. Burt, S.H. Boyd, and A.W. Morton. 1976. A multiple opening/closing net and environmental sensing system for sampling zooplankton. J. Mar. Sci. .34:313-326. Yabe, H., S. Ueyanagi, S. Kikawa, and H. Watanabe. 1959. Study on the life-history of the sword-fish, A'lp/i/asg/a- dius Linnaeus. Rep. Nankai Reg. Fish. Res. Lab. 10:107- 150. Yasuda, F., H. Kohno, A. Yatsu, H. Ida, P. Arena, F. L. Greci, and Y. Taki. 1978. Embryonic and early larval stages of the swordfish, Xiphias gladius. from the Mediterranean. .J. Tokyo Univ. Fish. 65:91-97. Xie, L., and L. J. Pietrafesa. 1995. .Shoreward intrusion of upper-layer warm water by pre- scribed shelf-break temperature perturbation and surface wind stress. Geophys. Res. Letters 22:2585-2588. 75 Abstract.— For fish populations with an annual breeding cycle, a biological ref- erence point based on the Leslie matrix is presented and compared with percent maximum spawning potential CJMSP) and F^^ reference points. For determin- istic population projections, the reference point is defined as the level of fishing mortality 1 the population will increase; If A = 1 the population will remain steady; If A < 1 the population will decrease. (Pielou, 1974; Caswell, 1989; Getz and Haight, 1989). Of particular importance to this reference point is the dominant eigenvalue which, in the deterministic case, is sufficient to determine the long-term trend in popu- lation abundance (Keyfitz, 1977; Cohen et al., 1983). Given these properties, the following assertion for the deterministic case can be made: (see Rothschild and Fogarty (1989) for cautions on this practice, however). Because data on spawning stock biomass are more commonly presented in fish- ery assessments than data on fecundity, I will pres- ent the model using age-specific fecundity equivalents (i.e. spawning biomass of individual fish) rather than fecundity. In this representation, the survival rate of age-0 fish is expressed as recruits per unit of spawn- ing stock biomass ( R/SSB ) rather than as actual sur- vival rate from egg to age 1, and SSB is used in place of egg production. It is easily shown that the use of R/SSB and SSB in the Leslie matrix is algebraically equivalent to using fecundity and survival from egg to age 1. Given the mapping of the vital rates [E{i), Sii)] of the unexploited population into a Leslie matrix, it is straightforward to represent the dynamics of the pop- ulation under exploitation. Observe that for the un- exploited population S(i)=e~'^"', where Mii) is the in- stantaneous natural mortality rate. For an exploited 1 A population under exploitation can maintain it- self at or above a given level of abundance only if the dominant eigenvalue of L^ (i.e. A^) > 1. From this assertion arises the proposed reference point: F^f (for F steady, after Quinn and Szarzi, 1993) is a fishing mortality pattern where A^ = 1. Note that F^i is actually a vector comprising two components: an overall level of fishing mortality (often termed fully re- cruited F) and the relative fishing mortality between age classes (often referred to as the partial recruit- ment vector or selection pattern), and that F^ = (fully recruited F) x (selection on age ;). By convention, I will use the fully recruited F as an index of the over- all level of fishing mortality, but stress that specifi- cation of the partial recruitment function is also nec- essary to determine the impact of harvesting on a population. Also note that there is an infinite set of fishing mortality patterns for which the condition that A =1 is satisfied. For a given partial recruitment fiinc- 78 Fishery Bulletin 98(1) tion, however, only one level of fishing will satisfy this condition. Note that the converse is not true; for a given level of fully recruited fishing mortality, nu- merous partial recruitment functions can satisfy the above condition. Because of the nature of these rela- tionships, I will focus on those situations where the partial recruitment function is specified and solve for the level of fishing that is sustainable. Once the selec- tion pattern and fully recruited fishing mortality are set, Ap can be found by the power method as described by Johnson and Riess ( 1981 ). Example of maintenance fishing mortality: deterministic case Data from Georges Bank haddock (Melanogrammus aeglefinus) are used to illustrate the computation and application of this reference point. For ease of discus- sion, I first present general results assuming knife- edge recruitment to the fishery at age t^, with full vul- nerability thereafter. Age-specific fecundity equivalents (X(i); Table 2) were computed as Xa) = W(i)xPM{i), where W(/) = mean weight (kg) at age /; and PMii) = proportion of females mature at age i. and spawning stock biomass was computed as the product of fecundity equivalents and number of fish at age. Mean weight at age iW{i))and proportion of fe- males mature ^PM^ i ) ) reported by O'Brien and Brown^ were used in this analysis. The instantaneous natural mortality rate (Mii)) of haddock age 1 and older is 0.2 (Clark et al., 1982), and a maximum age of 15 was used following Gabriel et al. ( 1989). I computed annual R/SSB (Table 3) from the ratio of number of female fish at age 1 to their parental female spawning stock biomass (Clark et al., 1982; O'Brien and Brown^; Hayes and Buxton^) for the pe- riod 1931-94. For the entire data series, R/SSB aver- aged 0.5902. As noted by Gabriel et al. (1989), how- ever, the R/SSB ratio (reflecting age-0 survival) ap- pears to have declined following the collapse of the Georges Bank haddock stock during the early 1960s. 2 O'Brien, L., and R. W. Brown. 1996. Assessment of the Georges Banlt haddock stock for 1994. Northeast Fisheries Sdence Center Reference Document 95-13. National Marine Fisheries Service, Northeast Fisheries Science Center, Woods Hole, MA. 107 p. ■■' Hayes, D. B., and N. G. Buxton. 1991. Assessment of the Georges Bank haddock stock. Papers of the Northeast Regional Stock Assessment Workshop, Research Document SAW 13/1, 126 p. Table 2 Age-specific characteristics of Georges Bank haddock. Fecun- dity equivalents (i.e. spawning biomass per individual) are denoted as X. instantaneous natural mortality rate as M. and partial recruitment as PR. The proportion mature and mean weights at age are from O'Brien and Brown (see Foot- note 2 in the main text). Age Proportion mature Mean weight (kg) X M PR 0.00 — 0.000 — 0.00 1 0.08 0.486 0.039 0.2 0.00 2 0.46 0.676 0.311 0.2 0.07 3 0.88 1.197 1.053 0.2 0.65 4 1.00 1.621 1.621 0.2 1.00 5 1.00 2.021 2.021 0.2 1.00 6 1.00 2.424 2.424 0.2 1.00 7 1.00 2.780 2.780 0.2 1.00 8 1.00 3.192' 3.192 0.2 1.00 9 1.00 3.548' 3.548 0.2 1.00 10 1.00 3.924' 3.924 0.2 1.00 11 1.00 4.297' 4.297 0.2 1.00 12 1.00 4.667' 4.667 0.2 1.00 13 1.00 5.034' 5.034 0.2 1.00 14 1.00 5.398' 5.398 0.2 1.00 1.5 1.00 5.758' 5.758 0.2 1.00 ' Predicted from von Bertalanflfy growth equation fitted to observed data for ages 1 to 7. At present the causes of the decline in the R/SSB ra- tio are not known. Several hypotheses explaining the obsei"ved decline in R/SSB have been put forth, in- cluding depensatory mortality on age-0 haddock (Col- lie and Spencer, 1993), changes in oceanographic con- ditions (Myers and Pepin, 1994), and increased pre- dation or competition with elasmobranchs (Collie and Spencer, 1993 1. Although the cause is not known, a cru- cial consideration is the choice of an appropriate time period where R/SSB is representative of current pop- ulation abundance and biomass and current environ- mental conditions. One strategy to obtain a mean R/SSB value repre- sentative of "current" conditions is to average R/SSB from the most recent data point back several years. The philosophy behind this strategy is to smooth an- nual variations in R/SSB by averaging over a suffi- ciently long time period. The problem, however, is to define a time period sufficiently long to achieve ad-, equate precision without introducing excessive bias. Averages over short time periods suffer from low pre- cision and can vary considerably because of annual variation in R/SSB. If averages are taken over a time period spanning a wide range of population levels or Hayes: A biological reference point based on the Leslie matnx 79 Table 3 Stock-recruitment data for Georges Bank haddock. 1931 to 1990. Data for 1931 to 1962 Hayes and Buxton (see Footnote 3 in the main text), and from 1963 to 1994 from O'Brien and Brown (see Footnote 2 in the main text). Spawning stock Number of age- 1 Spawning stock Number of age- 1 biomass (t) females produced biomass (t) females produced Year females only (millions) R/SSB Year females only (millions) RySSB 1931 75,767 22.793 0.3008 1963 82,128 235.939 2.8728 1932 61,586 26.052 0.4230 1964 64,278 16.577 0.2579 1933 52,643 30.239 0.5744 1965 72,512 2.013 0.0278 1934 47,187 29.436 0.6238 1966 90,260 6.426 0.0712 1935 52,104 29.323 0.5628 1967 56,051 0.211 0.00.38 1936 54,967 53.173 0.9674 1968 37,546 0.494 0.0132 1937 52,872 39.626 0.7495 1969 25,589 2.330 0.0911 1938 60,434 29.875 0.4943 1970 19,255 0.184 1939 70,067 55.903 0.7979 1972 13,456 9.707 0.7214 1940 67,418 57.052 0.8462 1973 6,135 5.270 0.8590 1941 76,187 30.731 0.4034 1974 10,906 3.827 0.3509 1942 77,248 11.783 0.1525 1975 9,190 51.616 5.6165 1943 75,511 32.411 0.4292 1976 11,031 6.891 0.6247 1944 68,957 20.806 0.3017 1977 20,717 3.029 0.1462 1945 62,785 46.548 0.7414 1978 34,427 41.941 1.2183 1946 65,712 30.486 0.4639 1979 33,841 5.052 0.1493 1947 57,269 16.676 0.2912 1980 31,703 3.600 0.1136 1948 55,659 62.918 1.1304 1981 27,757 1.230 0.0443 1949 45,791 29.447 0.6431 1982 22,682 1.519 0.0670 1950 55,919 52.810 0.9444 1983 17,531 8.548 0.4876 1951 54,039 21.795 0.4033 1984 12,583 0.865 0.0687 1952 55,497 66.748 1.2027 1985 10,313 7.233 0.7013 1953 54,772 26.276 0.4797 1986 10,186 0.882 0.0866 1954 65,764 46.429 0.7060 1987 9,540 8.024 0.8411 1955 68,949 30.534 0.4428 1988 8,781 0.580 0.0661 1956 78,979 31.155 0.3945 1989 9,061 1.244 0.1373 1957 75,364 29.630 0.3932 1990 9,762 0.991 0.1015 1958 75,995 63.270 0.8326 1991 8,682 4.125 0.4751 1959 77,394 61.685 0.7970 1992 6.231 7.219 1.1586 1960 93,667 26.950 0.2877 1993 5,001 3.754 0.7507 1961 113,366 19.368 0.1708 1994 7,324 3.938 0.5377 1962 109,001 95.348 0.8747 spanning a trend in environmental conditions, the estimate of future FJ/SSB may be biased. For this stock, mean R/SSB for periods of 5 to 15 years appear relatively stable. Over longer periods of time, mean R/SSB shows an upward trend, punctuated by sharp increases corresponding to the 1963 and 1975 year classes when R/SSB ratios were much higher than in any other years. Because of the trends observed over longer periods of time, I chose the time period from 1976 to 1994 as representative of "current" conditions for R/SSB, or equivalently, age-0 survival. For this time period, R/SSB averaged 0.4092, and had a me- dian of 0.1493. Based on the mean R/SSB for 1976 to 1994, the had- dock population would be expected to grow at a rate of 18.0% per year with no exploitation (i.e. A,^=1.180). For knife-edge recioiitment, I computed A^, for fishing mortality rates ranging from to 2.0 (Fig. 1) and for ages at entry (t^) from 1 to 5 years and for the commer- cial fishery age selectivity observed in 199li^-94 (Table 2). Additionally, I determined F^, and ^MSP (following the methods of Gabriel et al., 1989) for each age at entry (Table 4). It is apparent ft-om this analysis that as age at entry is delayed, the impact of fishing on the popula- tion is decreased (Fig. 1). Thus, higher fishing mortality rates can be sustained when recruitment to the fishery is delayed (Table 4). In fact, when the age at entry is 5 or greater, any level of fishing mortality is sustainable. These conclusions are not new; analysis of SSB/R yields similar insights into the response of populations 80 Fishery Bulletin 98(1 0.2 to harvesting. The analysis of the LesHe matrix offers information not available in the analysis of SSB/R, however. First, the consequences of ovei"fishing or "un- derfishing" are clearly evident from the graph of A, against fishing mortality rate. For example, for an age at entry of 3 years, F^^ is 0.465. If the fishing mortal- ity rate was limited to 0.20 (for exam- ple), the population would be expected to increase at a rate of about 8.8% per year (Fig. 1). Likewise if fishing mortal- ity was increased to 1.0, the population would be expected to decline at a rate of about 10.3% per year (Fig. 1). When %MSP is plotted against lambda resulting from various levels of fishing mortality and ages at entry into the fish- ery, it is apparent that equal %MSP val- ues are obtained for the same lambda only at two points along each of the curves. The first point is for the unfished population when lambda is at a maxi- mum and there is 100% MSP. The second point where all 7f MSP values are equal is when lambda is equal to 1.00 (Fig. 2). These results demonstrate the assertion that fishing mortality rates that result in equal %MSP values do not necessarily result in the same population dynamics (i.e. the same rate of increase or decrease). The reason for this disparity is that in a growing or declining population, the timing of reproduction dur- ing the lifetime is important, as well as the total life- time egg production. For example, when a population is growing, earlier realization of lifetime spawning poten- tial contributes more to population growth than later reproduction. This relationship is evident when the for- mula for lifetime spawning stock biomass (on which %MSP is based) is compared to the formula for repro- ductive value, upon which the rate of population change depends. Observe that lifetime spawning stock biomass per newborn individual is (Gabriel et al., 1989) Age at entry 5 Maintenance 4 level Current 3 0.0 —\ — 0.5 — I — 1.0 1.5 — I — 2.0 2.5 Instantaneous fishing mortality Figure T Rate of population increase (lambda) in relation to instantaneous fishing mortality (F) for a range of age at entry into the fishery. SSB/Nq = S(0) W( 1 )PM( 1 ) -t- S( )S( 1 ) W( 2 )PM ( 2 ) -i- S(0)S(1)S(2)W(3)PM(3) + ... . This formula is equivalent to that for the "net repro- ductive rate" (Caswell, 1989) which is the expected number of offspring produced by a newborn over its lifetime. With the above notation, the reproductive value (Caswell, 1989) of an age-1 individual can be ex- pressed as Reproductive value = S(0)S(1)W(2)PM(2)A-' + S( |S( 1 )S( 2 )W( 3 )PM( 3 )A-2 + S(0)S(l)S(2)S(3)W(4)PM(4)A-3 + Table 4 Sustainable fishing mortality (F^,) and %MSP as a function of age at entry (tj for Georges Bank haddock. t. F„ '?MSP 27.20 27.20 27.20 27.20 27.20 1 0.236 2 0.309 3 0.465 Current 0.519 4 0.956 5 >2.0 These formulae are similar except for the addition of the term A ', where / is the age index. Classical demo- graphic theory shows that the growth rate of a popu- lation is dependent on the reproductive value rather than on the net reproductive rate (Caswell, 1989). These two quantities are clearly related, however. Currently, Georges Bank haddock become vulner- able to the fishery at age 2 but are not fully recruited until age 4 (Table 2 1. With this partial recruitment vector, F, is 0.519. The graph of A, against F (Fig. 1) indicates that if F is held at its 1994 value of about 0.29, the population would be expected to increase at a rate of 6.35% per year If F is reduced to 0, the stock would be expected to increase at a rate of 18.0% per year. The expected rate of increase when F is below F^i is particularly pertinent to cases where stock re- building is desired because this analysis allows the de- Hayes: A biological reference point based on the Leslie matnx 81 Si termination of how rapidly the stock will be rebuilt under various levels of fishing mortality. Developing a Leslie matrix representation of harvesting: stochastic case One of the major challenges facing fish- ery managers is to determine appropriate reference points for fish populations that show variable recruitment. When one or more elements of the Leslie matrix vary in a stochastic fashion, no general closed- form expressions for the growth rate of the population are available (Tuljapurkar, 1989). Results of theoretical studies of sto- chastic Leslie matrices are useful, how- ever, in guiding the analysis and interpre- tation of matrices with entries that vary over time. Two results summarized by Tuljapurkar (1989) that are particularly useful in this analysis are 1 The analog to lambda for deterministic matrices is the mean population growth rate, in contrast to the growth rate of the average population. This is equivalent to the mean rate of change in the loga- rithm of population size (N). 2 The distribution of projected population size (AO over time tends towards a lognormal distribution when the dynamics are governed by a stochastic matrix. From these results, maintenance fishing mortality can be defined as the fishing mortality that results in an average population growth rate of (equivalent to Ag=l). Because this measure is analogous to lambda, I will use the symbol A for its representation but em- phasize that computationally the measures for deter- ministic and stochastic Leslie matrices differ. An im- portant corollary of the above two results is 3 A population growing deterministically at the mean growth rate does not produce the mean of the pop- ulation sizes produced in the stochastic represen- tation. Nor does a deterministic matrix composed of the means of the stochastic matrices produce a population with the same dynamics as applying the stochastic matrices. To illustrate these theoretical results, I performed a simulation of the Georges Bank haddock stock dy- namics using a stochastic Leslie matrix. In this case, I focused on the effects of stochastic age-0 survival as represented by R/SSB. I performed this simulation 1.15 - Age 1 ^^^^^""^"^ 1.10- Age 2 Age 3 ,:^p^ 1,05 - Age 4 Ages y f 1.00 - 0.95 - III) 20 40 60 % MSP 80 100 Figure 2 Relationship between rate of population increase (lambda) and percent maximum spawning potential C/rMSP) for a range of age at entry. by projecting a starting population forward in time, with the value for R/SSB for each year selected with equal probability from observed values fi-om 1976 to 1994. Five thousand replicates were simulated for a 150-year period. Results of these simulations are in accord with the theoretical assertion that TV, is distributed lognor- mally; for times greater than 110 years, the ln(A^,) did not differ significantly from a normal distribution at a=0.05. It is interesting to note that A^, is lognor- mally distributed, even though the stochastic element (R/SSB) was not normally or lognormally distributed itself The lognormal distribution of A^, arises from the fact that TV, is the result of the process of sequential multiplications of random elements (Aitchison and Brown, 1976). When the distribution of population size is plotted over time (Fig. 3), it is clear that the variance increases rapidly. When year-to-year popu- lation growth rates (i.e. TV^^/TV,) are computed for in- dividual simulation results, the distribution of growth rates shows an initial transient response for the first 5 years but thereafter settles into a stable distribution from year to year (Fig. 4). Because of the transient dy- namics, I began the evaluation of long-term dynamics with year 10. One of the critical theoretical results is that the growth rate of a population governed by stochastic rates tends towards a single value in the long run. This is what Tuljapurkar (1989) terms the "almost sure population growth rate." When the growth rate is computed over progressively longer intervals, the distribution shows a convergence on nearly the same 82 Fishery Bulletin 98(1) 750 - 500 J 250 . 500 250 . . 500 250 . . 500 >, 250 . J 0. g- 500 . £ 250 . . 500 . 250 . . 500 . 250 . . 500 . 250 . Year 3 )00 Stic bu- on i Year 4 \ Year 5 Year 6 I Year 7 1 Year 8 \ Year 9 1 Year 10 ( Distributi projection tion of po 5000 simi ) 5,000 10,000 15,000 20,000 25, Population size Figure 3 on of population sizes to year 10 for a stocha of the Georges Bank haddock stock. The distri pulation sizes for each time interval is based ilations. Arrows indicate mean for each year. Population growth rate Figure 4 Distribution of annual population growth rates (lambda! to year 10 for a stochastic projection of the Georges Bank Had- dock stock. The distribution of growth rates for each time interval is based on 5000 simulations. mean value (Table 5) with a decreasing variance (Fig. 5). This result has both theoretical and pragmatic im- plications. Of theoretical importance is the concept that although the variance of projected population abundance increases over time, the variance of the growth rates declines over time. Thus mean popu- lation growth rate can be used in defining a main- tenance fishing mortality rate. The practical conse- quence of the above result is that at least two different strategies can be used to compute A for a stochastic population. One strategy is to project the population for a long period of time (e.g. hundreds of years) to make a precise estimate of the long-term population growth rate. This strategy makes use of the fact that the variance of the long-term population growth de- clines as the period of projection is lengthened. A prob- lem with this approach, however, is that for projec- tions over a long period of time, population abundance can become so large or small that it cannot be directly represented on a digital computer, causing a numeric overflow or underflow. A preferable strategy is to com- pute A for a large number of simulations over a shorter time period (e.g. 150 years). This method avoids the problem of numeric overflow and achieves precision in the estimate of mean A by having a large number of simulations. Based on the current partial recruitment vector to the fishery, a fishing mortality of 0.450 (F,,) would result in an average population growth rate of zero (Table 6). The fishing mortality that results in a zero growth rate for the mean population size was higher, at 0.517 (Table 6). Interestingly, this is nearly the same as F^^ computed for the deterministic case by us- ing the mean R/SSB. The estimate of F^^^^y with these same data is much lower than F,, — only 0.069 (Table 6). When a deterministic Leslie matrix analysis was made with the corresponding median R/SSB, it re- sulted in an estimate of F,, that was nearly the same as F^^^ (Table 6). This finding illustrates the basic connection between these methods when they are op- erating on the same inputs. As an additional compari- son, I computed SSB/R as the ratio ISSB/SR instead of the mean (or median) of the individual year ratios. This method is based on sampling theory that sug- gests that the ratio of the sums is a less biased estima- Hayes: A biological reference point based on the Leslie matrix 83 1.200 - 800 _ Year 10-20 400 _ 1.200 ^ _^ 1^^^^_ 800 - Year 10-30 400 - 1 .200 - _^-7^^^ 800 - Year 1 0-40 400 - 1.200 - ^^T^N^ u 800 - Year 10-50 = 400 - a" 1! U. 1.200 - ./T^ 800 - Year 10-60 400 - 1.200 - yi V 800 - ^-^ Year 10-70 400 - 1.200 - yi \ n, 800 - / \ Year 10-150 400 - /A y t \ l.( ) 1.1 1.2 1.3 1.4 Population growth rale Figure 5 Distribution of population growth rates over various time scales for a stochastic projection of the Georges Bank had- dock stock. The distribution of growth rates for each time interval is based on 5000 simulations. Arrows indicate mean for each year. Table 5 Mean and coefficient of variation (CV) of 5000 simulations of | long run population growth rate for Georges Bank Haddock | without fishing. Time interval Mean CV Year 10-20 0.1552 37.02 Year 10-30 0.1555 23.28 Year 10-40 0.1558 18.03 Year 10-50 0.1559 15.23 Year 10-60 0.1559 13.34 Year 10-70 0.1560 12.07 Year 10-150 0.1.563 7.60 Table 6 Comparison of F^,, with deterministic analyses, analyses, and with % MSP calculations. stochastic Type of simulation Recruitment input F„ Deterministic Mean R/SSB 0.519 Deterministic Median R/SSB 0.070 Stochastic (mean of rates) Uniform probability 0.450 Stochastic (mean of populations) Uniform probability 0.517 '/fMSP Mean SSB/R 0.519 -XMSP Median SSB/R 0.069 7, MSP SSB/R^ISSB/lR 0.449 tor of the "true" ratio than is the mean of the individ- ual ratios (e.g. Cochran, 1977). The value for SSB/R obtained with this method was 2.685, and when used in place of the median SSB/R in the F„,^,^y calculations, re- sulted in an estimated reference point of 0.449 — essen- tially the same as in the stochastic simulation where the mean population growth rate was zero (Table 6). Discussion The primary purpose of this paper was to demonstrate how the Leslie matrix can be used to compute a refer- ence point for harvested populations and to contrast this method with the SSB/R method currently in use. One of the major findings was that the SSB/R method and the Leslie matrix approach produce similar es- timates of sustainable fishing mortality when they are based on the same inputs. This is not surprising given the similarity between reproductive value, on which the Leslie matrix is based, and lifetime spawn- ing stock biomass per recruit which the SSB/R ap- proach uses. Although the two methods produce similar esti- mates of sustainable fishing mortality, the Leslie ma- trix approach is preferable because of the additional direct information it provides regarding the popula- tion response to fishing at levels different from the ref- erence point. Furthermore, when population growth rate is different fi-om zero, equal levels of SSB/R do not result in the same population gi'owth rate for differ- ent partial recruitment vectors. These differences are small, however, in relation to changes in population growth rate owing to changes in fishing mortality. Given the various approaches illustrated (e.g. de- terministic vs. stochastic), the basic question is what method to use. On the basis of theoretical advances in the population dynamics literature and results pre- sented here, I recommend the use of a stochastic anal- ysis where the mean of the population growth rates is used as the 'Tjest" measure of growth for harvested populations. A stochastic analysis is preferable be- 84 Fishery Bulletin 98(1) cause it can fully represent the information contained in the distribution of observed R/SSB values. Also, a stochastic simulation can be used to provide a mea- sure of the uncertainty associated with estimates of the biological reference point. The use of the mean population growth rates instead of the mean of the population sizes is justified on theoretical grounds (Tuljapurkar, 1989). As shown in Table 6, use of the mean R/SSB in a deterministic analysis or the rate of growth of the mean population size in a stochastic simulation tends to result in higher estimates of sus- tainable fishing mortality (or, alternatively, a higher estimate of population growth rate for a given fishing mortality rate) than does the stochastic simulation where the mean of population growth rates are used as the output. 1 feel that the reason for this difference hinges on the distinction between projections and fore- casts (Caswell, 1989) . If we view the long-term simu- lation results as forecasts, this implies that we could use the distribution of population size ( and the mean of the distribution) as the best estimate of the future state of the population. In these simulations, however, the mean population size is strongly influenced by the very high values that occur in the right-hand tail of the lognormal distribution. These estimates are far above what has ever been observed for this stock, and are probably not biologically realistic. As such, they should not be treated as true forecasts of the future population. In contrast, if we view the analysis as a projection, the goal is not to forecast future popula- tion size but rather to use the results to detennine the population growth rate that is represented by the cur- rent Leslie matrix. By knowing the curi'ent population growth rate, it is possible to find the fishing mortality rate that maintains an expected value for population growth of zero, which would result in a statistically stationary population. In a stochastic setting, the entire distribution of FJ/SSB ratios is used to portray the reproductive suc- cess for the stock. When a deterministic analysis is de- sired, however, it is necessary to choose among several possible measures of central tendency for the R/SSB ratio. Sissenwine and Shepherd (1987) advocated the use of the median R/SSB ratio as a way of robustly portraying the reproductive success of a stock. Their rationale was that the frequency of relatively poorer recruitment is balanced by years of better recruitment when the median R/SSB is used. Table 6 illustrates, however, that using the median of the observed R/SSB ratios can result in estimates of sustainable fishing mortality that are substantially different from the sto- chastic simulations, which are arguably the best to represent the population's dynamics. The use of the mean of the observed R/SSB ratio as the measure of central tendency can likewise result in estimates of sustainable fishing mortality that differ from the standard set by stochastic simulations. The primary reason for this difference is that use of a mean of the observed ratios is biased high in relation to the preferred estimator of the ratio ( in this case the sum of recruitm'ent over the sum of spawning stock bio- mass; Cochran, 1977). As such, the use of the mean of the observed R/SSB values can also be inaccurate. Among the measures presented here, the estimator Ireciaiitment/Sspawning stock biomass should be used as the measure of central tendency for the R/SSB ratio in deterministic analyses. The use of this measure re- sults in point estimates of sustainable fishing mortal- ity that are essentially the same as a full stochastic analysis. Although the Leslie matrix is a useful tool to por- tray the dynamics of harvested populations and to de- termine appropriate reference points, several issues arise that are of considerable practical importance. As alluded to earlier, a significant challenge is how to detennine what is an appropriate distribution for the R/SSB ratio. Because of the variability in R/SSB over time and the occurrence of occasional large year classes, it is very difficult to determine what time frame is representative of the present. Although the answer to this question is beyond the scope of this pa- per, 1 feel that the best approach is to plot the mean R/SSB ratio over progressively longer time periods back from the present to determine if there are any temporal trends or epochs in the data set. The analyst should then use his or her judgment based on other biological information over time (such as stock size, mean weight per individual at age, and maturation schedule) to determine an appropriate period to use as the basis for stochastic simulations. It is important to emphasize that the dilemma of choosing a representa- tive time period is not unique to analyses in which the Leslie matrix is used, and the same problem arises for computing any biological reference point. In addition to the difficulty of determining what is a representative time period for the present population, a fundamental question is how to represent the dy- namics of populations with a density-dependent stock- recruitment relationship. In principle, this can be ap- proached by altering the R/SSB ratio as a function of stock size (e.g. Quinn and Szarzi, 1993). Although I agree with Quinn and Szarzi's (1993) approach, the challenge of accurately specifying the distribution of R/SSB ratios at different stock sizes is even greater than specifying the current distribution. As a final comment, biological reference points for fish populations are not necessarily targets for fish- ery management (Mace, 1994), nor are they inviolate boundaries that may not be crossed. Rather, they are most useful as a means of comparing the consequences Hayes: A biological reference point based on the Leslie matrix 85 of different choices among fishery management op- tions. For example, it is appropriate to allow fishing mortality to exceed the biological reference point if the goal is to reduce an overly abmidant fish stock. Like- wise, they can be useful in projecting the likely growth of a population under more restrictive fishery man- agement measures. In the end, however, they may be most useful as a reminder and a warning that there are limits to the productive capacity offish population and that if we consistently exceed their limits, popu- lation declines are almost certain to occur (Francis, 1997; Myers, 1997). Acknowledgments I wish to thank Jim Bence, Mike Jones, Mike Rutter, and Terrance Quinn II for their insightful reviews of this manuscript. The support of the National Marine Fisheries Service, the Michigan State University Agri- cultural Experiment Station, the Fisheries Division of the Michigan Department of Natural Resources, and the Department of Fisheries and Wildlife at Michigan State University is also gratefully acknowledged. Literature cited Aitchison, J., and J. A. C. Brown. 1976. The lognomial distribution with special reference to its uses in economics. Cambridge Univ. Press, Cambridge, England, 176 p. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. U.K. Ministry of Agriculture, Fisheries and Food Fishery Investi- gations (series 2) 19, 533 p. CasweU, H. 1989. Matrix population models. Sinauer Associates, Sun- derland, MA, 328 p. Clark, S. H., W. J. Overholtz. and R. C. Hennemuth. 1982. Review and assessment of the Georges Bank and Gulf of Maine haddock fishery. J. Northwest Atl. Fish. Sci. 3:1-27. Clark, W. G. 1991. Groundfish exploitation rates based on life history parameters. Can. .J. Fish. Aquat. Sci. 48:734-750. Cochran, W. G. 1977. Sampling techniques. Wiley and Sons, New York, NY, 428 p. Cohen, J. E., S. W. Christensen, and C. P. Goodyear. 1983. A stochastic age-structured population model of striped bass (Morone saxatilis) in the Potomac River. Can. J. Fish. Aquat. Sci. 40:2170-2183. CoUie, J. S., and P. D. Spencer. 1993. Management strategies for fish populations subject to long-term environmental variability and depensatory preda- tion. In G. Kruse, D. M. Eggers, R. J. Marasco, C. Pautzke, and T. J. Quinn. II (eds. ), Proceedings of the international symposium on management strategies for exploited fish pop- ulations, p. 629-650. Alaska Sea Grant College Program Report 93-02, Univ Alaska, Fairbanks, AK. DeAngelis, D. L., S. W. Christensen, and A. G. Clark. 1977. Responses of a fish population to young-of-the-year mortality. J. Fish. Res. Board Can. 34:2124-2132. Francis, R. I. C. C. 1997. Comment: How should fisheries scientists and manag- ers react to uncertainty about stock-recruit relationships? Can. J. Fish. Aquat. Sci. 54:982-983. Gabriel, W. L., M. P. Sissenwine, and W. J. Overholtz. 1989. Analysis of spawning stock biomass per recruit: an example for George Bank haddock. N. Am. J. Fish. Manage. 9:383-391. Getz, W. M., and R. C. Haight. 1989. Population harvesting: demographic models of fish, forest and animal resources. Princeton Univ. Press, Prince- ton, NJ, 391 p. Goodyear, C. P., and S. W. Christensen. 1984. Bias-elimination in fish population models with sto- chastic variation in survival of the young. Trans. Am. Fish. Soc. 115:627-632. Gulland, J. A., and L. K. Boerema. 1973. Scientific advice on catch levels. Fish. Bull. 71:325- 335. Johnson, L. W., and R. D. Riess. 1981. Introduction to Unear algebra. Addison-Wesley, Read- ing. MA, 358 p. Keyfitz, N. 1977. Introduction to the mathematics of populations with revisions. Addison-Wesley. Reading, MA, 490 p. Mace, P. M. 1994. Relationships between common biological reference points used as thresholds and targets of fisheries manage- ment strategies. Can. J. Fish. Aquat. Sci. 51:110-122. Myers, R. A. 1997. Comment and reanalysis: paradigms for recruitment studies. Can. J. Fish. Aquat. Sci. 54:978-981. Myers, R. A., and P. Pepin. 1994. Recruitment variability and oceanography stability. Fish. Oceanogr. 3:246-255. Overholtz, W. J., M. P. Sissenwine, and S. H. Clark. 1986. Recruitment variability and its implication for manag- ing and rebuildmg the Georges Bank haddock iMelanogram- mus aeglefinus) stock. Can. J. Fish. Aquat. Sci. 43:748-753. Pielou, E. C. 1974. Population and community ecology . Gordon and Breach Science Publishers, New York, NY, 424 p. Quinn, T. J., II, and N. J. Szarzi. 1993. Determination of sustained yield in Alaska's recre- ational fisheries. In G. Kruse. D. M. Eggers. R. J. Mar- asco, C. Pautzke, and T J. Quinn, II (eds.). Proceedings of the international symposium on management strategies for exploited fish populations, p. 61-84. Alaska Sea Grant Col- lege Program Report 93-02. Univ. Alaska, Fairbanks, AK. Rothschild, B. J., and M. J. Fogarty. 1989. Spawning-stock biomass: a source of error in recruit- ment/stock relationships and management advice. J. Cons. Int. Explor. Mer 45:131-135. Sissenwine, M. P., and J. G. Shepherd. 1987. An alternative perspective on recruitment overfishing and biological reference points. Can. J. Fish. Aquat. Sci. 44:913-918. Tuyapurkar, S. 1989. An uncertain life: demography in random environ- ments. Theor Pop. Biol. 35:227-294. 86 Abstract.— We conducted a large-scale field experiment to test whether clam and oyster harvesting applied alone and in combination on intertidal oyster reefs have impacts on resident shellfish pop- ulations. This experiment was conducted to resolve a long-standing conflict be- tween oyster iCrassostrea virginica (Gmelin, 1791)) and clam (Mercenaria mercenaria (Linneaus, 1758)) fishermen who contend that the other fishery causes high rates of mortality to their respective species. Intertidal oyster reefs located in two estuarine creeks near Wilmington, North Carolina, were harvested for clams only, oysters only, both clams and oys- ters, or were left undisturbed as controls. Experimental harvesting was conducted over a one-year period by a professional shellfisherman who used realistic fish- ing techniques (clam rakes and oyster tongs), intensity, and frequency. Har- vesting impact on hard clam and oyster populations was assessed by sampling naturally occurring oysters before and after harvesting, and sampling both nat- urally occurring clams (all size classes) and transplanted, hatchery-raised clams (20-37 mm in length) after harvesting. Clam and oyster harvesting had obvious negative effects on populations of oys- ters. There was a substantial decrease in the number of live oysters on clam-har- vested and oyster-harvested reefs com- pared with unharvested, control reefs. Clam and oyster harvesting, applied together, reduced oyster densities and killed unharvested oysters at a level sim- ilar to that caused by each type of har- vesting applied separately. The effects of the shellfish harvesting on populations of hard clams varied between the two sites (i.e. creeks). In both creeks, clam har- vesting, alone and combined with oyster harvesting, significantly decreased the number of live, naturally occurring clams. Oyster harvesting alone decreased the number of live, naturally occurring clams only at one site. Clam harvesting also decreased the number of live, trans- planted clams on reefs, but there was no effect of oyster harvesting, because the transplanted clams were juveniles too small to be harvested with oyster tongs. Overall, the combined effect of both types of harvesting applied together did not have a negative synergistic effect on clam and oyster populations. Conse- quently, both clamming and oyster har- vesting should be permitted on some reefs, but maintaining large populations of oysters and clams on intertidal oyster reefs will require protection of some reefs from both types of harvesting. Manuscript accepted 24 March 1999. Fish. Bull. 98:86-9.-) (2000). Biological effects of shellfish harvesting on oyster reefs: resolving a fishery conflict by ecological experimentation Hunter S. Lenihan Beaufort Laboratory National Marine Fishenes Sen/ice, NCAA 101 Piver's Island Road. Beaufort. North Carolina 28516 Present address Institute of Manne Sciences University of North Carolina at Chapel Hill, 3431 Arendell St., Morehead City, North Carolina E-mail address fisleniha(S)email uncedu Fiorenza Micheli National Center for Ecological Analysis and Synthesis Santa Barbara, California 93101 Marine fisheries are an important source of employment and protein for humans but can negatively affect marine organisms and ecosystems (Dayton et al., 1995; Engel and Kvitek, 1995; Botsford et al., 1997). The most obvious negative ecolog- ical effects of fishing result from over-harvesting of target species, incidental mortality of non target spe- cies ("bycatch"), and fishery-related disturbances to marine habitat ( FAO, 1993; Dayton et al., 1995). Of course, fisheries over-exploitation and hab- itat destruction also threaten the sustainability of the fishing indus- try. At present, 44'^ of the worlds fish stocks are fully to heavily exploited, and 22'7( are overexploited or depleted, indicating most fisher- ies are not managed for long-term sustainability (Botsford et al., 1997). The degradation and destruction of marine biogenic habitat (e.g. coral reefs, seagrass beds, mangrove for- ests, and oyster reefs) by dredging, trawling, use of explosives, and poi- soning reduces fishery production by removing habitat essential for the recruitment, growth, and survival of fishery and prey organisms ( Winslow 1881, a and b; Peterson et al, 1987; Norse, 1993; Rothschild et al., 1994; NRC, 1995; Lenihan and Peterson, 1998). The sustainability of a fishery is often threatened when competing fisheries exploit a common resource or negatively impact a commonly used habitat. For example, when the bycatch of one fishery is within a food web supporting another fishery (West and Gordon, 1994), or when a fishery destroys habitat impor- tant to the life history of other fish- ery species (Russ and Alcala, 1996), heated political battles arise and the livelihood of many people may be lost. Resolving such fishery conflicts has important ecological and eco- nomic consequences and is of major concern to fisheries managers and ecologists worldwide (McAllister and Peterman, 1992). This paper pres- ents the results of an experimental analysis of whether two economi- cally valuable fisheries conflict and provides management recommenda- tions to resolve the conflict. High productivity of fishery stocks in estuaries and shallow water coastal habitats often induces intense exploi- tation of a common species or habitat by multiple, potentially conflicting fisheries (Peterson et al., 1987). In many estuaries along the Atlantic coast of the USA, intertidal oyster reefs are harvested for hard clams {Mercenaria mercenaria ) year round, Lenihan and Micheli: Biological effects of shiellfisfi fiarvesting on oyster reefs 87 and for oysters iCrassostrea virginica ) in the fall and winter (i.e. October-March). In recent years, clam and oyster (i.e. "shellfish") harvesting on oyster reefs has led to conflict between the two fisheries, and between fishermen and habitat managers over the issue of hab- itat degradation, especially in the southeastern United States (e.g. Frankenberg^; Noble-^). Oyster fishermen claim that the harvesting of clams from intertidal oyster reefs decreases resident oyster populations, and vice-versa, because each type of fishing kills or removes the other species. Rakes and hand tongs used for the two types of shellfishing appear to increase the mortality of the sessile reef animals by burying them beneath sediments, fracturing their shells, or causing other physical damage (Noble^). In addition, bodies of dead and wounded animals left behind may attract scavengers and predators, thereby increasing preda- tion intensity on healthy live animals (Dayton et al., 1995). Habitat and fishery managers are concerned that the physical destruction of oyster reefs caused by shellfishing will negatively affect many other fishery organisms that recruit to and utilize oyster reef habi- tat, including many species of fishes (Breitburg et al. 1995, Lenihan et al., 1998, Luckenbach et al., 1998) and the blue crab (Callinectes sapidus (Rathbun)) (Bahr and Lanier, 1981; Zimmerman etal., 1989; Leni- han et al., 1998; MicheU and Peterson, 1999). Shell- fishing also reduces the overall size of reefs because shell material is broken or removed along with the target species (Lenihan and Peterson, 1998; Coen^). Reducing the size of reefs is thought to decrease the abundance of clams because less habitat is available for adults and recruits (Arnold, 1984; Sponaugle and Lawton, 1990; Peterson et al., 1995). Decreasing the size (i.e. height) of oyster reefs also reduces oyster production because flow speed over reefs diminishes, causing sediment to accumulate and oyster growth and survival to decrease (Lenihan, 1999; Lenihan et al., 1999). In contrast to the negative effects of shell- fish harvesting, many fishermen claim that "turning over" the shell matrix of oyster reefs during harvest- ing improves clam and oyster production because it removes accumulated sediment. In North Carolina and many other Atlantic coast states, both types of shellfishing are allowed on reefs and conflicts between the fisheries continue (e.g. Frankenberg'; Marshall''). ' Frankenberg, D. 1995. Report of North Carolina Blue Ribbon Advisory Council on oysters. North Carolina Department of Envi- ronmental Health, and Natural Resources, Raleigh. NC, 101 p. ^ Noble, E. B. 1995. Destruction of oyster rocks by individuals taking clams by legal hand harvest methods. Report of the North Carolina Division of Marine Fisheries, Morehead City, NC, 11 p. ■^ Coen. L. D. 1995. Areview of the potential impacts of mechani- cal harvesting on subtidal and intertidal shellfish resources. A report prepared by the South Carolina Department of Natural Resources, Marine Resources Research Institute, SC, 111 p. Whether oyster harvesting, clam harvesting, or both types of fishing activities together have negative impacts on shellfish populations of intertidal oyster reefs is a matter of opinion and has not been tested experimentally. A comparison of the biological impact of various fish- ing practices by using large-scale field experiments is an efficient method of resolving many fishery-related conflicts (McAllister and Peterman, 1992) and is an important component of adaptive management strat- egies (Walters, 1986). Such experiments are usually designed so that replicate areas (i.e. treatments) are fished, by using each technique separately and by using a combination of techniques, while other areas (i.e. con- trols) are closed to fishing. For these experiments to be meaningfial, they must be conducted on realistic tem- poral and spatial scales, and the fishing treatments must be applied through the actual fishery (McAllister and Peterman, 1992). The success of such experiments also depends heavily on close working relationships among fishermen, fisheiy ecologists, and fishery man- agers (Grumbine, 1997). For adaptive management, the results of initial (i.e. "prototype") experiments are used to design new management strategies that are subsequently tested on even larger temporal and spa- tial scales. Such adaptive management strategies and the use of experimental approaches are often discussed in fisheries management, but in reality are rarely attempted (e.g. Walters, 1986; Botsford et al., 1997). We conducted a large-scale field experiment to test the effects of hard clam and oyster harvesting, sepa- rately and in combination, on oyster and hard clam populations living on intertidal oyster reefs in North Carolina. Specifically, we tested whether 1) the den- sity of live and dead oysters varied among oyster reefs that were harvested for clams, harvested for oys- ters, harvested for clams and oysters, or were unhar- vested; 2) the density of live and dead clams varied among oyster reefs subjected to the same four harvest- ing treatments; and 3) the joint application of both shellfish harvesting practices has a synergistic (i.e. a multiplicative) effect on each species. If applying both types of harvesting to the same reefs enhances poten- tial negative effects of each harvesting type, a possible management option would be to allow clam and oyster harvesting only on separate reefs. This experiment was designed and conducted with the combined effort of a clam and oyster fisherman, "> ecologists,^ and habitat ^Marshall, M. 1996. North Carolina Division of Marine Fisheries, 3431 Arendell St.. Morehead City, NC, 28557. Personal commun. ■'' Cummings, R. A. 1996. For address contact H. S. Lenihan. Institute of Marine Sciences, 3431 Arendell, Morehead City, NC 28557. Personal commun. " Peterson, C. H., and H. C. Summerson. 1997. Institute of Marine Sciences, 3431 Morehead City, NC 28557. 88 Fishery Bulletin 98(1) managers J and is a prototype experiment for adaptive management of shellfisheries in southeastern North America. Methods Study sites The intertidal oyster reefs used in this study were located in two large tidal creeks, Pages and Whiskey Creeks, sit- uated on the Intercoastal Waterway near Masonborough Inlet, Wilmington, North Carolina. The two creeks con- sisted of well-flushed sandy to muddy bottom tidal chan- nels 0-2 m in water depth. Channels in each creek were separated by small to large patches of marsh iSpartina alterniflora ) habitat surrounded by oyster reefs created by Crassostrea virgimca. The two creeks were chosen because they have been permanently closed to fishing for about the last ten years owing to high coUform bacteria counts caused by the seepage of septic tanks from sur- rounding homes. Both creeks are highly productive, sup- porting large populations of fishes, birds, crabs, clams, and oysters. Tides in each creek are predominantly M-2 lunar tides, and the tidal range is 1-2 m in both creeks. Four large oyster reefs (9-13 m wide x 45-55 m long), each containing relatively high densities of oysters and hard clams, were chosen in each creek. The reefs were situated 150-200 m from the mouth of each creek. The salinity near the experimental reefs was 22-34 psu throughout the course of the experiment and water tem- perature was 3-30°C. Three to five permanent 6-m long x 1-m wide tran- sects were established haphazardly on each oyster reef by using PVC poles with rebar anchors between 1-14 June 1996. A total of sixteen transects were estab- lished in each creek at approximately 0.5 m above the mean low tide level. The sixteen transects pro- vided a total of four replicates of each of the following four harvest treatments: clam harvesting only, oyster harvesting only, clam harvesting and oyster harvest- ing combined, and no harvesting. Reefs and transects were located in areas where disturbances caused by shellfishing, boat traffic, or other human activities did not normally occur. We found no evidence suggesting that experimental reefs were physically or chemically disturbed throughout the course of the experiment. Sampling of clams and oysters The density of live and dead oysters on each experi- mental oyster reef was measured between 5 and 10 ^ Carpenter, R., and M. Marshall. 1996. North Carolina Division of Marine Fisheries, 3431 Arendell St., Morehead City, NC 28557. July 1996 before harvest treatments were applied. Oyster density was measured by counting (but not removing) oysters in three 0.25-m^ permanent plots established in each of the sixteen transects in each creek. Plots were established by stretching a measur- ing tape between the two PVC poles marking each transect and by placing a PVC quadrat at 0.5-, 2.5-, and 3.5-m distance. All live and dead oysters were counted in each quadrat. The density of naturally col- onized clams was not determined prior to the appli- cation of harvest treatments to avoid disturbing the reefs and potentially influencing the condition of the remaining clams and oysters. Instead, between 5 and 10 July, 16 hatchery-raised clams provided by ARC, Inc. of Atlantic, North Carolina, were placed in each of three 1-m- quadrats established within each 6-m transect. This introduction of transplanted clams was done to assure that enough clams were present for the experiment. The 1-m- quadrats were also placed at 0.5, 2.5, and 3.5 m distance along the transects. Before being transplanted, hatchery clams were dyed in Alizarin red dye in order to distinguish them from natural clam populations (Peterson et al., 1995). Of the 16 clams in each plot, eight were 20-25 mm in length, and eight were 27-32 mm in length. Between October 1996 and March 1997, oysters were harvested with hand tongs during low tides within the 1 ) oyster harvesting and 2 ) clam and oyster harvest- ing treatments. Oysters were harvested for the same total period of time (3.75-4.0 h/transect) along the entire length of each transect. From August 1996 to May 1997, clams were harvested during low tides with clam rakes and clam tongs from the 1 ) clam harvest- ing and 2) clam and oyster harvesting treatments, and for approximately the same total period of time (i.e. 3.75-4.0 h/transect). The total number of natu- rally occurring and transplanted clams and oysters removed during the harvest was recorded. All harvest- ing was conducted by R. A. Cummings, a professional shellfisherman. The density of live and dead clams and oysters remaining on experimental reefs was sampled 10-23 July 1997, after termination of the harvesting treat- ments. Clams and oysters were sampled several months after the last clam harvesting in May so that any poten- tial long-term effects of harvesting were realized. For example, unharvested clams and oysters remaining on reefs may have been injured during harvesting and died after several weeks. Oysters were sampled by plac- ing a measuring tape along the transects and counting all oysters within the three 0.25-m2 quadrats at 0.5-, 2.5-, and 3.5-m distance along each transect. Clams were sampled by digging up the top 25 cm of sediment from each 1-m- sampling plot. The sediment was then passed through a 1-mm sieve to remove all clams. Lenihan and Micheli: Biological effects of shellfish harvesting on oyster reefs 89 Table 1 Mean square errors (MS), F ratios, and corresponding significance levels (Pi of two-way fixed factor ANOVAs comparing densities of live and dead oysters, and proportions of dead oysters (per 0.25 m-) among intertidal oyster reefs before application of experimental harvest treatments (sampled 5-10 July 1996). The main factors in ANOVAs were creeks (Pages and Whiskey Creeks) and harvest treatment (clamming, oystering, both, and neither). Source df Live Dead Proportion ( lead MS F P MS F P MS F P Creek (C) 1 277.50 6.17 0.02 378.10 34.58 0.0001 0.03 30.11 0.0001 Harvest treatment (H ) 3 31.03 0.69 0.57 0.41 0.04 0.99 0.001 0.48 0.70 CxH 3 31.74 0.71 0.56 12.27 1.12 0.36 0.001 1.19 0.33 Residual 24 45.01 10.93 0.001 Statistical analyses The density of live and dead oysters, and the propor- tion of the total number of oysters that were found dead before harvesting, were compared among treat- ments by using two-way, fixed factor analysis of vari- ance (ANOVA) tests. The two main factors in the ANOVAs were creeks (Pages and Whiskey Creeks) and harvest treatments (clam harvesting, oyster har- vesting, both, or neither). The same ANOVA model was used to test for differences in 1 ) the density of live and dead oysters, and the proportion of dead oysters (i.e. number of dead oysters/live -i- dead oysters) after harvesting, 2 ) the density of live and dead, and propor- tion of dead, naturally occurring clams after harvest- ing, and 3) the density of live and dead, and number of missing transplanted, hatchery-raised juvenile clams after harvesting. Before all ANOVAs, homogeneity of variances was tested by using Cochran's test (at 0=0.05). When variances were heterogeneous, data were log transformed and homogeneity was retested. After ANOVAs, post hoc tests for differences among treatment means were conducted with Student-New- man-Keuls method (SNK) tests (at a=0.05). Results In July 1996, prior to the application of experimental harvests, the number of live and dead oysters (those observed with naked eye; >1 mm in length) and the proportion of dead oysters in experimental plots did not vary with the interaction of creeks and harvest treatment (ANOVA, creek x harvest treatment inter- action, P=0. 33-0.56; Table 1), nor among harvest treatments (P=0.57-0.99), but differed significantly between creeks (P=0.02-0.0001). Whiskey Creek had greater numbers of live and dead oysters and propor- tion of dead oysters than Pages Creek (Figs. 1 and 2). E in 80 70- 60 50 40 30- 20 10 80i 70 60 50 40 30 20 10 Pages Creek JL _Mi I, ^ i Whiskey Creek X n Live Dead u :MMl(iti Harvest type: O Before Oysters Oysters After Figure 1 Mean density of live and dead oysters (>1 mm in length) before (5-10 July 1996 ) and after ( 10-23 July 1997 ) applica- tion of experimental harvest treatments in Pages and Whis- key Creeks, NC. Data are means and one standard error (n=4) of counts taken within 0.25-m^ quadrats. Results of SNK post hoc comparisons are illustrated with letters above bars (a>b at P<0.05). Separate ANOVAs and SNK tests were used to compare numbers of live and dead oys- ters both before and after harvesting. Experimental clam harvesting conducted from Aug- ust 1996 to May 1997 removed only hard clams ft-om experimental plots (Table 2). In contrast, a few clams 90 Fishery Bulletin 98(1) were caught in oyster tongs during oyster harvest- ing, which was conducted from October 1996 to March E m i 100 90 80 70 60 50 40- 30 20- 10- Pages Creek o Q. 100 90 80 70- 60 50- 40 30 20- 10 Whiskey Creek ~r a a -r -r b T Harvest type: h o o o Before Clams + Oysters £ TO y; " c After 5 Clams •S, * >- Oysters Figure 2 Mean percentage of oysters found dead before ( 5-10 July 1996 ) and after ( 10-23 July 1997 ) application of experimen- tal harvest treatments in Pages and Whiskey Creeks. NC. Data are means and one standard error (n =4 ) of counts taken within 0.25-m'^ quadrats. Results of SNK pos< hoc compari- sons are illustrated with letters above bars (a>b at P<0.05). Separate ANOVAs and SNK tests were used to compare numbers of dead oysters before and after harvesting. Table 2 Mean number of clams and oysters removed from intertidal oyster reefs during experimental harvesting. Reefs were harvested for clams (clamming), oysters (oystering), both (clamming and oystering), or neither (controls). Transplanted, hatchery-raised clams were not removed during harvesting. Pages Creek Whiskey Creek Harvest treatments Clams Oysters Clams Controls Clammmg 3.47 ±1.1 11.77 ±7.37 Oystering 1.15 ±0.22 69.20 ±9.20 5.05+2.89 43.27 ±14.10 Clamming and oystering 3.46 ±0.75 89.40 ±58.32 12.59 ±5.84 34.97 ±8.26 1997. In both creeks, two to three times the number of clams were harvested during clam harvesting treat- ments than during oyster harvesting. Similar num- bers of clams were removed from reefs in the clam harvesting and the combined clam and oyster har- vesting treatments. Similar numbers of oysters were removed from plots harvested for oysters only and from those harvested for both oysters and clams (Table 2). According to visual observations, both types of har- vesting inflicted obvious wounds (holes and cracks) to the shells of oysters (range: 5-13 individuals within each plot) that were not removed by harvesting. In July 1997, after experimental clam and oyster harvesting, the density of live and dead oysters, and the proportion of dead oysters did not vary with the interaction of creeks and harvest treatment (AN OVA; creek x harvest treatment interaction, P=0.23-0.44; Table 3). There was also no significant difference in the density of live and dead oysters and the proportion of dead oysters between the two creeks (P=0. 16—0.65; Table 3). In contrast, there was a highly significant effect of harvest treatment on the density of live oysters and the proportion of oysters found dead (P=0.0001; Table 3). At both sites, plots harvested for clams, oys- ters, or both had 2-4.5 times lower densities of live oysters and 2-2.5 times higher proportions of dead oys- ters than did unharvested control plots (SNK, P<0.05 for both contrasts; Figs. 1 and 2). There were no dif- ferences in the number of dead oysters among harvest treatments. In July 1997, after experimental harvesting, the density of live, naturally occurring hard clams varied with the interaction of creeks and harvest treat- ments (ANOVA, creek x harvest treatment interac- tion, P=0.015; Table 4). At Pages Creek, there were greater numbers of live, naturally occurring clams in control reefs than in plots harvested for clams, oys- ters, or both (SNK; P<0.05; Fig. 3). At Whiskey Creek, there were more live, naturally occur- ring clams in both control and oyster- hai-vested plots than in plots harvested for clams and for both species (SNK, P<0.05 for both contrasts; Fig. 3). The number and proportion of dead, nat- urally occurring clams found in July 1997 did not vary with the interaction of creeks and harvest treatment (ANOVA, creek x harvest treatment interaction, P=0.09-0.87; Table 4), or between creeks (P=0.16-0.10; Table 4). There was also no significant effect of harvest treatment on the density of dead, nat- urally occurring clams (P=0.17; Table 4). However, there was a significant effect of harvest treatment on the pro- Oysters Lenihan and Michell; Biological effects of shiellfish harvesting on oyster reefs 91 Table 3 Mean square errors (MS), F ratios, and corresponding significance levels of (P) two-way fixed factor ANOVAs comparing densities of live and dead oysters, and proportions of dead oysters (per 0.25 m-) among intertidal oyster reefs afler application of experimental harvest treatments ( 10-23 July 19971. The main factors in ANOVAs were creeks (Pages and Whiskey Creeks) and harvest treatment (clamming, oystering, both, and neither). Live Dead Proportion dead Source df MS F P MS F P MS F P Creek (C) 1 18.45 Harvest treatment (H) 3 2192.00 C X H 3 132.90 Residual 24 86.82 0.21 0.65 55.52 25.25 0.0001 4.45 1.53 0.23 67.70 .54..54 1.02 0.32 0.08 0.97 1.24 0.32 0.02 0.12 0.01 0,01 2.10 13.86 0.93 0.16 0.0001 0.44 Table 4 Mean square erros ( MS I, F ratios, and corresponding significance levels (f ) of 2-way fixed factor ANOVAs comparing densities of live and dead hard clams, and proportions of dead clams (per 1.0 rn-^i among intertidal oyster reefs after application of harvest treatments ( 10-23 July 1997) The main factors in ANOVAs were creeks (Pages and Whiskey Creeks) and harvest treatment (clamming, oystering, both, and neither). Source Live Dead Proportion dead df MS F P MS F P MS F P Creek (C) 1 277.5 6.17 0.02 378.1 34.58 0.0001 0.03 30.11 0.0001 Creek (C) 1 0.13 0,03 0.85 10.7 2.15 0.16 0.07 2.94 0.10 Harvest treatment ( H ) 3 36.55 10.01 0.0002 8.97 1.80 0.17 0.16 6.19 0.003 CxH 3 15.60 4.27 0.015 12.28 2.47 0.09 0.01 0.23 0.87 Residual 24 3.65 4.98 0.03 portion of deaci, naturally occurring clams (P=0.003; Table 4). The proportion of dead clams in both creeks was much higher on harvested than on unharvested (i.e. control) reefs (SNK P<0.05 for both contrasts; Fig. 4) but was similar among the three harvest treatments (SNK, P>0.05 for both contrasts; Fig. 4). After harvesting, the density of live and dead hatch- ery-raised clams transplanted to reefs at the beginning of the experiment tended to vary with the interaction of creeks and harvest treatment, although not signif- icantly (ANOVA, creek x harvest treatment interac- tion, P=0.07-0.08; Table 5 ). However, the density of live transplanted clams varied between creeks (P=0.03; Table 5) and among harvest treatments (P=0.04; Table 5). More transplanted clams were recovered alive in Pages Creek (mean ± ISD: 3.21 ±1.62/m2) than in Whis- key Creek (2.22 ±1.45/m''). Fewer live transplanted clams were recovered from clam-harvested plots than from control plots in both creeks ( SNK, P the interac- tion of <0.05; Fig. 5). The number of dead transplanted clams found after harvesting also varied between creeks (Pages Creek>Whiskey Creek; P=0.0001; Table 5) but did not vary significantly with harvest treat- ment (P=0.10; Table 5). At Pages Creek, there was a slight trend for greater mortality of transplanted clams on clam-harvested and clam- and oyster-har- vested plots than in oyster-harvested and control plots only (Fig. 5). Most transplanted clams placed on reefs at the beginning of the experiment were not found at the end of the experiment ("missing" clams; Fig. 5). The number of missing transplanted clams differed with the interaction of creeks and harvest treatment (ANOVA, creek x harvest treatment interaction, P=0.03; Table 5) because fewer clams were recovered in our census in the oyster-harvested plots than in clam-harvested plots at Whiskey Creek only (SNK; P<0.05; Fig. 5). Discussion Our results clearly demonstrate that both clam and oyster harvesting significantly reduce oyster popula- 92 Fishery Bulletin 98(1) E E O Whiskey Creek Harvest type Figure 3 Mean density of live and dead naturally-occurring hard clams found after ( 10-23 July 1997 1 application of experi- mental harvest treatments in Pages and Whiskey Creeks, NC. Data are means and one standard error (^=4) of counts taken within 1.0-m'^ quadrats. Results of SNK post hoc comparisons are illustrated with letters above bars (a>b at P<0.05). Separate ANOVAs and SNK tests were used to compare numbers of live and dead clams. E o 100 90 80 70 60 50- 40- 30 20 10 en Q. Pages Creek lOOi 90 SO- TO 60- 50 40 30- 20 10- Whiskey Creek X o 2 i;? Clams ^ tl (U + o 5 g. Oysters Harvest type Rgure 4 Mean percentage of naturally-occurring hard clams found dead after (10-23 July 1997) application of experimental harvest treatments in Pages and Whiskey Creeks, NC. Data are means and one standard error (?!=4) of counts taken within 1.0-m'-^ quadrats. Results o{ SNK post hoc comparisons are illustrated with letters above bars (a>b at P<0.05 ). Table 5 Mean square errors (MS), F ratios and corresponding significance levels b at P<0.05). Separate ANOVAs and SNK tests were used to compare numbers of live, dead, and missing clams. alone also reduced the density of live clams but only at one site. Pages Creek. At Whiskey Creek, the density of live clams after harvesting was similar between oys- ter-harvested and control plots, indicating that oyster harvesting had little effect on clam survival (Fig. 3). A negative effect of oyster harvesting on clams may be caused both by direct removal of clams as bycatch (Table 2) and enhanced clam mortality through mecha- nisms analogous to those hypothesized for oysters (see above). Some clams may also have emigrated from the oyster harvesting treatments following harvesting. Patterns of survival and mortality of hatchery- raised clams transplanted to experimental reefs varied with site and harvest type (Table 5). Fewer live and dead transplanted clams were recovered from reefs at Whiskey Creek than at Pages Creek (Fig. 5). In con- 94 Fishery Bulletin 98(1) trast, there were greater numbers of missing trans- planted clams at Pages Creek than at Whiskey Creek. Harvest type, specifically clam harvesting, influenced the number of live transplanted clams but had no sig- nificant effect on the number of dead or missing trans- planted clams. Fewer live, transplanted clams were found in clam-harvested plots than were found in con- trol plots at both sites (Fig. 5). Because few of the transplanted clams were removed from reefs by exper- imental harvesting, the negative effects of clam har- vesting on densities of live, transplanted clams may be explained by increased clam mortality caused by clam harvesting. Overall, the effects of shellfish har- vesting appear to be more variable and unpredictable for clams than for oysters. Our results indicate that both types of shellfish harvesting can have negative impacts on clam populations, but that this is a site- specific phenomenon. PJesults of this study do not support the hypothesis that harvesting reefs for both clams and oysters has a negative synergistic impact on clam and oyster popu- lations. Clam and oyster harvesting alone had similar negative effects on densities of live oysters, and the joint harvesting of both species on the same reefs did not decrease the density of live oysters any fuither. Simi- larly, the negative effects of clam harvesting on the den- sity of live clams, and on survival of hatchery-raised clams were not enhanced when oyster harvesting was applied in combination wdth clam harvesting. Thus, the combined harvesting of both clams and oysters on inter- tidal reefs does not cause greater direct or indirect mor- taUty of shellfish populations than that caused by clam or oyster harvesting conducted separately. This experimental analysis has important implica- tions for the management of intertidal oyster reefs and their associated molluscan fishery resources. First, maintaining high densities of oysters on some inter- tidal reefs, by preventing both clam and oyster har- vesting, may help to preserve future oyster harvests and brood stock. Protecting some reefs from shellfish- ing will also help preserve the many ecological ser- vices that oysters and oyster reefs provide, such as improving water quality through the filtration of sus- pended particles (Officer et al., 1982; Dame et al., 1984; Newell, 1988) and creating essential recruit- ment, refuge, and foraging habitat for economically valuable fishes and crabs (Bahr and Lanier, 1981; Zimmerman et al., 1989; Lenihan et al., 1998). Pre- venting oyster and clam harvesting on some intertidal reefs will also potentially conserve clam populations from both the direct and indirect negative effects of shellfish harvesting, thereby protecting future clam harvests and brood stock. Overall, allowing the har- vest of both clams and oysters on some natural and restored oyster reefs is a rational option because the combined effect of both clam and oyster harvesting is no greater than the effect of each harvesting activity conducted alone. Thus, we recommend that both types of harvesting be allowed on some reefs but that other reefs be protected as refuges for shellfish populations and other reef-associated fauna. In adaptive fishery and habitat management, the results of relatively small-scale, prototype experi- ments, like the one reported here, are used to design larger-scale comparisons of potential management options. Therefore, we recommend that the results of our experiment be used to design alternative shell- fishery management options that can be implemented and monitored on relatively large spatial and tempo- ral scales in North Carolina and other coastal states of North America. Our recommendation that some natu- ral and restored oyster reefs be closed from shellfish harvesting and others opened or restored for the pur- pose of both clam and oyster harvesting can be used to identify potential management options. Further test- ing of the generality of our findings on larger spatial and temporal scales is necessary because our study was conducted at only two sites and over a one-year period. Therefore, our results may not apply to areas with different environmental conditions (e.g. different flow and sedimentary regimes, areas of low recruit- ment) and hai-vesting intensities (e.g. very low and high levels of hai-vesting). It is necessary to determine with experiments and simulation models how much oyster reef habitat should be preserved from harvesting to maintain sustainable oyster and clam brood stock populations and habaitat for the successful recruitment and sunaval of other fishery organisms. The following steps should now be taken by fishery and habitat managers to improve management of the clam and oyster populations and intertidal oyster reef habitat: 1 ) identify overall management goals and pos- sible options; 2) derive specific predictions based, at least in part, on the experiment results reported in this study; and 3) design monitoring programs to quantify the effect of each management option. Whenever pos- sible, it is highly recommended that fishermen, fish- ery managers, and ecologists be included in designing and monitoring large-scale management experiments because collectively they will provide the highest level of rigor and reality. Acknowledgments We thank R. A. Cummings for helping to design and conducted the experimental harvests. We thank M. Marshall, D. Meyers, C. H. Peterson, H. C. Summer- son, and G. W. Thayer for advice; R. Carpenter for helping to select experimental sites; and D. Bockus, Lenlhan and Michell: Biological effects of sfiellfisfi harvesting on oyster reefs 95 L. Hill, C. Ingram, and C. Lund for help in the field. We thank J. B. Pearce, J. Merriner, and three cinonymous reviewers for improving the manuscript. This work was funded by a Fishery Resource Grant awarded to F. Micheli, H. S. Lenihan, and R. A. Cummings by the North Carolina Marine Fisheries Commission. H. S. Lenihan also received support from a National Research Council Postdoctoral Associateship. Literature cited Arnold, W. S. 1984. The effects of prey size, predator size, and sediment composition on the rate of predation of the blue crab, Calli- nectes sapidus Rathbun, on the hard clam, Mercenaria mer- cenaria (Linne'). J. Exp. Mar Biol. Ecol. 80:207-219. Bahr, L. M. and W. P. Lanier. 1981. The ecology of intertidal oyster reefs of the South Atlan- tic Coast: a community profile. U.S. Fish and Wildlife Ser- vice. Biological Sendees Program FWS/OBS/81/1.5. 63 p. Botsford, L. W., J. C. CastiUa, and C. H. Peterson. 1997. The management of fisheries and marine ecosystems. Science (Wash. D.C.) 277:509-514. Breitburg, D. L, M. A. Palmer, and T. Loher. 1995. Larval distributions and the spatial patterns of settle- ment of an oyster reef fish: response to flow and structure. Mar Ecol. Prog. Ser 125:45-64. Dame R. F., R. G. Zingmark, and E. Haskin. 1984. Oyster reefs as processors of estuarine materials. J. Exp. Mar Biol. Ecol. 83:239-247. Dayton P. K., S. F. Thrush, M. T. Agardy, and R. J. Hofman. 1995. Environmental effects of marine fishing. Aquat. Cons. 5:205-232. Engel, J., and R. G. Kvitek. 1995. Effects of otter trawling on a benthic community in Monterey Bay National Marine Sanctuary. Conserv. Biol- ogy 12:1204-1214. FAO (Food and Agriculture Organization of the United Nations). 1993. Annual report. Food and Agriculture Organization, Rome, Italy 321 p. Grumbine, R.E. 1997. Reflections on "What is ecosystem management?" Con- serv. Biol. 11:41-47. Lenihan, H. S. 1999. Physical-biological coupling on oyster reefs: how habitat structure influences individual production. Ecol. Monogr. 69:251-276. Lenihan, H. S., J. H. Grabowski, and G. W. Thayer. 1998. Recruitment to and utilization of oyster reef habitat by fishes and blue crabs. National Research Council, National Marine Fisheries Service, Beaufort, NC. Tech. Rep., 130 p. Lenihan, H. S., F. Micheli, S. W. Shelton, and C. H. Peterson. 1999. The influence of multiple environmental stressors on susceptibility to parasites: an experimental determination with oysters. Limnol. Oceanogr. 910-924. Lenihan, H. S., and C. H. Peterson. 1998. How habitat degradation through fishery disturbance enhances impacts of hypoxia on oyster reefs. Ecol. Appl. 8:128-140. Luckenbach, M. A., R. Mann, and J. E. Wesson (eds). 1998. Oyster reef restoration: a symposium and synthesis of approaches. Virginia Institute of Marine Sciences Press, Gloucester Pt., VA, 266 p. McAllister, M. K., and R. M. Peterman. 1992. Experimental design in the management of fisheries: a review. N. Am. J. Fish. Manage. 12:1-18. Michelli, F., and C. H. Peterson. 1999. Estuarine vegetated habitats as corridors for predator movements. Conserv. Biol. 13:869-881. NRG (National Research Council). 1995. Understanding marine biodiversity: a research agenda for the nation. National Academy Press, Washington, D.C, 203 p. NeweU, R. I. E. 1988. Ecological changes in Chesapeake Bay: Are they the result of overharvesting the American oyster, Crassostrea vir- ginica? In Understanding the estuary: advances in Chesa- peake Bay research: proceedings of a conference, p 536-546. Chesapeake Research Consortium Publication 129, CBP/TRS 24/88. Baltimore, MD. Norse, E.A.(ed.). 1993. Global marine biological diversity strategy. Center for Marine Conservation, Washington, D.C, 321 p. Officer C. B., T. J. Smaya, and R. Mann. 1982. Benthic filter feeding: a natural eutrophication control. Mar Ecol. Prog. Sen 9:203-210. Peterson, C. H., H. C. Sununerson, and S. R. Fegley. 1987. Ecological consequences of mechanical harvesting of clams. Fish. Bull. 85:281-298. Peterson, C. H., H. C. Summerson, and J. Huber. 1995. Replenishment of hard clam stocks using hatchery seed: importance of bottom type, seed size, planting season, and density. J. Shell. Res. 14:293-300. Rothschild B. J., J. S. Ault, P. GouUetquer, and M. Heral. 1994. Decline of the Chesapeake Bay oyster population: a century of habitat destruction and overfishing. Mar Ecol. Prog. Ser 111:29-39. Russ, G., and Alcala, R. 1996. Marine reserves: rates and patterns of recovery and decline of large predatory fish. Ecol. Appl. 6:947-961. Sponaugle, S., and P. Lawton. 1990. Portunid crab predation on juvenile hard clams: effects of substrate type and prey density. Mar Ecol. Prog. Ser. 67:43-53. Underwood, A. J. 1990. Experiments in ecology and management: their logic, functions and interpretation. Austr J. Ecol. 15:365-389. Walters, C. J. 1986. Adaptive management of renewable resources. Mac- Millan Press, New York, NY, 415 p. West, R. J., and G. N. G. Gordon. 1994. Commercial and recreational harvest of fish from two Australian coastal rivers. Austr. J. Mar Fish. Res. 45:1259-1279. Winslow, F. 1881a. Deterioration of American oyster beds. Pop. Sci. Mon. 20:29-43. 1881b. Deterioration of American oyster beds. Pop. Sci. Mon. 20:45-1.56. Zimmerman, R., T. J. Minello, T. Baumer, and M. Castiglione. 1989. Oyster reefas habitat for estuarine macrofauna. NOAA Tech. Mem.NMFS-SEFC-249, 217 p. 96 Abstract.-In 1996 we surveyed the fishes living on and around seven olT- shore oil platforms in the Santa Bar- bara Channel area. We conducted belt transects at various depths in the mid- water and around the bottoms of each platform using the research submers- ible Delta. The bottom depths of these platforms ranged from 49 to 224 m and the midwater beams ranged from 21 to 196 m. We found that there were several distinct differences in the fish assemblages living in the midwater and bottom habitats around all of the platforms. Both midwater and bottom assemblages were dominated by rock- fishes. Platform midwaters were dom- inated by young-of-the-year (YOY) or juveniles up to two years old. Rockfishes larger than about 18 cm total length were rarely seen in the midwater. The fish assemblages around the bottoms of the platforms were dominated by larger individuals, primarily subadults or adults. Density of all fishes was sim- ilar between the bottoms and midwa- ter of any given platform. However, the total biomass was much greater on the bottoms, owing to larger fish living there. There was a consistently greater number of species on the bottom than in the midwater of each platform, likely because of a larger variety of habitat types on the bottom. The fish assem- blages also differed among platforms. We found significantly higher densities of young-of-the-year rockfishes around platforms north of Pt. Conception com- pared with those in the Santa Barbara Channel, probably because the more northerly platforms are located in the more productive waters of the Califor- nia Current. Fish assemblages around seven oil platforms in the Santa Barbara Channel area Milton S. Love ^: Jennifer E. Caselle Linda Snook Marine Science Institute University ol California Santa Barbara, California 93106 E-mail address (for M S Love) loveio'lifesci.ucsb edu Manuscript accepted 16 August 1999. Fish. Bull. 98:96-117 (2000), Petroleum production has been a part of the southern Cahfornia econ- omy since the nineteenth century. The earhest drilUng took place on land, but by the early twentieth cen- tury a large number of piers lined the coast, tapping into offshore oil deposits. Hazel, the first offshore oil platform, was constructed off Summerland in 1958 (Carlisle et al., 1964). At the peak of oil drill- ing in the early 1980s, there were 30 platforms operating in southern and central California. Currently, there are 19 platforms in operation in the Santa Barbara Channel and off Point Conception (Fig. 1). Oil platforms provide considerable habitat for marine organisms. The earliest structures were relatively small (23 m long at the surface), newer platforms, however, are over 100 m long (MBCi). Sessile inverte- brates (primarily mussels, barnacles and anemones) encrust the pilings and well pipes and cover the bottom to form additional habitat. Oil platforms have a finite eco- nomic lifespan and a number of them are becoming uneconomical to operate. In 1996, four platforms were removed from the Santa Bar- bara Channel, although not without controversy. There is considerable debate regarding the fate of these structures. Some interest groups would like to leave part or all of them in place, claiming protection of fish habitat; others favor complete removal. Understanding the biologi- ical communities on the platforms is crucial to making rational deci- sions regarding the fates of these structures. In addition, research on these platforms could also address questions regarding the role that artificial reefs might play in coastal fish communities. Ultimately, this research will allow us to contrast the fish assemblages on platforms with those of nearby reefs. Currently, very little is known about the fish fauna around these platforms. One relatively compre- hensive SCUBA survey examined fish populations around two shallow inshore platforms. Hazel and Hilda, during Hazel's first three years and Hilda's first year of operation (Car- lisle et al., 1964), Additional cur- sory surveys were conducted around these two platforms in 1970 and 1975; Bascom et al., 1976; Allen and Moore'^). With the exception of a short-term study of fishes around platform Hidalgo using a remotely operated vehicle (ROV) (Love et al., 1994) and a survey of recre- ational fishing around Santa Bar- • MBC( Marine Biology Consultants). 1987. Ecology of oil/gas platforms offshore Cal- ifornia. Outer Continental Shelf (OCS) Studv Minerals Management Service (MMS) 86-0094. - Allen, M. J, and M.D.Moore. 1976. Fauna of offshore structures. South. Calif Coast. Water. Res. Proj. Annu. Rep., Long Beach, CA, p. 179-186. Love et al : Fish assemblages around oil platforms in the Santa Barbara Channel area 97 Irene .\ pt Argi"" 72wyf Hidalgo-^ 130m .^ Harvest Hermosa 176m 182m icep'""' Figure 1 Locations of 19 oil platforms in the Santa Barbara Channel and off Pt. Conception. The seven platforms sur- veyed in this paper are denoted with stars and labeled. bara Channel platforms (Love and Westphal, 1990), no other research has been published on the fishes of any California oil platform. In 1995, we began a survey of the fishes living on and around several platforms in the Santa Bar- bara Channel area. The surveys were of two t3^es: a scuba-based study in the surface waters (to 30 m) of the platforms and a submersible survey that examined the deeper sections of these structures. However, in 1995, we could not survey any platform bottoms because of inclement weather. This paper discusses the results of the 1996 deep survey. Channel are typically cool because the California Current flows equatorward from high latitudes year- round and upwells in the Point Conception and Point Arguello areas during spring and summer. At the same time, the cyclonic circulation pattern in the southern California bight brings warm water flow- ing poleward along the coast from the east and south of the Santa Barbara Channel. In general, water is cooler and more productive in the area of Points Arguello and Conception than in the Santa Barbara Channel, particularly compared with the more east- ern end of the channel. Materials and methods Study sites We surveyed fish assemblages around oil platforms situated in and just northwest of the Santa Barbara Channel. Surveys were conducted around the bottom of six platforms and in the midwater of seven plat- forms in 1996 (Fig. 1; Table 1). The bottom depth of these platforms ranged from 49 to 224 m. The mid- water depths ranged from 21 to 196 m. The platforms are situated in an area with a com- plex oceanographic regime. The Santa Barbara Chan- nel is semi-enclosed, faces east-west, and is bordered by the Northern Channel Islands on the south and the mainland on the west. It is embedded within the much larger California-Baja California coastal current regime (Brink and Muench, 1986; Hickey, 1992). Surface waters to the north and west of the Surveys Using the submersible Delta, we conducted belt tran- sects around each platform. The submersible main- tained a speed of approximately 0.5 knots and stayed approximately 2 m from the structure. Transects were made around the bottom of the platform (from the substrata to approximately 2 m above the sub- strata) and around each set of cross beams to a mini- mum depth of about 20 m below the surface. Dives were conducted during daylight hours, between one hour after sunrise and two hours before sunset. During the transects, researchers made their observations from the central starboard-side viewing port. An externally mounted Hi 8-mm video camera with associated lights filmed the same viewing field as seen by the observers. Observers identified and counted all fishes and verbally recorded those data on the video. All fishes within 2 m of the submarine were counted. Fish lengths were estimated by using 98 Fishery Bulletin 98(1) Table 1 Latitude, longitude, to southeast. bottom and midwater depths and date of sampling of the seven oil platforms. Platforms li sted from northwest Platform Location Bottom depth (m) Midwater depths (m) Date surveyed Irene 34''36.62'N, 120°4.3.77'W 72 29,50 2 Nov 1996 Hidalgo 34°29.70'N, 120°42.13'W 130 36. 59, 83, 107 28 Oct 1996 Harvest' 34°28.15'N, 120°40.85'W 176 38,61,84, 113, 141 28 Oct 1996 Hermosa 34°27.33'N, 120°38.78'W 182 63. 84, 106, 131, 1.56 28 Oct 1996 Holly 34°23.38'N, 120°54.33'W 49 21. 35. .50 29 Oct 1996 Grace 34''10.77'N, 120°28.12'W 97 45. 69. 82 30 Oct 1996 Gail 34°07.50'N, 120»24.02'W 224 71.95, 115, 141, 166, 196 30 Oct 1996 ' Midwater survey on] V, a pair of lasers mounted on either side of the exter- nal video camera. The projected reference spots were 20 cm apart and were visible to the observer. An environmental monitoring system aboard the sub- marine continuously recorded date and time, depth, and altitude of the vessel above the sea floor. After the dive, these data were overlaid on the original videotape. Transect videos were reviewed either aboard the research vessel or in the laboratory. For each fish, we recorded 1) its species, to lowest identifiable taxa; 2) its estimated total length to the nearest cm; and 3) the microhabitat it occupied (e.g. pipe, sand, mussel shell mounds, mud). We defined young-of-year fishes (YOYs) from published estimates of size at age. Sub- adults are defined as juveniles in their second year up to, but not including, maturity. During the survey at platform Gail, all greenspot- ted iSebastes chlorostictus) and greenblotched rock- fishes (S. rosenblatti) were inadvertently identified as greenspotted rockfish. In reviewing the video- tape, it was clear that some of the individuals that were recorded as greenspotted rockfish were in fact greenblotched rockfish. In order to correct for this potential misidentification, the total number of both species was adjusted by using the proportion of greenblotched to greenspotted rockfishes (ratio=2.2) obsei"ved at platform Gail during the following year's survey (Love, unpub. data). Similar numbers of the two species combined were observed during the two years (1996: /z = 186,1997: «=209). Analyses We estimated length of those transects conducted on the bottom by first determining the submersible speed. This was done by evaluating a ten-second seg- ment for every one minute of transect. The video was manually forwarded frame by frame and the number of 20-cm segments passing the lasers in a ten-second section was counted. The number of 20 cm segments per 10 seconds was divided by 2 to obtain speed in centimeters per second. All subsamples were then averaged to obtain mean transect speed (cm/s). The mean speed was then multiplied by the number of seconds in the transect and divided by 100 to obtain transect length in meters. The length was then mul- tiplied by 2 m (the transect width) to obtain transect area, allowing us to present both densities (fish/m^l and biomass (kg/m^). Biomass was estimated for all species by using length-weight relationships derived empirically or obtained from the literature. No bio- mass estimates were made for species that could not be identified to the family level. In the midwater, we could not see the lasers pass before fixed points; therefore, we could not directly measure the length of the midwater transects. With- out knowledge of the length of the midwater tran- sects, we could not calculate density or biomass per unit area as done on the bottom transects. However, we were able to estimate the length of midwater transects for use in estimating both fish density and biomass. We did this by converting density and bio- mass on the midwater transects from number and kilogram per minute to number and kilogram per m^, respectively. This conversion was accomplished by calculating the equation for the regression of den- sity in terms of number per m- on density in terms of number per minute for the bottom transects where both values were known (Fig. 2A). The same rela- tionship was calculated for biomass (Fig. 2B). Given the regression equations, density per m^ and bio- mass per m- could be calculated from number per minute and kilogi'ams per minute. We called these Love et a\ Fish assemblages around oil platforms in the Santa Barbara Channel area 99 calculations "estimated density" (number/m^) and "estimated biomass "{kg/m-). This method of estimating transect length and hence fish density and biomass relies on the assump- tion that the submersible travels at the same speed in both habitats. Although we did not have data on submersible speed, every attempt was made to main- tain the submersible at the same speed on all tran- sects during the survey. However, because of debris on the bottom and water currents in the midwater, if there were differences in speed, the submersible was likely to travel slightly faster in the midwater habitats than on the bottoms. In this case, the sub- mersible would cover more area per unit time and the true fish density in the midwater may actually be slightly lower than our estimated density. We consider the potential bias introduced by differences in submersible speed to be minor in relation to the magnitude of the observed differences in fish densi- ties between the midwater and bottom transects (see "Results" section). We calculated both species richness (number of species) and species diversity. We used the Shan- non-Weiner diversity index (H') for all species diver- sity comparisons (Shannon and Weaver, 1949). We also calculated a percent similarity index (PSD that quantifies how similar two assemblages are in terms of their species composition (i.e. the relative abun- dance of those species). The index ranges from (no species shared) to 100'7f (identical composition and relative abundances). The formula for PSI is PS/ = {^min(p„,pj)x 100. where, p^, and p^.^ are the proportion of the ith spe- cies in habitat x and habitat y. PSI was calculated for each pair of platform bottom assemblages. Results Bottom versus midwater transects We found that there were several distinct differ- ences in the fish assemblages living in the midwater and bottom habitats around all of the platforms. We calculated percent similarity indices (PSI) between the bottom and midwater assemblages for each plat- form. These PSIs ranged from 19( to 349c (mean 13.3%). Although both midwater and bottom assem- blages were dominated by rockfishes, platform mid- waters were dominated by young-of-the-year (YOY) or slightly older juveniles (<10cm). Rockfishes larger than about 20 cm were rarely seen in the midwater cm E o Density (number/minute) 20-1 Biomass (kg/minute) Figure 2 Regressions of (A) density in terms of number per minute on density in terms of number per m- and (B) biomass in terms of kilograms per minute on bio- mass in terms of kilogram per m- for bottom tran- sects on all platforms. The regression equations were used to calculate density and biomass from number and kilogram per minute into number and kilogram per ni'^ on the midwater transects. See "Methods" section for explanation of the conversion. (Fig. 3). The fish assemblages around the bottoms of the platforms were dominated by subadults or adults (11-20 cm) and occasionally harbored very large individuals (up to 48 cm) (Fig. 3). Average density per platform of all fishes com- bined was not significantly different on the bottom versus the midwater transects (bottom mean density 100 Fishery Bulletin 98(1) E s Z 1,500-1 1,000- 500- IZI Midwater Bottom 18 23 28 33 38 Midpoint length (cm) — t- 43 48 — 1 53 Figure 3 Length-frequency distribution for all roclifish species and all platforms combined. Midpoint length is the midpoint of .5-cm length bins. (SE)=141.4 fish/lOOm'- (49.0), n=6 platforms; mid- water mean density (SE)= 115.8 fish/lOOm^ (32.2), «=7 platforms; Mest, /=0.44, P=0.66). On three plat- forms, density was higher on the bottom than in mid- water and on three other platforms the reverse was true. However, there was a much larger and consis- tent difference in biomass between bottom and mid- water transects. For most families and all platforms, total biomass was higher on bottom than midwater transects (Table 2). Average biomass per platform (SE) for all species combined was 19.06 kg/m^' (2.5) on the bottom and 6.47 kg/m- (2.3) in the midwater (^test, ;=3.75, P=0.003). This consistent difference was due to the lack of adult fishes in the midwater. Fewer species lived on the midwater structures than on the bottom. Species richness for all rigs combined was 24 in the midwater versus 40 on the bottom. Average species richness per platform was significantly higher on the bottom than in the midwater (bottom mean richness (SE)=14.7 species ( 1.5); midwater mean richness (SE)=8.2 species ( 1.4); ^-test, ^=3.26, P=0.008). Average species diversity (H') across platforms was identical between bottoms and midwaters (bottom mean H' (SE)=:1.2 (0.2); mid- water mean H' (SE)=1.2 (0.2); ^test, ^=0.09. P=0.99). We present the remaining results for bottom and midwater habitats separately. Bottom habitat All platforms We identified at least 40 fish species around the platform bottoms (Table 2). Twenty-seven species were rockfishes; they were by far the most speciose group. Rockfishes made up 92. 7*^ of all fishes on the bottom (Table 3) and represented 96.7% of the biomass (Table 2). Halfbanded, greenspotted, copper, vermilion, widow, and fiag rockfishes, and bocaccio were among the most commonly observed rockfishes (Table 3). Our observations indicated that vermilion I'ockfish, flag rockfish, and bocaccio of all sizes were always closely associated with the platform structure (Fig. 4A). Larger copper and greenspotted rockfishes also were more likely to be very close to the platform. In par- ticular, flag rockfish were most often seen tucked well into the space formed by the bottom of the lowest crossbeam and the bottom (Fig. 4B). Flag and greenspotted rockfishes were almost always seen on or very close to the bottom. Halfbanded rockfish, as well as smaller greenspotted and copper rockfishes, were less bound to the platform and were often seen well away from the structure. Juvenile greenspotted and copper rockfishes were usually nestled within or just above the mussel, shell-covered substrata. Vermilion rockfish, and to a certain extent copper rockfish and bocaccio, would occasionally ascend up platform legs as much as 5 m. Flag rockfish, as well as larger bocaccio and ver- milion rockfish, often were solitary or found in small groups. The exception occurred at platform Gail, where one school of bocaccio comprised at least 100 individuals. Smaller adult or subadult vermilion and copper rockfishes tended to aggi'egate, often in mixed groups containing 50 or more individuals. The few Love et al,: Fish assemblages around oil platforms in the Santa Barbara Channel area 101 Table 2 Biomasses of fishes observed, by species, around all platforms in October-November of 1996. Biomasses platforms and midwater, with percent of totals for those parts of the platforms. Biomass is kilograms/m- in boldface. YOY means "young-of-year." are given for bottoms of Family totals are given Family Common name Scientific name Bottom Midwater Biomass % Total Biomass % Total Scorpaenidae Rockfishes 96.83 84.66 38.81 85.90 Kelp rockfish Sebastes atrovirens 0.34 0.75 Brown rockfish S. auriculatus 0.82 0.71 Gopher rockfish S. caniatus 0.01 <0.1 0.23 0.51 Copper rockfish S. caurinus 12.12 10.59 0.47 1.04 Greenspotted rockfish S. chlorostictus 8.80 10.94 0.56 1.23 Starry rockfish S. constellatus 0.02 <0.1 Darkblotched rockfish S. crameri 0.07 <0.1 Calico rockfish S. dalli 1.40 1.22 0.08 0.18 Greenstriped rockfish Swordspine rockfish S. elongatus S. ensifer 0.32 0.03 0.28 <0.1 0.08 0.17 Widow rockfish S. entomelas 1.86 1.62 24.08 53.15 Yellowtail rockfish S. flavidus 0.08 <0.1 Chilipepper S. goodei 0.82 1.82 Squarespot rockfish S. hopkinsi 0.46 0.40 0.51 1.12 Vermilion rockfish S. miniatus 20.84 18.22 Blue rockfish S. mystinus 0.74 0.65 0.29 0.64 Bocaccio rockfish S. paucispinis 14.35 12.55 3.68 8.12 Canary rockfish Rosy rockfish S. pinniger S. rosaceus 1.18 0.46 1.03 0.40 0.07 0.15 Greenblotched rockfish S. rosenblatti 3.89 0.15 Yelloweye rockfish S. ruberrimus 0.06 <0.1 Flag rockfish S. rubrivinctus 1.44 1.26 0.55 1.21 Bank rockfish S. rufus 0.29 0.25 Halfbanded rockfish S. semicmctus 26.21 22.91 0.04 <0.1 Olive rockfish S. serranoides 0.03 <0.1 Treefish Pygmy rockfish S. serriceps S. wilsoni 0.19 0.04 0.17 <0.1 0.04 <0.1 Sharpchin rockfish S. zacentrus 0.10 <0.1 0.04 <0.1 Shortspine thornyhead Sebastolobus alascanus <0.] <0.1 Sebastomus group' 0.31 0.27 0.37 0.82 Rockfish YOY Sebastes spp. 0.72 0.63 6.56 14.47 Hexagrammidae Greenlings 12.88 11.25 2.72 6.01 Kelp greenling Lingcod Hexagrammos decagrammus Ophiodon elongatus 12.35 10.80 <0.1 <0.1 Painted greenling Shortspine combfish Combfish sp. Oxylebius pictus Zanwlepis frenata Zamolepis sp. 0.51 0.01 0.01 0.44 <0.1 <0.1 2.72 6.01 Pomacentridae Damselfishes 0.54 1.20 Blacksmith Chromis punctipinnis 0.54 1.20 Embiotocidae Seaperches 4.57 3.99 3.02 6.65 Pile perch Damalichthys vacca 3.71 3.24 1.18 2.60 Sharpnose surfperch Phanerodon atripes 0.49 0.43 1.84 4.05 Unident. sea perches 0.20 0.17 Pink surfperch Zalembius rosaceus 0.17 0.15 Gadidae Cods 0.20 0.44 Pacific hake Merhtccius productus 0.20 0.44 continued 102 Fishery Bulletin 98(1) Table 2 (continued) Family Common name Scientific name Bottom Midwater Bioniass % Total Biomass % Total Cottidae Bathymasteridae Agonidae Flatfish Sculpins Unidentifed sculpin Ronquils Unidentified ronquil Poachers Unidentified poacher Flatfish Sanddabs Citharichthys sp. Unident. flatfish 0.03 0.03 0.01 0.01 0.06 0.01 0.05 0.03 0.03 0.01 0.01 0.06 0.01 0.05 0.03 0.03 0.07 0.07 ' Sebastomus group may include greenblotched. greenspotted, pinkrose, rosethorn, rosy. starry, or swordspine rockfishes. canary rockfish we noted tended to associate with vermilion rockfish. Halfbanded rockfish were almost always seen in schools, sometimes containing hun- dreds of individuals (Fig. 4C). In the greenling family, Hexagrammidae, both ling- cod and painted greenling, were common; together they represented about 5.3% of all fishes seen. Larger lingcod were solitary and tended to remain near the bottom of the platform (Fig. 4D). They were usu- ally seen sitting motionless on the bottom or slowly swimming just above it. Juvenile lingcod rarely came within a meter of the platform, they were usually seen Ijdng among the mussel shells away from the structure (Fig. 4E). Painted greenling sat on the crossbeams, along the pilings and on the mussel shells, always found as solitary individuals. Among platform comparisons The bottom fish as- semblages around each platform were all different (Tables 4 and 5). Pairwise percent similarity indices (PSD for each combination of platforms ranged from OVc (platforms Gail and Holly) to 70.1% (platforms Grace and Hidalgo) (Table 4). The average percent similarity was 20.0%. Despite a low average sim- ilarity value, rockfishes, as measured by number, density and biomass, dominated the bottom assem- blages around all of the platforms (Table 5). Ling- cod were the only nonrockfish species among the top four most common species at any platform. Around platform Irene, subadult and adult copper and vermilion rockfishes were most abundant. Irene also was unique among the platforms in having large numbers of juvenile lingcod. Halfbanded rockfish, painted greenling, and pile perch were also com- monly seen. Halfbanded and greenspotted rockfish were most common at platform Hidalgo, along with flag rockfish, lingcod, bocaccio, and vermilion rock- fish. Similar to that around Hidalgo (PSI=60%), the bottom fish assemblage around platform Hermosa was characterized by greenspotted and halfbanded rockfish, with lesser numbers of flag rockfish and lingcod (Table 4). Vermilion, calico, widow, copper, and squarespot rockfishes were most often seen at Holly, along with lesser numbers of halfbanded rock- fish, pile perch, rosy rockfish, and painted green- ling. Very large schools of halfbanded rockfish were observed at Grace, along with some flag, greenspot- ted, and vermilion rockfishes. The dominance of half- banded rockfish at Hidalgo and Grace resulted in the highest PSI among platform pairs (70.1% ). Mem- bers of the rockfish subgenus Sebastomus, primar- ily gi'eenblotched and greenspotted rockfishes and bocaccio were most abundant at platform Gail. Gail had by far the highest number and density of bocac- cio of any of the platforms. We observed between 8 and 21 species around the bottom of the platforms (Fig. 5A). We found no significant relationship between species number or diversity (H') and platform bottom depth (linear regression: species richness vs. depth, r^^O.SS, P=0.07, diversity (H') vs. depth, r2=0.19, P=0.37, Fig. 5A). Although neither relationship was signif- icant, there was a tendency for platforms in shal- lower water to have both higher species richness and species diversity. Location of the platforms within the Santa Barbara Channel and Santa Maria Basin also did not explain the differences among plat- forms. There was no correlation between northwest to southeast orientation and either species richness or diversity (Spearman rank correlation: species rich- ness vs. orientation, /•,,=-0.26, P=0.6, diversity (H') vs. orientation, r^.=-0.6, P=0.2). In fact, the two most Love et al : Fish assemblages around oil platforms in the Santa Barbara Channel area 103 Figure 4 Fishes typical of offshore oil platforms in the Santa Barbara Channel and Santa Maria Basin: (A) bocaccio, Sebastes paucispinis, (B) flag rockfish. S. rubnvinctus, (C) halfbanded rockfish, S. semicinctus, (D) adult lingcod, Ophiodon elongatus, (E) juvenile lingcod. A-E all on bottom transects, and (F) young-of-the-year rockfish, Sebastes spp,. on midwater crossbeam. 104 Fishery Bulletin 98(1) Table 3 Numbers of fishes observed, by species, arour d all platforms in October-November of 1996. Numbers are given for bottoms of | platforms and mi dwater, with percent of total 3 for those parts of the platforms. Family totals are given in boldface. YOY means "young-of-year." .„. Family Common name Scientific name Bottom Mid water Biomass 7c Total Biomass % Total Scorpaenidae Rockfishes 4212 92.7 2753 91.4 Kelp rockfish Sebastes atrovirens 1 <0.1 Brown rockfish S. auriculatus 7 0.2 Gopher rockfish S. carnatus 2 <0.1 4 0.1 Copper rockfish S. caurinus 347 7.6 11 0.4 Greenspotted rockfish S. chlorostictus 365 8.0 18 0.6 Starry rockfish S. constellatus 1 <0.1 Darkblotched rockfish S. crameri 1 <0.1 Calico rockfish S. dalli 68 1.5 2 <0.1 Greenstriped rockfish S. elongatus 12 0.3 Swordspine rockfish S. ensifer 2 <0.1 2 <0.1 Widow rockfish S. entomelas 115 2.5 1054 35.0 Yellowtail rockfish S. flavidus 1 <0.1 Chilipepper S. goodei 0.00 68 2.. 3 Squarespot rockfish S. hopkinsi 47 1.0 22 0.7 Vermilion rockfish S. miniatus 307 6.8 Blue rockfish S. my St in us 7 0.2 6 0.2 Bocaccio rockfish S. paucispinis 85 1.9 264 8.8 Canary rockfish S. pinmger 10 0.2 Rosy rockfish S. rosaceus 31 0.7 1 <0.1 Greenblotched rockfish S. rosenblatti 129 2.8 Yelloweye rockfish S. ruberrimus 2 <0.1 Flag rockfish S. rubrivmctiis 113 •2.5 15 0.5 Bank rockfish S. rufiis 2 <0.1 Halfbanded rockfish S. semicinctus 2491 54.8 1 <0.1 Olive rockfish S. serranoides 1 <0.1 Treefish S. serriceps 5 0.1 1 <0.1 Pygmy rockfish S. wilsoni 4 <0.1 Sharpchin rockfish S. zacenti-us 11 0.2 1 <0.1 Shortspine thornyhead Sebastolobiis alascan us 1 <0.1 Sebastomus group' 19 0.4 13 0.4 Rockfish YOY Sebastes spp. 26 0.6 1269 42.1 Hexaprammidae Greenlings 244 6.4 187 6.2 Kelp greenling Hexagra m m os decagra m mus 1 <0.1 Lingcod Ophwdon elongatus 193 4.3 Painted greenling Oxylebius pictus 46 1.0 186 6.2 Shortspme combfish Zaniolepis frenata 2 <0.1 Combfish sp. Zaniolepis sp. 3 <0.1 Pomacentridae Damselfishes 12 0.4 Blacksmith Chromis punctipinnis 12 0.4 Embiotocidae Seaperches 65 1.4 20 0.7 Pile perch Damalichlhys vacca 46 1.0 6 0.2 Sharpnose surfperch Phanerodon atripes 9 0.2 14 0.5 Unident. sea perches 1 <0.1 Pink surfperch Zalembius rosaceus 9 0.2 Gadidae Cods 2 <0.1 18 0.6 Pacific hake Mcrluccnis product us 2 <0.1 18 0.6 continued Love et al : Fish assemblages around oil platforms in the Santa Barbara Channel area 105 Table 3 (continued) Family Common name Scientific name Bottom Midwater Biomass % Total Biomass % Total Cottidae Sculpins 1 <0.1 Unidentifed sculpin 1 <0.1 Bathvmasteridae Ronquils 2 <0.1 Unidentified ronquil 2 <0.1 Agonidae Poachers 2 <0.1 Unidentified poacher 2 <0.1 Flatfish Flatfish 13 0.3 Sanddabs Citharichthys sp. 1 <0.1 Unident. flatfish 12 0.3 ' Sebastomus group may include greenblotched, greenspotted, pinkrose, rosethom, rosy. starry, or swordspine rockfishes. similar assemblages were on platforms near the geo- graphic extremes (Hidalgo and Grace). Density and biomass of all species combined also varied among rigs but in a pattern different from species richness and diversity (Fig. 5B). However, similar to richness and diversity, density and bio- mass differences could not be explained by bottom depth or by geography (linear regression; density vs. depth, r-=0.22, P=0.35, biomass vs. depth, 7-2=0.06, P=0.64; Spearman rank correlation: density vs. ori- entation, rj=-0.31, P=0.54, biomass vs. orientation, r,=-0.37, P=0.46). Although bottom depth did not explain the pat- terns of abundance of all species combined, the abun- dance patterns of individual species did relate more strongly to bottom depth. Among the more com- monly observed species, eight showed depth-related patterns of abundance (Fig. 6). Copper and vermil- ion rockfishes, lingcod, and painted greenling were most dense around the bottoms of some of the shal- lower platforms (especially platform Irene). Half- banded and flag rockfishes were most dense on the bottoms of the middepth structures and bocaccio and greenspotted rockfish were most common at the bottom of the deeper platforms. Midwater habitat All platforms Rockfishes also dominated the mid- water portions of the platforms, but were primarily YOYs and slightly older juveniles. Rockfishes repre- sented 91.4% of the individuals (Table 3) and 85.9% of the biomass (Table 2) in the midwater. Although it was difficult to identify many of the smaller individ- uals, widow rockfish were by far the most common Table 4 Percent sim bottom only form Harvet larity indices for each pair of platforms for the in 1996. No bottom surveys were done on plat- t. Platform Gail Grace Holly Hermosa Hidalgo Grace 2.3 Holly 8.5 Hermosa 24.7 34.5 8.5 Hidalgo 17.9 70.1 11.8 60.0 Irene 1.4 4.7 41.5 3.8 10.0 species, representing 35.0% of all fishes seen. It is likely that many of the small, unidentifiable YOYs also were widow rockfish. YOY bocaccio also were fairly abundant around some of the platforms and occasionally schooled with widow rockfish. Both spe- cies formed relatively tight, polarized schools, loosely associated with the pilings and crossbeams (Fig. 4F). When disturbed, the schools immediately swam inward underneath the platform structure. We also saw small numbers of what we tentatively identified as YOYs of the complex of kelp, copper, gopher and black-and-yellow rockfishes (S. chrysomelas). These were found in smaller, much less coherent aggrega- tions and were more likely to move in closer to the substrata when disturbed. Painted greenling, primarily small individuals, were the most commonly seen nonrockfish species. We often saw solitary individuals resting on the crossbeams. Other species occasionally seen near or 106 Fishery Bulletin 98(1) Table 5 Number, densities, and biomasses of fishes observed around the bottoms of six oil platforms off central and southern California. | Platforms are listed geographically, from northwest to southeast. Species are ranked by number observed. YOY means "young-of- year." We computed minimum number of species by assuming that each unidentified taxa (flatfish, poacher , ronquil and seaperch I represented one species. Platform Species Number Density (fish/1 00m-) Biomass (kg/m^) Irene Copper rockfish 297 55.99 10.01 Vermilion rockfish 198 37.33 11.81 Lingcod 152 28.65 1.23 Halfbanded rockfish 25 4.71 0.16 Painted greenling 20 3.77 0.24 Pile perch 20 3.77 0.83 Rosy rockfish 4 0.75 0.02 Sebastomus group' 2 0.38 0.01 Brown rockfish 2 0.38 0.05 Bocaccio 0.19 0.24 Flag rockfish 0.19 0.01 Gopher rockfish 0.19 0.00 Rockfish YOY 0.19 0.01 Widow rockfish 0.19 0.02 Yellowtail rockfish 0.19 0.08 Total 726 136.86 ^ 24.73 Minimum number of species 13 Hidalgo Halfbanded rockfish 552 94.62 9.35 Greenspotted rockfish 109 18.68 3.48 Flag rockfish 58 9.94 1.00 Lingcod 29 4.97 6.52 Bocaccio 17 2.91 2.10 Vermilion rockfish 13 2.23 2.83 Rosy rockfish 10 1.71 0.27 Sharpchin rockfish 10 1.71 0.09 Canary rockfish 7 1.20 1.03 Greenstriped rockfish 7 1.20 0.11 Painted greenling 5 0.86 0.07 Pygmy rockfish 4 0.69 0.04 Widow rockfish 4 0.69 0.64 Squarespot rockfish 3 0.51 0.03 Rockfish YOY 2 0.34 0.06 Shortspine combfish 2 0.34 0.01 Yelloweye rockfish 2 0.34 0.06 Sebastomus group ' 1 0.17 0.01 Bank rockfish 1 0.17 0.02 Unidentified poacher 1 0.17 0.01 Total 837 143.47 27.73 Minimum number of species 18 Hermosa Greenspotted rockfish 179 25.72 3.24 Halfbanded rockfish 98 14.08 0.71 Flag rockfish 16 2.30 0.20 Lingcod 7 1.01 2.60 Rockfish YOY 5 0.72 0.10 Copper rockfish 4 0.57 0.03 Pacific hake 2 0.29 0.00 Sebastomus group' 1 0.14 0.04 Greenblotched rockfish 1 0.14 0.15 continued Love et al : Fish assemblages around oil platforms in the Santa Barbara Channel area 107 Table 5 (continued) Platform Species Number Density (fish/lOOm-) Biomass (kg/m^) Hermosa Greenstriped rockfish 1 0.14 0.01 continued Unidentified poacher 1 0.14 0.00 Sharpchin rockfish 1 0.14 0.01 Starry rockfish 1 0.14 0.02 Widow rockfish 1 0.14 0.02 Total 318 45.70 7.13 Minimum number of species 12 Holly Vermilion rockfish 87 21.98 5.87 Calico rockfish 68 17.18 1.40 Widow rockfish 47 11.88 1.14 Copper rockfish 45 11.37 2.05 Squarespot rockfish 43 10.87 0.41 Halfbanded rockfish 29 7.33 0.12 Pile perch 26 6.57 2.88 Rosy rockfish 16 4.04 0.11 Painted greenling 15 3.79 0.17 Sharpnose surfperch 9 2.27 0.49 Blue rockfish 7 1.77 0.74 Pink surfperch 7 1.77 0.15 Unident. flatfish 6 1.52 0.02 Brown rockfish 5 1.26 0.77 Sebastomus group ' 4 1.01 0.08 Canary rockfish 3 0.76 0.15 Rockfish YOY 3 0.76 0.15 Treefish 2 0.51 0.10 Ronquils 0.25 0.00 Unident. sea perches 0.25 0.20 Combfish sp. 0.25 0.00 Gopher rockfish 0.25 0.01 Olive rockfish 0.25 0.03 Shortspine thomyhead 0.25 0.00 Unidentified fish 0.25 • Total 429 108.40 17.04 Minimum number of species 21 Grace Halfbanded rockfish 1787 351.16 15.87 Flag rockfish 30 5.90 0.18 Greenspotted rockfish 18 3.54 0.38 Vermilion rockfish 9 1.77 0.33 Rockfish YOY 7 1.38 0.05 Unident. flatfish 6 1.18 0.03 Painted greenling 6 1.18 0.03 Widow rockflsh 5 0.98 0.05 Treeflsh 3 0.59 0.09 Combfish sp. 2 0.39 0.01 Lingcod 2 0.39 0.03 Pink surfperch 2 0.39 0.01 Ronquils 0.20 0.03 Copper rockfish 0.20 0.03 Greenblotched rockfish 0.20 0.02 Rosy rockfish 0.20 0.06 Sanddabs 0.20 0.01 Squarespot rockfish 0.20 0.01 continued 108 Fishery Bulletin 98(1) Table 5 (continued) Platform Species Number Density (fish/lOOm-l Biomass (kg/m-) Grace Unidentified fish 1 '0.20 • continued Total 1884 370.22 17.21 Minimum number of species 16 _ Gail Greenblotched rockfish 127 19.8 3.71 Bocaccio 67 10.46 12.01 Greenspotted rockfish 59 9.2 1.70 Sebastomus group ' 10 1.9 0.18 Rockfish YOY 5 0.78 0.35 Greenstriped rockfish 4 0.62 0.20 Lingcod 3 0.47 1.96 Flag rockfish 2 0.31 0.03 Swordspine rockfish 2 0.31 0.03 Bank rockfish 1 0.16 0.27 Darkblotched rockfish 1 0.16 0.07 Total 281 44.17 20.51 Minimum number of species 9 ' Sebastomus group may include greenblotched. green spotted, pinkrose, rosethorn rosy. starry, or swordspine rockfishes on midwater structure included juvenile greenspot- ted and flag rockfishes, as well as sharpnose seap- erch, pile perch, and blacksmith. Among platform comparisons The midwater assem- blages also differed among rigs, although the vari- ability was less than among the bottom assemblages. Species richness ranged from 6 to 11 species per platform (Fig. 7A). Species diversity also showed less variability among platform midwaters than platform bottoms (midwater H' range; 0.7 to 1.8, Fig. 7A). The midwater around platform Irene was dom- inated by widow rockfishes (primarily YOYs, but also one-year-old fishes), unidentified YOY rockfishes (probably primarily widow rockfish) and YOY bocac- cio. Almost no other fishes were noted (Table 6). The species composition at platforms Hidalgo and Harvest was similar to that at platfrom Irene, although painted greenling were also occasionally seen. Far fewer fishes were noted at platform Hermosa, although the spe- cies composition was similar, with the addition of small numbers of Pacific hake. Fewer fishes were seen at platform Holly. Here, YOY rockfish (probably widow rockfish), painted greenling, sharpnose sea- perch and squarespot rockfish were the most common species. Smaller numbers of juvenile widow rockfish, YOY rockfish, and juvenile chilipepper characterized platform Grace. We saw fewest fishes in the midwa- ters around platform Gail where YOY rockfish (again probably widow rockfish) were the most common. In general, the platforms at the western end of the Santa Barbara Channel harbored a higher den- sity of fishes in the midwater than did those towards the east (Fig. 7B). There was a significant rela- tionship between density and northwest-southeast rank (Spearmans r^=0.89, P=0.006). This pattern was due to higher density of YOY rockfishes, espe- cially widow rockfish and bocaccio, at platforms Irene and Hidalgo. YOY rockfishes were abundant only at Irene and Hidalgo, they were much less common at the platforms farther east. There was not a signifi- cant relationship between biomass and northwest- southeast rank (Spearmans r^-0.64, P=0.11). Length-frequency comparisons Relatively few species were abundant in both the mid- water and bottom assemblages. For those species that were found in both environments, such as copper, flag, and greenspotted rockfishes, there was a ten- dency for juveniles to be found in the midwater and older individuals on the bottom. Bocaccio were the extreme example, with smaller juveniles occurring only in midwater and larger individuals only on the bottom (Fig. 8). The painted greenling was one of the few species that occurred in virtually all size classes in both the midwater and on the bottom (Fig. 8). There were considerable differences in the size fre- quencies of the major species around the platforms (Fig. 8). Some species, such as copper rockfish and vermilion rockfish, were found primarily as juveniles and subadults. At the extreme, we did not identify any mature widow rockfish. Numerous other species (i.e. painted greenling, bocaccio, greenspotted rockfish, Love et al Fish assemblages around oil platforms in the Santa Barbara Channel area 109 3- A • Species diversity (H') p25 ^_^ . A Number of species -20 Z a ^x 1 '' \ ^-'f'''''^^^ 3 > \i^-^^^^ ^\^ -'5 3 "3 • ^^"^^ '* V5 (A "S I ■- \ / ~ ~ ~ ~ - -• '^^ -10 2 a \ / C/5 iA ^ / \ / - 5 \ / A • * _ A U- 1 1 1 1 1 1 1 U >^ w w O K « — = c o 00 SJ g 3 o j^ s 13 e 2 o 3= - O 3 S E a X ^ 400 -| B • Density -30 <^ ,«„ A A Biomass -25 t» A ' ^ / \ E 300- 8 /\ ,' y \ 3 &3 S / ' \ / ^ \ / -20 ^ (/3 15 200- M mC \ / fr \ \ / (!5_ >-. ' ^ \ / -15 3 'Jfl -* • ^ \ / Is) c Q 100- *' ""■-\/ ^ ^\ / -10 n — * 1 "'"• c 0-' 1 1 1 1 1 1 1 J >^ u s; fl :^ = c M £ s; S H 2 -S £ g a: - :2 S g X X ^ .„„ Figure 5 (A) Species richness (number of species) and species diversity (Shannon- Weaver diversity index, H') and (B) density lfish/100 m-) and biomass (kilograms/m-) on the bottoms of seven platforms in the Santa Barbara Channel area. Platforms are ordered by bottom depth from shallow (Holly) to deep (Gail). * = no bottom transects done. flag rockfish, and halfbanded rockfish) were found over a wide size range, encompassing most life stages. Although a wide size range of lingcod was observed, it is noteworthy that most of the small fish were found around platform Irene; relatively few of these young individuals inhabited the other platforms. Discussion Although we found large variability in many of the attributes of the fish assemblages living around these seven oil platforms, several consistent patterns were evident. Around all of these structures, the midwa- ter fish assemblage was quite different from that inhabiting the platform bottoms. Juvenile rockfishes were by far the dominant group occupying the mid- water. Although the density of all species combined was similar between the bottom and midwater of any given platform, the biomass was much greater on the bottom, owing to larger fish living around the bottom. In addition, there was a consistently greater number of species on the bottom than in the midwater around each platform. The bottom of the no Fishery Bulletin 98(1) Copper rockfish 60-1 • 50- 40- 30- Vermilion rockfish 40-1 • 30- 20- • 20- 10- • 10- 0- 30- • o • o 0- 4- • • o O 1 1 1 Lingcod • 1 1 1 Painted greenhng • • 20- 3- 2- „ 10- "b • I- • • % 0- — ^ I ■= 400-1 c Q 300- o • • • 0- 10- 7.5- o o 1 1 1 lalfbanded rockfish • Flag rockfish • -J 200- 5H • 100- • 2.5- • 0- 15- • • • 0- 30-1 o • • 1 1 1 Bocaccio I I I Greenspotted rockfish 1 1 X • 10- • 20- • 5- • 10- • 0- O • O o 0- o o O D £ X - o own: ■ago 1 1 1 X ~ o s. -a ■a X 1 n o X 1 "a o Figure 6 Densities (fish/lOOm-) of eight common species on the bottoms of seven platforms in the Santa Barbara Channel area. Platforms are ordered by bottom depth from shallow (Holly I to deep (Gail). Data for greenspot- ted rockfish around platform Gail (noted by *), may include observation of greenblotched rockfish. Empty symbols represent zero values. platforms provided a larger variety of habitat types than did the midwater. Bottoms- are often largely composed of shell mounds that have fallen from the upper parts of the platforms. These mounds, in com- Love et al ; Fish assemblages around oil platforms in the Santa Barbara Channel area 111 2-, • Species diversity (H') ^ Number of species X > ■5 CO 1.5 1 - 0.5- 15 -10 — I- o H ■a o 300 - B • Density ^^ 250 - ^ Biomass E ^ 200 - \ * :? \ V W5 c 150 - \ ^ >^ \ ^ (/5 C too - V ^ i> A ^, a '^A-— '■'■''*• -•\ 50 - " -*- N, - 1 1 1 1 1 1 1 West o East 3 a- -20 -15 -10 CO 5' Figure 7 (A) Species richness (number of species) and species diversity (Shannon- Weaver diversity index, H') and (B) density (fish/100 m^) and biomass ( kilograms/m- ) on the midwater transects of seven platforms in the Santa Barbara Channel area. Platforms are ordered by geography from northwest (Irene) to southeast (Gail). bination with the wells, crossbeams, and pilings pro- vide a greater degree of habitat complexity and thus, may allow a greater number of species to coexist. Platforms north of Pt. Conception in the Santa Maria Basin contain far more YOY rockfishes than those in the Santa Barbara Channel to the south. This geographic difference is almost certainly due to the difference in water masses of the two areas. Platforms north of Pt. Conception are more exposed to the California Current; those south of the Point are more influenced by Southern California Bight water (Brink and Muench, 1986; Hickey, 1992). There is considerable evidence that, within much of the Southern California Bight, juvenile rockfish recruit- ment has been poor for a number of years (Ste- phens et al., 1984, 1994; Love et al., 1998), probably due to decadal-long changes in oceanographic condi- tions. Since the late 1970s, waters off Southern Cali- fornia have warmed significantly and upwelling has declined. This situation has led to reduced zooplank- ton production (Roemmich and McGowan, 1995) and a reduction in the survival of many marine fish spe- cies in early life stages (Holbrook and Schmitt, 1996). The present regime is probably part of a long-term 112 Fishery Bulletin 98(1) Table 6 Number, densities, and biomasses of fi shes observed on the midwater transects of seven oil platforms off central and southern Cali- fomia. Platforms are listed geographically, from northwest to southeast. YOY means "young-of-year." We computed Tiinimum number of species by assuming that each unidentified taxa ( flatfi sh, poacher, ronquil. and seaperchi representee one species Both density and biomass on midwater transects are estimates c alculated from transect minutes (see text f&r explanation of the conversion). Estimated density Estimated biomass Platform Species Number (fish/lOOm^) (kg/m-) Irene Widow rockfish 447 127.25 12.47 Rockfish YOY 271 78.22 1.36 Bocaccio 162 46.56 2.73 Blue rockfish 2 0.85 0.07 Copper rockfish 2 0.85 0.11 Pile perch 2 1.14 0.19 Painted greenling 1 0.57 0.04 Total 887 255.44 16.97 Minimum number of species 6 Hidalgo Rockfish YOY 647 137.88 1.80 Widow rockfish 286 50.57 3.62 Bocaccio 78 19.70 0.29 Painted greenling 29 8.60 0.28 Greenspotted rockfish 2 0.71 0.04 Unident. sculpin 1 0.53 0.03 Flag rockfish 1 0.46 0,04 Total 1044 218.46 6.10 Minimum number of species 6 Harvest Widow rockfish 171 39.45 2.51 Rockfish YOY 102 24.80 0.42 Bocaccio 43 11.53 0.18 Painted greenling 36 11.78 0.51 Unidentified fish 17 5.09 • Blacksmith 8 2.27 0.05 Greenspotted rockfish 5 1.80 0.11 Calico rockfish 1 0.55 0.05 Flag rockfish 1 0.45 0.03 Total 384 97.72 3.87 Minimum number of species 7 Hermosa Painted greenling 77 21.24 1.06 Rockfish YOY 63 17.70 1.41 Widow rockfish 36 11.01 0.99 Pacific hake 18 3.76 0.20 Bocaccio 16 4.60 0.38 Greenspotted rockfish 6 1.94 0.17 Squarespot rockfish 6 2.67 0.20 Sebastomua group ' 4 1..56 0.12 Blue rockfish 3 1.46 0.16 Unidentified fish 3 1.45 • Copper rockfish 1 0.46 0.08 Flag rockfish 1 0.46 0.04 Halfbanded rockfish 1 0.45 0.04 Sharpchin rockfish 1 0.46 0.04 Total 236 69.20 4.90 Minimum number of species 11 Holly Rockfish YOY 62 20.22 0.54 Painted greenling 33 13.75 0.53 continued Love et a\: Fish assemblages around oil platforms in the Santa Barbara Channel area 113 Table 6 (continued) Estimated density Estimated biomass Platform Species Number (fish/lOOm^) (kg/m2) Holly Sharpnose seaperch 14 6.22 1.84 continued Squarespot rockfish 11 4.02 0.12 Copper rockfish 3.77 0.28 Blacksmith 1.82 0.49 Gopher rockfish 2.13 0.23 Pile perch 2.23 0.99 Widow rockfish 1.80 0.31 Bocaccio 0.57 0.04 Kelp rockfish 0.67 0.34 Total 145 57.21 5.70 Minimum number of species 10 Grace Widow rockfish 103 28.92 3.90 Rockfish YOY 76 25.18 0.32 Chilipepper 25 6.73 0.64 Sebastomus group' 9 4.67 0.25 Painted greenling 8 3.37 0.16 Squarespot rockfish 5 1.58 0.18 Flag rockfish 2 0.88 0.03 Greenspotted rockfish 2 0.80 0.06 Swordspine rockfish 2 1.09 0.08 Calico rockfish 1 0.62 0.04 Kelp greenling 1 0.54 0.05 Rosy rockfish 1 0.54 0.07 Treefish 1 0.54 0.04 Total 236 75.47 5.76 Minimum number of species 11 Gail Rockfish YOY 48 21.49 0.72 Flag rockfish 10 4.92 0.40 Widow rockfish 8 3.77 0.27 Bocaccio 7 3.07 0.25 Greenspotted rockfish 3 1.42 0.18 Painted greenling 2 1.29 0.14 Unidentified fish 2 1.10 • Blue rockfish 1 0.63 0.06 Total 81 37.69 2.00 Minimum number of species 6 ' Sebastomus group may include greenblotched, green ^potted, pinkrose. rosethorn, rosy starry, or swordspine rockfishes. alternation of warm- and cold-water conditions that have occurred over millennia (MacCall, 1996). Previous surveys of rockfishes at the two most inshore platforms of the Santa Barbara Channel, Hilda and Hazel, provide some evidence for the plas- ticity of rockfish populations in the Santa Barbara Channel. In the late 1950s, Carhsle et al. (1964) found large numbers of bocaccio and olive, copper, and brown rockfishes. Most of these fishes were either YOYs or older juveniles. By 197.5, olive and brown rockfishes were still abundant, but bocaccio and copper rockfishes were uncommon (Bascom et al., 1976). In this latter survey, blue rockfish, not reported by Carlisle et al. (1964), were abundant. Thus, we believe that the relative dearth of juve- nile rockfishes around Southern California Bight platforms is not a permanent condition but repre- sents a fluctuating system. It is likely that as ocean- ographic conditions in the Southern California Bight become more favorable to rockfish recruitment, off- shore platforms in the Santa Barbara Channel may well harbor far greater numbers of juvenile rock- fishes than at present. In fact, indirect evidence implies that juvenile rockfishes were at one time 114 Fishery Bulletin 98(1) far more abundant around southern California plat- forms. This conclusion comes from observations we made in the mid-1970s, a period of relatively strong juvenile rockfish recruitment off California (Love and Westphal, 1990). During that period, we observed a significant recreational fishery directed at juvenile widow rockfish and bocaccio (and to a certain extent olive and blue rockfishes) at platform Holly, as well as at a number of other Santa Barbara Channel plat- forms. We estimate that tens of thousands of these YOY and 1- and 2-yr-old fishes were caught over the course of about three years. The absence or relative rarity of such common nearshore species as kelp bass (Paralabrax clath- ratus), opaleye iGirella nigricans), black seaperch (Embiotoca jacksoni), and white seaperch (Phaner- odon furcatus) from the upper waters was partic- ularly striking. This is in contrast to the inshore platforms and reefs of this area that harbor many of these species (Carlisle et al., 1964; Ebeling et al., 1980; Schroederf ). A most important cause for the absence of nearshore species is the isolation of these offshore structures; relatively deep water separates them from the mainland. This distance may effec- tively cut these species off from source populations of many shallow-water species. Thus, it may be dif- ficult for the young of many species to either reach these platforms or become established there. Sea- perches are viviparous and produce fully developed ^ Schroeder, D. 1997. Marine Science Institute, University of California, Santa Barbara, CA 93106. Personal commun. 00- Painted grc ■en ling 75- 50- T 25- 0^ ^ u 13 — t— 23 l.OOfri 750- I— u XI E 500- 3 z 250- 0- Widow rockfish I -M-P- 1 — I — I — i— 13 18 23 28 33 38 43 48 150 100 50 Bocaccio I -I— t- X^ -r—r 3 8 13 18 23 28 33 38 43 48 53 58 Copper rockfish 13 18 23 28 33 38 300- 250- 200- 150- 100- 50- Greenspotted rockfish I MYn. 13 18 23 28 33 38 Flag rockfish 13 18 23 28 33 38 Midpoint length (cm) 80 Lingcod Bottom n Midwater 0-^ f tT I f TT 8 1318232833384348535863687378 150- 100- Vermilion rockfish 50- 8 13 18 23 28 33 38 43 48 53 58 Halfbanded rockfish 1 1 — ^- 13 18 23 28 33 Figure 8 Length-frequency distributions of nine common species on midwater and bottom tran.sects on all platforms combined. Midpoint length is the midpoint of .S-cm length bins. Love et al,: Fish assemblages around oil platforms in the Santa Barbara Channel area 115 young that do not disperse widely, making it unlikely that they commonly find their way to platforms. Kelp bass and opaleye produce pelagic larvae and although it is likely that some may settle to the plat- forms, conditions at these structures may preclude their survival after settlement. Young opaleye seem to require quiet intertidal waters and kelp bass YOY may need algae or thick benthic turf to avoid preda- tion (Carr, 1994; Stephens'*). Both of these conditions are lacking at platforms. Moreover, in the study area kelp bass recruitment is only sporadic and may not have occurred in the recent past. Thus, strong cur- rents and lack of suitable habitat around platforms may reduce the amount of successful recruitment of these and other nearshore species. A few species, notably painted greenling, do seem to be well adapted to a substrate-associated life in the midwaters. Judging from the very small individ- uals we observed, it is likely that lai-vae of this spe- cies recruit directly to the platform and settle out in the shallower portions. We saw a wide range of sizes, from newly settled individuals to adults, sitting on the crossbeams and hanging vertically on the pil- ings. Other than painted greenling, only a few juve- nile flag, greenspotted, copper, swordspine, gopher, and rosy rockfishes were seen sitting on the platform in midwater. Although juvenile rockfishes dominated the plat- form midwater, for some species platform bottoms tended to harbor a wider range of life stages. For some rockfishes (such as copper and greenspotted rockfishes), the entire range of stages from YOY to adults were present. In these species, the smaller individuals tended to live somewhat away from the legs and crossbeams and more among those parts of the mussel shell mounds a few meters from the platform. Although juvenile vermilion rockfish were common on several of the shallower platforms, we saw no YOYs around any of these structures. Ver- milion rockfish tend to settle out in the nearshore, relatively shallow waters, and it is likely that even the shallowest of the surveyed platforms were situ- ated in waters too deep for successful recruitment. This supposition was born out in our SCUBA diver surveys of platform Gina, located off Port Huen- eme, southern California. Platform Gina is located in waters about 33 m deep. Divers have surveyed the entire structure and on several occasions have noted YOY verm.ilion rockfish at the bottom of the platform. The situation with lingcod is particularly interest- ing. Including observations from all platforms, we observed all life stages from YOYs to large adults. However, almost all the young fish lived around plat- form Irene, in relatively high densities. From the lengths of these animals (Miller and Geibel, 1973), we determined that these fish were either YOYs or one-year-olds. We noted that most were sitting in the mussel shells on the bottom slightly away from the structure. In an underwater survey that encompasses seven platforms and 61 natural reefs in central and southern California, we have never encountered juvenile lingcod densities approaching the levels noted around platform Irene. Similar submersible research farther north, off Big Sur- Monterey (Yoklavich^) and Alaska (O'Connell^) also implies that such aggregations are very rare. The aggregation around Irene may also be relatively stable because we saw similar high densities in the subsequent 1997 survey. It is unclear what attracts young lingcod to this location. A large juvenile aggre- gation was noted off Big Sur on a sandy bottom cov- ered with ripple marks (Yoklavich"*). Perhaps young lingcod seek out substrate with at least some verti- cal relief and, at Irene, mussel shell mounds provide this type of relief. Many bottom fishes tended to be patchily distrib- uted around individual platforms. This is particu- larly true of the aggregating species, such as bocaccio and vermilion and halfbanded rockfishes. Whether this is in response to current pattern, variations in platform structure, or to other parameters is not clear at this point. We have also observed a ten- dency for small individuals, such as halfbanded rock- fish or juvenile greenspotted rockfish, to be found away from larger, presumably predacious, individu- als. Smaller fishes also tend to be found farther away from the platform, again probably to avoid the larger fishes nestled in the structure. Fishing pressure is intense over most of the natural reefs in southern California and platforms may act as refuges for rockfishes and lingcod. An example is the relatively high numbers of bocaccio living around platform Gail. Historically, bocaccio were very impor- tant recreational and commercial fish along all of Cal- ifornia and owing to a combination of over-fishing and poor juvenile recruitment, their populations have drastically decreased ( Ralston et al. , 1996 ). Our survey of the fish assemblages of 61 natural reefs off south- em and central California shows that platform Gail has by far the highest density of adult bocaccio of all of these sites ( 10.5 fish/100 m^ on platform Gail ■• Stephens, J. 1997. Department of Biology, Occidental College, 1600 Campus Rd., Los Angeles, CA 90041. Personal commun. ■^ Yoklavich. M. 1997. Pacific Fisheries Environmental Labora- tory, National Marine Fisheries Service, 1352 Lighthouse Ave., Pacific Grove, CA, 93950. Personal commun. '^ O'Connell, T. 1998. Alaska Department of Fish and Game, 304 Lake St., Rm. 103, Sitka. AK, 99835. Personal commun. 116 Fishery Bulletin 98(1) compared with 4.4 fish/100 m^ on the highest density natural reeD. The reef was located on the northern side of the passage between San Miguel and Santa Rosa islands. The average density of bocaccio across all natural reefs surveyed in 1996 was only 1.26 fish/ 100 m-. The large numbers of bocaccio around Gail may reflect the minimal fishing pressure around this platform. Fishing by recreational or commercial ves- sels near platforms is generally discouraged by plat- form operators. In addition, because larger fishes tend to live close to or inside the platforms, they are difficult to catch because the habitat close to or inside the platforms eludes most fishing gear. We realize that the data presented in this paper represent a "snapshot" in time and thus issues of seasonality or interannual variation in assemblage structure remain to be addressed. Longer-term sur- veys of the fish fauna on two platforms in the Gulf of Mexico as well as one in the Santa Barbara Channel showed considerable diel and seasonal variation in the number of species present (Carlisle et. al 1964; Hastings et. al. 1975). In addition, monthly SCUBA observations on one shallow-water platform indicate that there may be large temporal changes in assem- blage structure (Schroeder-). Despite this, the differ- ences we observed in fish assemblages among and within platforms suggest that each platform may have unique characteristics. There has been considerable discussion regarding the role of artificial structures in aggregation or enhanced production of marine species (or both) (Carr and Hixon, 1997). Based on this study, it appears that oil platforms may serve to do both. First, large adult fishes of several species were pres- ent on several platforms where no juveniles of those species had previously been observed, e.g. vermilion rockfish. It appears that those adults may have set- tled away from the platforms and migrated to them at some life stage. On the other hand, several plat- forms had very large numbers of very young fish that presumably settled to the platforms directly from the plankton, e.g. widow rockfish. If we assume that some of these young fishes would not have found appropriate settling habitat, then platforms, at least in the short term, do play some role in enhancing pro- duction. To ultimately assess the role of platforms in production of reef fishes, it will be necessary to under- stand the fate of the young fish settling to them. Acknowledgments We would like to thank Bob Lea, Mary Nishimoto, Donna Schroeder, Rick Starr, and Mary Yoklavich, all of whom were instrumental in -helping us collect data. We would also like to express our appreciation to the crew of the RV Cavalier, Douglas Morse, Jona- than Blackman, Don Chesnut, Don Tondro, Nancy Stewart, Erik Kohnhorst, and the pilots of the sub- mersible Delta, Chris Ijames, and Dave Slater, for their very professional handling of the technical aspects of this survey. Lyman Thorsteinson was, as always, extremely supportive and we thank him. This research was based on an information need identified by the Minerals Management Service's Pacific OCS Region and funded through the U. S. Geological Survey Biological Resources Division's National Offshore Environmental Studies Program (1445-CA-0995-0386). Literature cited Bascom, W., A. J. Mearns, and M. D. Moore. 1976. A biological survey of oil platforms in the Santa Bar- bara Channel. J. Pet. Tech. 28:1280-1284. Brink, K. H., and R. D. Muench. 1986. Circulation in the Point Conception-Santa Barbara Channel region. J. Geophys. Res. 91(C1 1:877-895. Carlisle, J. G., Jr., C. H. Turner, and E. E. Ebert. 1964. Artificial habitat in the marine environment. Calif Dep. Fish and Game, Fish. Bull. 124, 93 p. Carr, M. H. 1994. Effects of macroalgal dynamics on recruitment of a temperate reef fish. Ecology 7.5:1.320-13.33. Carr, M. H., and M. A. Hixon. 1997. Artificial reefs: the importance of comparisons with natural reefs. Fisheries 22:28-33. Ebeling, A. W., R. J. Larson, W. S. Alevizon, and R. N. Bray. 1980. Annual variability of reef-fish assemblages in kelp for- ests off Santa Barbara, CaHfornia. Fish. Bull. 78:361-77. Hastings, R. W., Ogren, L. H., and M. T. Mabry. 1975. Observations on the fish fauna associated with off- shore oil platforms in the Northeastern Gulf of Mexico. Fish. Bull. 74:387-402. Rickey, B. M. 1992. Circulation over the Santa Monica-San Pedro basin and shelf Prog. Oceanog. 30:37-115. Holbrook, S. J., and R. J. Schmitt. 1996. On the structure and dynamics of temperate reef fish assemblages — are resources tracked? In M. L. Cody and J. A. Smallwood (eds ), Long-term studies of vertebrate communities, p. 19-48. Academic Press, San Diego, CA. Love, M. S., J. Caselle, and K. Herbinson. 1998. Declines in nearshore rockfish recruitment and pop- ulations in the southern California Bight as measured by impingement rates in coastal electrical generating sta- tions. Fish. Bull. 96:492-.501, Love, M. S., J. Hyland, A. Ebeling, T. Herrlinger, A. Brooks, and E. Imamura. 1994. A pilot study of the distribution and abundances of rockfishes in relation to natural environmental factors and an offshore oil and gas production platform off the coast of southern California. Bull. Mar Sci. 55:1062-1095. Love, M. S., and W. Westphal. 1990. Comparison of fishes taken by a sportfishing party vessel around oil platforms and adjacent natural reefs near Santa Barbara, CaHfornia. Fish. Bull. 88:599-605. Love et al Fish assemblages around oil platforms In the Santa Barbara Channel area 117 MacCall, A. D. 1996. Patterns of low-frequency variability in fish popula- tions of the California Current. Calif Coop. Oceanic Fish. Invest, Rep. 37:100-110. Miller, D. J., and J. J. Geibel. 1973. Summary of blue rockfish and lingcod life histories; a reef ecology study; and giant kelp, Macrocystis pyrifera, experiments in Monterey Bay, California. Calif Dep. Fish Game. Fish Bull. 158, 137 p. Ralston, S., J. N. lanelli, R. A. MiUer, D. E. Pearson, D. Thomas, and M. E. Wilkins. 1996. Status of bocaccio m the Conception/Monterey/Eureka INPFC areas in 1996 and recommendations for manage- ment in 1997, Appendix B. In Status of the Pacific Coast groundfish fishery through 1996 and recommended accept- able biological catches for 1997, stock assessment and fish- ery evaluation, p. 1^8. [Available from Pacific Fisheries Management Council, 2130 SW 5th Ave, Suite 224, Port- land, OR 97201.1 Roemmich, D., and J. McGowan. 1995. Climatic warming and the decline of zooplankton in the California Current. Science (Wash. D.C.) 267:1324-1323. Shannon, C. E., and W. Weaver. 1949. The mathematical theory of communication. Univ. Illinois Press, Urbana, IL, 117 p. Stephens, J. S. Jr., P. A. Morris, D. J. Pondella, T. A. Loonce, and G. A. Jordan. 1994. Overview of the dynamics of an urban artificial reef fish assemblage at King Harbor, USA, 1974-1991: a recruit- ment driven system. Bull. Mar. Sci. 55:1224-1239. Stephens, J. S. Jr., P. A. Morris, K. Zerba, and M. Love. 1984. Factors affecting fish diversity on a temperate reef: the fish assemblage of Palos Verdes Point, 1975-1981. Env. Biol. Fish. 11:259-275. 118 Abstract.-The Gulf of Mexico is the only known spawning area for bluefin tuna {Thunnus thynnus thynnus) in the western Atlantic. Although it is known from tag recaptures that east- em Atlantic bluefin tuna travel to the western Atlantic, whether or not these fish spawn in the western Atlantic is of critical importance in interpreting the significance of this movement. East Atlantic bluefin tuna mature at a younger age (4-5 yr) and smaller size (45 kg) than western bluefin tuna (8 yr and 135 kg), and tag recaptures indi- cate that some young fish make the trans-Atlantic swim. Thus the presence of small (<135 kg) bluefin tuna in the Gulf of Mexico during spawning season would constitute evidence that bluefin tuna of east Atlantic origin spawn in the west. We used size-frequency analysis to test the hypothesis that Atlantic blue- fin tuna of eastern and western origins mingle on the Gulf of Mexico spawning grounds. We created a simple model to estimate the proportion of small east- ern spawning fish that should be found in the Gulf of Mexico catch, assuming a 2% east-to-west transfer rate and com- plete mixing of eastern and western fish. Using conservative assumptions, the model predicts that between 5% and 10% of the bluefin tuna catch in the Gulf should consist of fish that are less than 135 kilograms in weight, and thus are presumably eastern migrants. We analyzed Gulf of Mexico catch records from 1980 to 1992 for the presence of bluefin tuna less than 135 kg. These small fish represented from 0'7c to 0.9% of the catch annually, and only 0.3% for the entire period. We conclude that eastern migrant tuna do not mix com- pletely, if at all, with western bluefin tuna on the Gulf of Mexico spawning grounds. Spawning site fidelity in Atlantic bluefin tuna, Thunnus thynnus: the use of size-frequency analysis to test for the presence of migrant east Atlantic bluefin tuna on Gulf of Mexico spawning grounds David Nemerson National Audubon Society 550 South Bay Avenue Ishp, New York 11751 Present address Marine Field Station Institute of Manne and Coastal Sciences Rutgers University 132 Great Bay Boulevard Tuckerton, New Jersey 08087-2004 E-mail address nemersoniatmcsrutgers edu Steven Berkeley Oregon State University Hatfield Manne Science Center Oregon State University Newport, Oregon Carl Saflna National Audubon Society 550 South Bay Avenue Islip, New York 11751 Manuscript accepted 1 December 1998. Fish Bull. 98:118-126 12000). Atlantic bluefin tuna ( Th un n us thyn - nus thynnus) is a highly migratory pelagic species that ranges through- out the Atlantic between 60°N lati- tude and the equator, although it has not been encountered south of 20°N since the 1960s. Two blue- fin tuna breeding sites are known in the North Atlantic: the Gulf of Mexico and the Mediterranean Sea. No other regular spawning site has been identified in the North Atlantic (Richards, 1976;McGowan and Richards, 1989; NRC, 1994). Intensive fisheries exist for bluefin tuna along the North American and European coasts, and to a lesser degree in the high seas of the North Atlantic. Although fish tagged on both sides of the ocean have been recovered on the side opposite from their release, it is not known if blue- fin tuna return to their natal spawn- ing ground to reproduce (Turner and Powers, 1995; Cooke and Lank- ester, 1996). This question is of utmost importance in evaluating the significance of trans-Atlantic movement and the scale at which management must operate to be effective. The behavior of trans-Atlantic migrating bluefin tuna is unknown, but the possibilities are bounded by two extremes. At one extreme, an emigrant may join the popula- tion on the side of the ocean to which it migrates, becoming indis- tinguishable from the population it joins with respect to the proba- bility, timing, and locale of future life history events, such as matura- tion, spawning, and migration. At the ether extreme, a migrant may always return to its natal side prior to the next spawning season. Nemerson et al.: Spawning site fidelity in Thunnus thynnus 119 These two extremes, and the terms used to describe them, have been the subject of some confusion in the bluefin tuna hterature. The permanent transfer of individuals from one side of the Atlantic to the other has been called the "no-memory condition" (Punt and Butterworth, 1995; Powers and Cramer, 1996) or the "diffusion model" (Cooke and Lankester, 1996). The case where migrants always return to their natal side prior to the next spawning season is termed the "overlap model" by Cooke and Lankester (1996). We follow the convention of Cooke and Lankester (1996) and use the terms diffusion and overlap to refer to the two models. We use "transfer rate" to refer to the per- manent transfer of an emigrant from one population to the other and "migration rate" to refer generally to the trans-Atlantic movement of individuals. Finally, we use the term "memory" to refer only to an individ- ual's behavior with respect to spawning location, not to other life history attributes. That is, under the dif- fusion (no-memory) model, a migrant will spawn on the side of the ocean to which it migrates, regardless of its birth location, but will retain other life history attributes such as size or age at maturity. The permanent transfer of individuals can be con- sidered a migration for dispersal (Greenwood and Harvey, 1982), whereas the overlap model can be assumed to be a feeding migration, and is free of implications for reproductive mixing. One can envi- sion intermediate scenarios combining varying degrees of memory, or philopatry. For example, migrants may remain on the opposite side for a period of years, while either participating in or foregoing spawning, before ultimately returning to their natal side. Fur- thermore, some migrants may exhibit spawning site fidelity while others may stray, joining previ- ously established spawning populations (e.g. Curry, 1994). Simulation models have shown that the dynamics of the two populations are potentially very sensitive to even low trans-Atlantic migration rates, partic- ularly for east-to-west transfer (NRC, 1994; Porch et al., 1995; Punt and Butterworth, 1995; Powers and Cramer, 1996) because the average size of the eastern population has been about 6 to 13 times that of the western population over the past 20 years (ICCAT^'^; Fig. 1). Recent spawning biomass estimates for the western population are based on ' ICCAT ( International Commission for the Conservation of Atlan- tic Tunas). 1994a. WestAtlanticbluefin tuna. Biennial report of the ICCAT Standing Committee on Research and Statistics, 41 p. ICCAT, Estebanez Calderon 3, E-28020, Madrid. Spain. ^ ICCAT (International Commission for the Conservation of Atlan- tic Tunas). 1994b. East Atlantic bluefin tuna. Biennial report of the ICCAT Standmg Committee on Research and Statistics, 31 p. ICCAT, Estebanez Calderon 3, E-28020, Madrid, Spain. East Atlantic Population West Atlantic Population Figure 1 Representation of the effect of migration on the eastern and western populations of bluefin tuna. Migration from the larger eastern population to the west has a larger effect on the western population than does migration from the smaller western population to the east. In this schematic, the eastern population is about six times the size of the western population, and the migration rates are about 1% in each direction. catches throughout the fishing area, which includes the entire North Atlantic west of 45°W longitude. If fish of eastern origin are included in these catch statistics but do not spawn in the west Atlantic, then western spawning biomass will be substantially overestimated (Powers and Cramer, 1996; American Fisheries Society'^). Determining the spawning site fidelity of itero- parous pelagic species that occur over a wide area of open ocean is difficult. Population differentiation can be inferred from tag-return data, comparisons of life history parameters and morphometric characters, or from genotypic variation. Several studies have attempted to analyze the population structure of Atlantic bluefin tuna with these methods (Calaprice, 1986; NRC, 1994; Cooke and Lankester, 1996). Several investigators have reviewed and ana- lyzed trans-Atlantic tag returns to estimate rates of migration (NRC, 1994; Punt and Butterworth, 1995; Turner and Powers, 1995; Cooke and Lank- ester, 1996). These studies have estimated annual migration rates of between 1% and 10% and have considered both diffusion and overlap models. Gen- erally, these studies have sought to find interpreta- tions of tag-return data that agree best with other estimates of population size. •* American Fisheries Society. 1995. Marine Fisheries Section statement on bluefin tuna, 2 p. Am. Fish. Soc, 5410 Grosvenor Lane, Ste. 110, Bethesda, MD 20814-2199. 120 Fishery Bulletin 98(1) Punt and Butterworth (1995) estimated west-to- east transfer at about 7% and east-to-west transfer at about 1.5-3%, assuming a diffusion model. They also state that the higher end of the range (3%) sug- gests a far larger size for the western population than do models that assume no migration. Cooke and Lankester (1996) test both diffusion and over- lap models and concluded that the overlap model fits the data better. Under that model, they estimated exchange rates at 7.3% east-to-west and 9.8% west- to-east, but with no statistical difference between the two. Powers and Cramer ( 1996) examined the impli- cations of a range of migration rates and degrees of spawning site fidelity. Although they made no con- clusions about which scenario is most likely, they pointed out the extreme sensitivity of the results to the assumptions. Eastern and western Atlantic bluefin tuna popu- lations have markedly different life history parame- ters (Turner, 1994). The western population spawns from mid-April to mid-June (Richards, 1976). West- ern bluefin tuna sometimes mature as early as age 6 and are considered fully mature by age 8, at a weight of 135 kg ( Baglin, 1982; NRC, 1992 ). The east- ern population spawns from June through August (Dicenta and Piccinette, 1980) and matures at an earlier age and smaller size than the western pop- ulation. Eastern bluefin tuna mature as early as age 3, at a weight of 15 kg ( Rodriguez -Roda, 1967; Baglin, 1982), and are fully mature by age 5 (Rodri- guez-Roda, 1967; Baglin, 1982; ICCAT-). The contrast in size and age at maturity of western and eastern Atlantic bluefin tuna allows an inferen- tial test of spawning site fidelity. Because the Gulf of Mexico is the only known spawning ground for west- ern Atlantic bluefin tuna, and the vast majority of fish collected in the Gulf are large adults that are present only during and just prior to the spawning season (January-June), we assumed that all bluefin tuna in the Gulf during this time period are there to spawn. If eastern Atlantic bluefin tuna that migrate to the west mature according to the eastern Atlantic matu- ration schedule, then the size distribution of bluefin tuna in the Gulf of Mexico should reveal the pres- ence of eastern migrants within the western spawn- ing population. Finding small fish (<135 kg) on the Gulf of Mexico spawning grounds would support the diffusion hypothesis and suggest that trans-Atlan- tic migrants from the east mix with western fish during spawning. In contrast, the absence of small fish on the Gulf of Mexico spawning grounds would imply that eastern migrants do not spawn in the west, supporting the overlap model and indicating strong spawning site fidelity. We know that small bluefin tuna from the east Atlantic swim west at least occasionally. All tagged eastern Atlantic blue- fin tuna recaptured in the west have been small fish (n=19, all captured outside the Gulf), although very few large fish, and relatively few bluefin tuna over- all, have been tagged in the east, compared with tag- ging in the west (NRC, 1994). Methods We analyzed the size distribution of bluefin tuna caught in the Gulf of Mexico prior to and during the spawning season (the only time of year when bluefin tuna are present in the Gulf) for fish between the known size of first breeding in the Mediterranean and the known size of first breeding for west Atlan- tic bluefin tuna. Any individuals smaller than the known size of first spawning in the west would pre- sumably be of eastern Atlantic origin. A weight-frequency distribution (WED) of bluefin tuna on the Gulf spawning grounds was constructed by using data reported to National Marine Fisheries Ser- vice (NMFS) by the commercial fishing industry oper- ating in the Gulf This data set included the weight and date of capture for every bluefin tuna legally caught and landed in the Gulf We used data from 1980 through 1992, because beginning in 1993 only bluefin tuna over 178 centimeters ( 70 inches) fork length were legally permitted to be retained and sold.^ To estimate the proportion of smaller eastern spawning fish expected at a given east-to-west annual transfer rate (i.e. fish remain with the western popu- lation), we created a simple model of the number of sexually mature eastern migrants that arrive in the west each year: where S^^. = the number of age-7 or younger spawn- ing fish of eastern origin arriving in yeary; P„ = the percentage of sexually mature adults in eastern age class a; T^ = the east-to-west transfer rate; and A^^^ ^ - the number in eastern age class a in yeary. East-to-west transfer was modeled as an instan- taneous process that occurs prior to the spawning season. The parameter P was taken from the lit- ■• National Marine Fisheries Service. 1995. Supplemental draft environment impact statement for a regulatory amendment for the western Atlantic bluefin tuna. U.S. Dep. Commer. NMFS. NOAA, Silver Spring, MD, 131 p. Nemerson et a\ Spawning site fidelity in Thunnus thynnus 121 erature, and is for ages 0-3, 0.5 for age 4, and 1 for ages 5 and beyond (Baglin, 1982; ICCAT"). We assumed that any migrant of age 8 or greater would be the same size as western spawning fish and would not be distinguish- able from western spawning fish of the same size (Cort, 1991; Turner et al., 1991; Table 1 ). We used a transfer rate T of 29^ per year, east-to- west. This rate is at the low range of published estimates. In this initial test, we did not consider fish less than age 4 that could have migrated to the west as immature fish in prior years and then reached age 4 and maturity in the current year. Thus, our esti- mate of the expected number of spawning fish of eastern Atlantic origin in the western Atlantic should represent a minimum estimate and provide a conservative test for the presence of east- ern migrants. N^a.\ was taken from yearly age-spe- cific population estimates supplied by NMFS from a run of the ADAPT virtual population analysis (VPA) program with 27c east-to-west and a 1% west-to-east transfer rates, assuming no memory. Note that the population estimates from this VPA run resulted in poor fits to the indices of abundance used to tune the VPA. '5 We used these population estimates because they provided a conservative test of our assump- tions. Results and discussion Bluefin tuna smaller than the accepted size at first spawning of western fish are very rare in the Gulf. Catches of fish less than 135 kg ranged from 0% to 0.9% of annual catch from 1980 to 1992 and aver- aged 0.3*^ over the entire period (Table 2). A com- plete weight frequency distribution is presented in Figure 2. These percentages are not consistent with the low end of published migration rate estimates under the diffusion model. That is, if 27( of each age class Weight (kg) Figure 2 Weight-frequency distribution of bluefin tuna caught in the Gulf of Mexico between 1980 and 1992. The numbers over the bars indicate the total number of fish caught in the weight interval indicated. Table 1 Estimated length at age for eastern (Cort. 1991) and west- 1 ern (Turner et al. 1991) Atlantic bl uefin tuna. Age (yr) Length (cm) East West 1 53.4 48.6 2 77.0 73.8 3 98.4 97.0 4 118.0 118.5 5 135.8 138.4 6 152.1 156.8 7 166.9 173.7 8 180.4 189.4 ^ Porch, C. 1995. National Marine Fisheries Service. South- east Fisheries Science Center, 75 Virginia Beach Drive. Miami, FL 33149. Personal commun. migrated from east to west and joined the western pop- ulation, we would expect to see many more small fish, i.e. fish of eastern origin, spawning in the west. ( Recall that the diffusion, or no-memory model, implies that the migrant does not "remember" its natal spawning gi'ound but does "remember" its maturation sched- ule.) The model predicts that between 8483 and 14,655 sexually mature migrants smaller than 135 kg would have arrived each year in 1980-92. We compared these numbers with the numbers of sex- ually mature fish estimated for the west from the 122 Fishery Bulletin 98(1) Table 2 Comparison of the number and proportion of s pawning fish less than 135 kg actually caught in the Gulf of Mexico from 1980 to 1992 and predictions on the ba sis of a hypothesized 2'7c annual trans-Atlantic transfer rate. The number of small spawning fish observed is significantly less than the predicted number (X'^=353, P<0.0001). Data in columns 2 and 3 from NMFS ADAPT VPA with 2% east-to-west and 1% west-to-east transfer rates and no memory columns 4 and 5 are unpublished data, NMFS. Number of Total Actual proportion Predicted Western Predicted bluefin tuna bluefin tuna of catch in proportion of catch population migrant <135kg caught the Gulf in the Gulf Year age 8 or greater spawning fish caught in the Gulf m the Gulf <135 kg <135 kg 1980 97,824 10,745 19 0.099 1981 97,698 9,732 255 0.091 1982 88,781 9,402 228 0.096 1983 96,739 8,483 316 0.081 1984 92,440 8,691 320 0.086 1985 83,659 10,623 429 0.113 1986 89,444 13,151 1 395 0.003 0.128 1987 106,182 14,561 3 474 0.006 0.121 1988 117,440 14,655 3 516 0.006 0.111 1989 134,585 13,035 1 273 0.004 0.088 1990 167,115 10,631 4 469 0.009 0.060 1991 191,036 8,525 1 596 0.002 0.043 1992 260,192 9,436 1 399 0.003 0.035 ADAPT VPA run supplied by NMFS (Table 2). Table 2 shows that, if the no-memory assumption is cor- rect and the trans-Atlantic transfer rate is at least 2%, we would expect about 5% to 10% offish spawn- ing in the west to be smaller than 135 kg. (Note that for this comparison to be valid, either all mature fish in the west and all small migrant spawning fish must go to the Gulf of Mexico to spawn or the same proportion of each group must go there each year. ) In fact, only 0% to 0.3% of fish on the Gulf spawn- ing grounds between 1980 and 1992 weighed less than 135 kg (Fig. 3), significantly less than predicted (X2=353, P<0.0001). Of 4688 fish for which NMFS has recorded weights, 15 were less than 135 kg, and 10 of those were between 120 and 135 kg (Fig. 2). Note that this analysis considered only newly arrived migrants each year, and ignored the possible accumulation of sexually mature migrants from pre- vious years that had not yet reached 135 kg. The con- tinued presence of prior migrants would have raised the expected number of small spawning fish. Even without the cumulative effect of small migrants, the actual proportion of small spawning fish in the Gulf catch was about 5—10% of that predicted by the model with a 2% transfer rate. Higher transfer rates would imply that even greater numbers of small spawning fish should appear in the Gulf It is clear from our results that small bluefin tuna are not present among spawning fish in the Gulf of Mexico in the numbers that would be expected for even the lowest of hypothesized trans-Atlantic trans- fer rates. Our interpretation is that young adult blue- fin tuna of eastern origin seldom or never spawn in the Gulf of Mexico and presumably do not contribute significantly to the spawning biomass of the western population. There are at least three possible alterna- tives: 1) eastern migrants may either delay spawn- ing in the west until they reach 135 kg or remain in the east until they reach 135 kg, making them indis- tinguishable from western spawning fish; 2) migrant eastern tuna may be spawning in the west but not in the Gulf of Mexico; or 3) small migrants may be spawning in the Gulf but are avoiding detection or are being under-reported. Size and age at maturity If size and age at maturity are environmentally determined, then eastern migrants might follow the west Atlantic maturity schedule and thus be unde- tected with our methods. For example, changes in size and age at maturity may be a response to differ- ences in interspecific or intraspecific population den- sity. A lower population density reduces competition for food and increases per capita food intake, resulting in faster growth. When experiencing such low inter- or intraspecific population densities and enhanced growth, fish may mature at about the same size but would attain this size at a younger age (Trippel, 1995). However, for bluefin tuna, the reported dif- Nemerson et al.: Spawning site fidelity in Thunnus thynnus 123 ference in size and age at maturity between east and west Atlantic populations does not appear related to dif- ferences in growth rate be- cause recent growth models indicate little difference be- tween populations (Cort, 1991; Turner et al., 1991; Table 1). Similarly, if differences in age or size at maturity are affected by environmental conditions, for example tem- perature, we would expect this effect to be manifested pri- marily by changes in growth rate. Again, the similarity in growth rate between east and west Atlantic bluefin tuna suggests that environmental conditions are unlikely to explain the difference in size and age at maturity. Alternatively, if age at ma- turity is a heritable trait, then a long period of size- selective fishing mortality could shift genotype fre- quencies in the population because few late-maturing fish are likely to survive to reproduce (Trippel, 1995), resulting in a younger age or smaller size at matu- rity, or both (Policansky, 1993; Trippel, 1995). Experi- ments with guppies indicate that increased mortality (as through fishing) selects for earlier maturity at smaller size (Reznick, 1993). Bluefin tuna in the east Atlantic has a longer history of exploitation and a much larger population than west Atlantic bluefin tuna. Assuming that the very large difference in population sizes results in a comparable difference in stock density, then an eastern bluefin tuna with a genetic propensity to mature at or before age 5 in the east Atlantic should, upon migrating to the west Atlantic, find itself in a relatively resource-rich, lower-density environment, which should certainly not delay maturation or inhibit spawning. Thus, although genetic effects are difficult to establish, such a large difference in size and age at maturity as between east and west Atlantic bluefin tuna is unlikely to be a result of density-dependent or envi- ronmental differences. Further, it seems unlikely that a sexually mature five- or six-year-old east Atlantic bluefin tuna would revert to immaturity upon migrating to the west Atlantic, and then remain immature for two or three years until finally spawn- ing at age eight. Another possible explanation for these results is that the size-at-maturity data on which this analy- 14- 12- 10- 0) u 0- 6- 4- 2- 1 Actual Catch < 135 kg 1 1 Predicted Catch <1 35 kg r-| r-| B r-| 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 Year Figure 3 Comparison of actual catch of bluefin tuna less than 135 kg in the Gulf of Mexico between 1980 and 1992 and catch predicted with a 2% annual eat-to-west transfer rate and the diffusion model. sis depends are incorrect. In fact. Clay (1990) crit- icized both the Baglin (1982) and Rodriguez-Roda (1967) studies, citing small sample sizes and inad- equate temporal and spatial coverage. As Clay (1990) pointed out, the Rodriguez-Roda (1967) study dealt with fish that were on their way to the spawning grounds and thus may have over-estimated the per- centage of small, mature fish in the total population. Although this is a valid criticism, it is worth noting that the collection of Rodriguez-Roda also contained immature fish, impl3dng that not all fish in his sample were on their way to the spawning grounds. Further, although our method clearly relies on the assumption that bluefin tuna of eastern origin first spawn at a smaller size than western fish, only a small proportion of age classes 4 through 7 need be mature for our results to prevail. We investigated the sensitivity of our model to a ten-fold reduction in the maturity parameter P (i.e. to 0.05 on age-4 and 0.1 on ages 5 through 7) and found that small blue- fin tuna would still be significantly rarer (chi-square test, P<0.005) in the Gulf than predicted under the diffusion model with a 2% east-to-west transfer rate, given our assumptions. Finally, for the purpose of this study, we have made the most conservative assumption, i.e. that any bluefin tuna (except larvae) found in the Gulf of Mexico is spawning. Thus, it is possible that some of the smaller bluefin tuna in the Gulf of Mexico land- ing records were not actually spawning fish, or were of west Atlantic origin, or both. 124 Fishery Bulletin 98(1) Alternative spawning grounds Our results might also be otabined if small migrant tuna of eastern origin spawn elsewhere in the west other than the Gulf of Mexico. A recently discovered concentration of medium and large tuna off the coast of North Carolina from January through April caused speculation that perhaps this concentration represents another spawning area. However, the lack of gonad development in a sample of seventeen fish (weighing between 65 and 183 kg) suggested that these fish were unlikely to spawn in the year they were captured and were probably immature (Belled). Several other workers have searched for evidence of bluefin tuna spawning in the west Atlantic. Mather et al. (1995) reported finding ripening small fish but no larvae. If the overlap hypothesis does pre- vail, then these potentially mature but nonspawning smaller fish may be eastern migrants that although capable of spawning, will return to the Mediterra- nean before actually doing so. McGowan and Rich- ards (1989) reported on the sporadic presence of larvae in the Gulf Stream as far north as North Caro- olina but concluded that most larvae found in the Gulf Stream were either advected out of the Gulf or spawned by tuna exiting the Gulf Furthermore, they stated that conditions are poor for larval devel- opment in the Gulf Stream and that the occasional occurrence of larvae there does not indicate an addi- tional spawning ground. On the matter of alternative spawning grounds, the National Research Council concludes that "extensive searching has detected only two spawning localities: the Gulf of Mexico and the Mediterranean Sea" (NRC, 1994, p. 18). Underreporting or low catchability ' Two other possibilities for the lack of small bluefin tuna in the Gulf of Mexico catch are that they are present in the Gulf of Mexico and are either being caught but not recorded, or are not being caught owing to a lack of appropriate fishing effort. To test for the first possibility, we acquired records of all bluefin tuna recorded by longline observers in the Gulf of Mexico during 1993-95. Of 31 bluefin tuna recorded by observers for which actual or estimated weights were recorded, all were greater than 135 kg. We also reviewed ICCAT data for the Japanese long- line fishery in the Gulf of Mexico for the period 1973 to 1981 and found that only 58 records out of 14,530 (0.4%) were for fish under 180 cm ( 135 kg). These data are particularly significant in light of the fact that there were no regulations concerning the reten- tion and sale of small bluefin tuna during this period as there have been in recent years. Therefore, the Japanese would have had no incentive to intention- ally misidentify or underreport small bluefin tuna. Mather etal. (1995), after reviewing longline catches in the Gulf and Caribbean prior to 1973, found only fish larger than 185 cm. They also reported very young bluefin tuna (less than 2 kg) in the Gulf of Mexico from July into November (Mather et al., 1995); fish presumably spawned a few months ear- lier. Similarly, Hisada and Suzuki (1982) presented length-frequency distributions of Japanese longline catches from the Gulf of Mexico which appear to show essentially no fish smaller than 200 cm. The possibility that small bluefin tuna are present in the Gulf but are not being caught cannot be com- pletely eliminated. However, there is a considerable accumulation of evidence that suggests that this is highly unlikely. For example, although there cur- rently is no directed fishery for either small or large bluefin tuna in the Gulf, there is a widespread, year- round yellowfin tuna (Thunnus albacares) longline fishery. This fishery targets yellowfin tuna of the same size as the small bluefin tuna of east Atlantic origin that we hypothesize would be present in the Gulf if the diffusion model is correct. This fishery does have a bycatch of bluefin tuna, none of which have ever been recorded by observers as less than 175 cm.^ Furthermore, longline operations in the northwest Atlantic do catch small bluefin tuna, indicating that they are potentially vulnerable to this gear. Cramer and Turner*^ reported length frequencies for observer data from the U.S. longline fishery in the northwest Atlantic from 1992 to 1995, showing that over 30% of fish hooked were less than 150 cm straight fork length (Fig. 4). Similarly, catch data from the Japa- nese northwest Atlantic longline fishery in the 1970s and 1980s show that the catch dominated by blue- fin tuna between 100 and 150 cm in several years (Fig. 8 in Hisada and Suzuki, 1982). Although fail- ure to catch a given species or size class in an area can never rule out its presence, given the extent and diversity of fishing activity in the Gulf, it is unlikely that any significant aggregation of small bluefin tuna ^ Belle, S. 1996. Biological sampling of bluefin tuna off Cape Hatteras, North Carolina. Final report to the New England Aquarium Corporation (NOAA requisition no. 43AANF503279I, Boston, MA. 12 p. ' Lee, D. 1996. National Marine Fisheries Service, Southeast Fisheries Science Center, 7.5 Virginia Beach Drive. Miami. FL 33149. Personal commun. " Cramer, J., and S. C. Turner 1996. Standardized catch rates for bluefin tuna, Thunnus thynnus, from the U.S. pelagic long- line fisherv in the northwest Atlantic. ICCAT working docu- ment SCRS/96/69. Nemerson et a\. Spawning site fidelity In Thunnus thynnus 125 30 n 25 20- 3 15 10 5- D n there would have been entirely missed. Finally, we tested the sensitivity of our results to a range of selec- tivities of longline gear set in the Gulf of Mexico. With selectivities on fish smaller than 135 kg rang- ing from to 1 (where the selec- tivity on fish greater than 135 kg is 1), the sensitivity analysis indi- cated that at selectivities greater than about 0.13, small fish were significantly less abundant (chi- square test, P<0.05 ) in the catch in the Gulf than would be expected given our assumptions and a 2% east-to-west transfer rate. Future research on this topic must seek to address both the annual rate of trans-Atlantic movement as well as the degree of philopatry exhibited by mi- grants to achieve a full under- standing of the population dynam- ics of east and west Atlantic bluefin tuna. Currently, there are studies underway to iden- tify nuclear and mitochondrial DNA markers that may have variations specific to east and west popu- lations (Graves et al., 1995). Examinations of otolith chemistry (microconstituent analysis) may provide information on stock differentiation and mixing rates, and researchers are currently deploying archi- val tags on bluefin tuna caught in the west Atlan- tic, primarily off Cape Hatteras, North Carolina. ^'i" These tags will record geolocation data and, if recov- ered, should yield a complete record of each fish's movement since its release. Finally, additional stud- ies on the maturation schedules of fish in the east and west are still needed. Clearly, it will take years before these studies yield sufficient information on transfer rates and philopatry to provide robust management advice. The depleted state of bluefin tuna populations world- wide, and in the west Atlantic particularly (Safina, 1993), make these issues of considerable practical importance. Until such time as these questions are answered definitively we believe that spawning site fidelity should be assumed and the stocks managed accordingly. Landed Catch Observed Catch ■n 70 90 110 130 150 170 190 210 230 250 Fork Length (cm) 270 290 310 330 350 Figure 4 Length-frequency distributions of bluefin tuna caught on longlines in the north- west Atlantic (landings. n=403) and measured by observers on longline vessels (observer. n = \\2) between 1992 and 1995 (see Footnote 7 in the main text). ^ Prince, E. 1996. National Marine Fisheries Service, South- east Fisheries Science Center, 75 Virginia Beach Drive, Miami. FL 33149. Personal commun. 10 Block, B. 1997. Hopkins Marine Station, Stanford Univer- sity, Oceanview Boulevard, Pacific Grove, CA93950-3094. Per- sonal commun. Acknowledgments We would like to thank Clay Porch and Dennis Lee of the Southeast Fisheries Science Center, National Marine Fisheries Service, for kindly providing assis- tance in acquiring the data sets that made this paper possible. Literature cited Baglin, R. E. 1982. Reproductive biology of western Atlantic bluefin tuna. Fish. Bull. 80:121-1.34. Calaprice, J. R. 1986. Chemical variability and stock variation in northern Atlantic bluefin tuna. ICCAT Coll. Vol. Sci. Pap. 24(2); 222-254. Clay, D. (ed.). 1990. Atlantic bluefin iur\a[Thunnus thynnus (L. )): a review. In World bluefin tuna meeting. May 2.5-31, La Jolla, CA, p. 89-180. Cooke, J. G., and K. Lankester. 1996. E.xamination of alternative stock distribution models for the interpretation of bluefin tuna ( Thunnus thynnus)iag- recovery data. ICCAT Coll. Vol. Sci. Pap. 45(21:135-138. Cort, J. L. 1991. Age and growth of the bluefin tuna, Thunnus thynnus (L.) of the northeast Atlantic. ICCAT Coll. Vol. Sci. Pap. 35(2):213-230. Curry, P. 1994. Obstinate nature: an ecology of individuals. Thoughts on reproductive behavior and biodiversity. Can. .J. Fish. Aquat. Sci. 51(7):1,664-1,673. 126 Fishery Bulletin 98(1) Dicenta, A., and C. Piccinetti. 1980. Comparison between the estimated reproductive stocks of bluefin tuna (Thunniis thynnus) of the Gulf of Mexico and western Mediterranean. ICCAT Coll. Vol. Sci. Pap. 9:442^48. Graves. J. E., J. R. Gold. B. Ely, J. M. Quattro, C. Woodley, and J. M. Dean. 1995. Population genetic structure of bluefin tuna in the north Atlantic ocean. I. Identification of variable genetic markers. ICCAT Coll. Vol. Sci. Pap. 45(2):155-157. Greenwood, P. J., and P. H. Harvey. 1982. The natal and breeding dispersal of birds. Annu. Rev. Ecol. Syst. 13:1-21. Hisada, K., and Z. Suzuki. 1982. Catch, fishing effort and length composition of the Atlantic bluefin tuna caught by Japanese longline fishery. ICCAT Coll. Vol. Sci. Pap. 17:307-314. Mather, F. J., J. M. Mason, and A. C. Jones. 1995. Historical document: life history and fisheries of Atlantic bluefin tuna. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC 370, 165 p. McGowan, M. F., and W. J. Richards. 1989. Bluefin tuna, Thunnus thynnus, larvae in the Gulf stream off of the southeastern United States: satellite and shipboard observations of their environment. Fish. Bull. 87:615-631. National Research Council (NRC). 1994. An assessment ofAtlantic bluefin tuna. National Aca- demy Press, Washington. D.C., 148 p. Policansky, D. 1993. Fishing as a cause of evolution in fishes. In T. K. Stokes, J. M. McGlade, and R. Law (eds.). The exploitation of evolving resources, p. 2-18. Springer- Verlag, Berlin. Porch, C. E., V. R. Restrepo, S. C. Turner, and G. P. Scott. 1995. Virtual population analyses ofAtlantic bluefin tuna incorporating movement and tagging data. ICCAT Coll. Vol. Sci. Pap. 44(11:183-190. Powers, J., and J. Cramer. 1996. An exploration of the nature of bluefin tuna mixing. ICCAT Coll. Vol. Sci. Pap. 45(2):173-181. Punt, A. E., and D. S. Butterworth. 1995. Use of tagging data within a VPA formalism to esti- mate migration rates of bluefin tuna across the north Atlan- tic. ICCAT Coir Vol. Sci. Pap. 44(1): 166-182. Reznick, D. N. 1993. Norms of reaction in fishes. In T K. Stokes, J. M. McGlade, and R. Law (eds.). The exploitation of evolving resources, p. 72-90. Springer- Verlag, Berlin. Richards, W. J. 1976. Spawning of bluefin tuna [Thunnus thynnus) in the Atlantic Ocean and adjacent seas. ICCAT Coll. Vol. Sci. Pap. 5(2):267-278. Rodriguez-Roda, J. 1967. El atun, Thunnus thynnus (L.) del sur de Espana, en la eampana almadrabera del ano 1966. Invest. Pesq. 31(2):349-359. Safina, C. 1993. Bluefin tuna in the west Atlantic: negligent manage- ment and the making of an endangered species. Conserv. Biol. 7(2):229-233. Trippel, E. A. 1995. Age at maturity as a stress indicator in fisheries. Bio- science 45(11 ):759-771. Turner, S. C. 1994. A review of recent information on size at age and the calculation of age from size for Atlantic bluefin tuna. ICCAT Coll. Vol. Sci. Pap. 42( 1 ):97-99. Turner, S. C, and J. E. Powers. 1995. Review of information related to Atlantic bluefin tuna east-west movement. ICCAT Col. Vol. Sci. Pap. 44(1): 191-197. Turner, S. C, V. R. Restrepo, and A. M. Eklund. 1991. A review of the growth ofAtlantic bluefin tuna, Thun- nus thynnus. ICCAT Coll. Vol. Sci. Pap. 35(2):271-293. 127 Abstract.— Genetic information per- taining to stock structure in red drum ISciaenops ocellatus) is equivocal, com- plicating attempts to develop sound fishery management and stock enhance- ment plans. In this study, genetic stock structure was examined by using mito- chondrial DNA(mtDNA) control region sequences of 209 individual red drum from six locations in the Gulf of Mexico and five locations in the nearshore Atlantic Ocean off the southeastern United States. Eighty-one polymorphic sites within a 369 base-pair portion of the control region defined 134 dif- ferent haplotypes which differed by up to 26 nucleotide substitutions. Red drum showed high average within-sam- ple haplotype (0.98) and nucleotide (0.030) diversities. Sequence diver- gences between pairs of haplotypes ranged from 0.27'7, to 7.06% (J=3.17% ). Cluster analysis of haplotypes revealed very little phylogeographic structure among mtDNA lineages. However, a neighbor-joining tree based on nucleo- tide divergence between pairs of sam- ples showed cohesion among Atlantic samples and, to a lesser degree, among Gulf samples. In contrast to a prior study, we found no evidence that red drum in Mosquito Lagoon, Florida, con- stitute a self-contained, reproductively isolated population. Hierarchical anal- ysis of molecular variance supported the hypothesis that red drum are subdi- vided into two weakly diverged popula- tions with a genetic transition in south Florida between Sarasota Bay and Mos- quito Lagoon. This area forms a zone of differentiation between two geneti- cally semi-isolated populations between which the structuring of heterogeneity differs from that under the assump- tion of panmixia. In addition, the analy- sis of molecular variance also indicated that red drum from Apalachicola Bay are genetically divergent from all other samples. The Atlantic and Gulf red drum populations are likely to respond independently to harvest regulations; these fisheries should continue to be managed separately. Additional subdi- vision of the Gulf stock between pen- insular Florida and the northern and western Gulf may also be warranted. An analysis of genetic population structure in red drum, Sciaenops ocellatus, based on mtDNA control region sequences Seifu Seyoum Michael D. Tringali Theresa M. Bert Department of Environmental Protection Florida Marine Research Institute 100 Eighth Avenue SE St. Petersburg, Florida 33701-5095 E-mail address (for 5 Seyoum, contact autfior) Seyoum_S'a'epic7dep state fl us Doug McElroy Department of Biology Western Kentucky University Bowling Green, Kentucky 42101 Rod Stokes Department of Marine Science University of South Florida 140 7th Avenue S, St. Petersburg, Flonda 33701-5095 Manuscript accepted 5 March 1999. Fish. Bull. 98:127-138 (2000). Red drum (Sciaenops ocellatus) is a pelagic marine fish that is dis- tributed over a large geographic range that extends throughout the northern Gulf of Mexico and along the Atlantic coast of the southeast- ern United States to Cape Cod, Massachusetts (Ross et al., 1983). Juveniles grow rapidly in estuarine nurseries and reach reproductive maturity by age 4. At this age they join large schools of highly dispers- ing adults and for the remainder of their approximately 35-year life span (Murphy and Taylor, 1990), maintain a pelagic existence except to spawn during annual congrega- tions at the mouths of bays and estuaries. The census size of the breeding population in the Gulf of Mexico has been estimated to be greater than 7 million individuals (Nichols^ ). Abundance in the Atlan- tic is thought to be of a similar mag- nitude (Gold et al., 1993). Red drum supports highly valu- able commercial and recreational fisheries throughout its range (Mercer, 1984). Fishing pressure is directed principally on subadult year classes (ages 2-4). A high rate of annual mortality among some cohorts (Murphy and Taylor, 1990) and an overall decline in abundance and recruitment during the 1980s (Goodyear^) have led to concerns regarding the status of red drum spawning stocks. Because there have been no prolonged offshore fisheries for adult red drum, biological and fishery-dependent data pertaining to their spawning stocks have been limited. Therefore, fishery managers have had to rely principally on vir- tual population analysis and simi- ' Nichols, S. 1988. An estimate of the size of the red drum spawning stock using mark/ recapture. Southeast Fisheries Center, Natl. Mar. Fish. Serv., Pascagoula, MS. - Goodyear, C. P. 1989. Status of the red drum stocks of the Gulf of Mexico. Report for 1989. Contract no. CRD 88/89-14. Coastal Resources Div, Miami Laboratory, Southeast Fisheries Center, Natl. Mar. Fish. Serv., Miami, FL. 128 Fishery Bulletin 98(1) lar analyses (e.g. Vaughan-^) to manage the red drum fishery. Knowledge of stock structure in red drum, i.e. the geographic relation between spawning and recruitment, is needed to facilitate the management of this fishery. Molecular studies have been used to define appro- priate geographic scales for monitoring and man- aging exploited animal populations (Moritz, 1994), including marine fishes (e.g. Bentzen et al., 1996; Graves, 1996; Tringali and Bert, 1996). Although several population genetic studies have been con- ducted on red drum, the existing data are equivo- cal. On the basis of significant differences in allele frequencies at a single allozyme locus, Bohlmeyer and Gold ( 1991) concluded that red drum are weakly subdivided between the Atlantic Ocean (North Car- olina and South Carolina) and the Gulf of Mexico. However, data from other allozyme surveys (Ramsey and Wakeman, 1987; Campton'*) did not permit the rejection of the null hypothesis that red drum form a single panmictic gene pool. In contrast, on the basis of small but statistically significant differences in the frequencies of several composite mitochondrial DNA (mtDNA) haplotypes between red drum collected from North Carolina and South Carolina waters and red drum collected from the Gulf of Mexico, Gold and Richardson (1991) and Gold et al. (1993, 1994) reas- serted that red drum are weakly subdivided between these regions, ostensibly along the south Florida coast. In addition, Gold et al. (1993) reported a pat- tern of mtDNA differentiation in Gulf red drum con- sistent with the isolation by distance model (Wright, 1943) for samples ranging from Florida to Texas. However, samples from a significant portion of the species range, including locations near the putative Atlantic-Gulf division (i.e. the eastern Florida sea- board), were not assayed in any of the above studies. An equally tenable but untested hjrpothesis is that isolation by distance occurs over the entire range of the species, perhaps in the absence of a genetic break at a particular geographic location. Localized population subdivision in red drum has also been postulated. From comparisons with sam- ples from North Carolina and South Carolina waters and samples from the Gulf of Mexico, Gold and Rich- ardson (1994) proposed that red drum inhabiting Mosquito Lagoon, Florida, form a genetically distinct population. Red drum in Mosquito Lagoon report- edly have a life history uncharacteristic of other red drum. Adults occupy this coastal lagoon throughout the year and may complete their life cycle within the lagoon (Johnson and Funicelli, 1991). Adult red drum also occur throughout the year in other coastal lagoons adjacent to Mosquito Lagoon (e.g. Banana, Indian, and Halifax Rivers), but these have not been surveyed genetically. Rather than forming a self-con- tained population, red drum from Mosquito Lagoon may belong to a larger subpopulation occupying Flor- ida Atlantic waters. Owing to the perceived decline of red drum abun- dance in the 1980s, state agencies in Alabama, Flor- ida, South Carolina, and Texas studied the feasibility of stock enhancement as a means of supplementing wild populations. Hatcheries in Florida, South Caro- lina, and Texas currently employ stocking on a large scale (McEachron et al., 1995; FDEP"^). Hatchery programs potentially affect the gene pools of indige- nous red drum populations by way of an inappropri- ate introduction of non-native individuals (Hindar et al., 1991) and by hatchery-induced inbreeding effects (Tringali and Bert, 1998). For example, because brood- stocks for large-scale stock enhancement programs along the Atlantic seaboard have been obtained from Mosquito Lagoon and nearby estuaries (Halstead^), there is a potential for artificial genetic exchange between putatively separate gene pools (e.g. those of Mosquito Lagoon and the Carolinas). State and regional fishery managers (Vaughan'^; FDEP'') and hatchery managers (FDEPM have adopted the stock structure scenario proposed by Gold et al. (1993) and Gold and Richardson (1994) in which red drum are divided into Gulf of Mexico and Atlantic populations, and those fish in Mosquito Lagoon comprise a unique, self-contained Atlantic subpopulation. However, several important ques- tions regarding the genetic structure of red drum remain unanswered; each has serious implications for fishery management and stock enhancement pro- grams. First, are red drum populations in the Gulf of Mexico and Atlantic really subdivided or do the observed genetic differences solely reflect isolation by distance over the range of the species? Second, if a genetic break does exist somewhere between the Gulf and the coast of the Carolinas, where is it? ■* Vaughan, D. S. 1995. Statu.s of thi- ri-d drum stock on the Atlan- tic coast: stock a.sse.ssment report for 1995. Southeast Fisheries Science Center, Natl. Mar Fish. Serv., Beaufort, NC. 50 p. '' Campton, D. E. 1992. Gene flow estimation and population structure of red drum iSciaenops ocellatiis) in Florida. Final report, cooperative agreement no. 14-16-009-1522, National Fisheries Research Center, U.S. Fish and Wildlife Serv.. Gaine.s- ville, FL. ' FDEP (Florida Department of Environmental Protection). 1993. A stock assessment of red drum iSciaein>ps ocellatiis) in Florida. Florida Marine Research Inst., Dep. Natural Resources, 100 Eighth Ave. SE, St. Petersburg, FL, 24 p. •^ Halstead, B. 1997. Stock Enhancement Research Facility, Florida Department of Environmental Protection, 14495 Harlee Road, Port Manatee, FL 34221. Personal commun. ' FDEP (Florida Department of Environmental Protection l. 1993. Marine fi.sh stock enhancement and hatchery executive sum- mary. Report to the legislature. FDEP, St. Petersburg, FL, 17 p. Seyoum et a\ : Genetic population structure in Saaenops ocellatus 129 Atlantic Ocean Gulf of Mexico FB Figure 1 Sampling sites for red drum in waters off the southeastern United States. Florida: AP = Apalachicola Bay; TB = Tampa Bay; OF = offshore, Tampa Bay; SA = Sarasota Bay; CH = Charlotte Harbor; FB = Florida Bay; IR = Indian River; ML = Mosquito Lagoon; PO = Ponce de Leon inlet; TP = Tomoka Basin. South Carolina: SC = Charleston Harbor. Third, are red drum in Mosquito Lagoon reproduc- tively and genetically distinct or do they belong to a larger and heretofore unsampled east Florida pop- ulation? To examine these questions, we obtained sequence data from the rapidly mutating mtDNA control region. Sequencing the control region has proven useful in intraspecific phylogeographic and population genetic studies of fishes (e.g. Fajen and Breden, 1992; Brown et al., 1993; Stepien, 1995; Stabile et al., 1996). We employed sampling and analytical regimes designed to test the various com- peting hypotheses of red drum population structure. In addition, by gathering baseline data for mtDNA control region diversity in red drum populations, we explored the potential for using the control region as a marker to assess and monitor ongoing stocking programs for wild red drum populations. Materials and methods Sample collection and DNA purification Samples of red drum were collected with hook-and- line gear, trammel nets, and purse seines from riv- erine, estuarine, and offshore waters of the South Carolina coast (one location) and the east coast of Florida (four locations, sampled prior to stock enhancement activities), referred to collectively as the Atlantic samples; and the west coast of Florida (six locations), referred to collectively as the Gulf sam- ples (Fig. 1). All specimens were collected between February 1992 and February 1997. Somatic muscle and liver tissue were dissected from each individual. Total length (range 280-1070 mm) of each individual was recorded prior to dissection. Tissues were frozen in liquid nitrogen and stored at -80°C in the labora- tory until processing. Approximately 100—400 milligrams of muscle or liver tissue were digested in 900 microliters of lysis buffer (O.IM Tris, pH 8.0, 0.05M ethylenedia- minetetraacetid acid (EDTA), 0.2M NaCl, 1% weight by volume of sodium dodecyl sulfate (SDS), contain- ing 1-2 milligrams of proteinase K) with moderate shaking for 3-5 h at room temperature. Following the addition of 150 microliters of chilled 8M potas- sium acetate, the SDS and cellular debris were precipitated for 30 min at A'C and removed by centrifugation. Total genomic DNA was purified by phenol/chloroform extraction ( Sambrook et al., 1989). 130 Fishery Bulletin 98(1) The DNA was concentrated by isopropanol precipita- tion, resuspended in 75 microliters of sterile water, and stored at -20°C. mtDNA control region sequencing Initially, we used the polymerase chain reaction (PCR; Saiki et al., 1988) and published primer sequences L15926 (Kocher et al., 1989) and H16498 (Meyer et al., 1990) to amphfy a portion of the mtDNA con- trol region of red drum. Double-stranded PCR was performed with Perkin Elmer AmpliTaq in a 50-pL reaction volume for 32 cycles in a DeltaCycler II System (Ericomp Inc., San Diego, CA), according to the methods described by Kocher et al. (1989). We amplified a 455-base-pair (bp) fragment of the con- trol region for several individuals of red drum. How- ever, because direct sequencing of the amplicon with the same PCR primers in the sequencing reaction yielded unsatisfactory results, we used a process of cloning and sequencing to design specific primers for red drum. The 455-bp amplicon was cloned in pBluescript by TA cloning (Marchuk et al., 1990). Following denatur- ation of the plasmid DNA(Hattori and Sakaki, 1986), sequencing was done from both directions by the dide- oxy termination method (Sanger et al., 1977) by using Sequenase version 2.0 (U.S. Biochemicals, Cleveland, OH) and [a-S^S] dATP (Dupont Biotechnology Sys- tems, Wilmington, DE). Products of the sequencing reactions were resolved in 6% polyacrylamide/7-M urea gels that were vacuum dried at 80°C and auto- radiogi'aphed with Kodak X-Omat AR film. We used the ESEE program (Cabot and Beckenbach, 1989) to align sequences. From these sequences, (Genbank accession no. AF054671), highly specific internal primers were designed for the control region of red drum. These primers partially overlapped the initial primers and were designated L15943 (5'-GTA AACCGGATGTCGGGGGTTAG-3') and H16484 (5'-GGAACCAGATACCAGGAATAGTTCA -3' ). We used these custom primers to amplify a por- tion of the control region for 209 individuals in SO-pL reaction volumes. The PCR products were run on 1.2% low-EEO (Fisher Scientific, Norcross, GA) aga- rose gel during electrophoresis. The resulting bands were excised and then purified with GeneClean (Bio 101, La Jolla, CA). Double-stranded sequencing was conducted as described by O'Foighil et al. ( 1996). Data analyses Base composition, number of transitions (TSs), and number of transversions (TVs) were determined by using MEGA 1.01 (Kumar et al.-, 1993). Further analysis of base substitutions was conducted as in Brown and Clegg (1983). Each different haplotype was assigned a number, and the distribution of the dif- ferent haplotypes was determined for each sample. We used MEGA to generate a pairwise matrix of sequence divergence values between pairs of haplo- types and to construct an unrooted neighbor-joining tree; 200 replicates were used to estimate bootstrap values for the nodes. Sequence divergences were computed by using the pairwise-deletion option in MEGA; this distance estimator excludes sites at which indels occur on a pairwise basis. Haplotype and nucleotide diversity within samples and nucleo- tide divergence (D) between pairs of samples were estimated according to Nei and Tajima (1981) and Nei ( 1987 ) by using the DA option of REAP 4.0 ( McEl- roy et al., 1992). The nucleotide divergence values were clustered by using the NJTREE program (Jin and Ferguson, 1990) based on the neighbor-joining method of Saitou and Nei ( 1987 ). Geographic structuring of molecular variance among samples was examined by using the matrix of sequence divergences between all pairs of haplo- types in AMOVA 1.55 (Excoffier et al., 1992). In this analysis, the haplotype correlations ((^statistics) and their variance components were estimated in a hier- archical fashion: between regions, among samples within a region, and among individuals within sam- ples. Statistical significances of values were com- puted by performing randomization tests with 500 replicates. Goldet al. ( 1993) concluded that red drum was subdivided into Atlantic and Gulf of Mexico pop- ulations. To determine the validity of this conclusion, we examined the spatial partitioning of molecular variance as follows. The between-region component of variance and (J^.^ was first calculated for red dium samples divided into Atlantic (SC, TP, PO, ML, and IR) and Gulf (FB, CH, SA, OF, TB, and AP) regions. The compositions of the two groups were then adjusted by sequentially adding Atlantic sam- ples to the Gulf group and then sequentially adding Gulf samples to the Atlantic group. After each addi- tif)n. the apportioning of molecular variance between the resulting groups was recalculated. For example, IR ( the Atlantic sample closest to the Gulf) was added to the Gulf group and tested against the remaining Atlantic samples (ML, PO, TP, and SC). ML (the second closest sample) was then added to the Gulf- plus-IR gi'oup; that grouping was then tested against PO, TP, and SC. This process was repeated until only a single sample remained in one group. Finally, to test for an association between interpopu- lation D values and geogi-aphic distance (isolation-by- distance), we performed the Mantel test (BIOMstat, version 3.0; Rohlf and Slice, 1995) for samples gi-ouped Seyoum et al. Genetic population structure in Saaenops ocellatus 131 by region (Atlantic or Gulf) and for all samples combined. The statistical significance of the association was tested by random permutation analysis by using 500 replicates (Sokal andRohlf, 1995). Results -AP FB ■-OF c CH TB ■SA Gulf r-SC 54 We analyzed sequence data from a 369-bp portion of the mtDNA control region for 209 red drum. A total of 81 poly- morphic sites were observed among all individuals. Of these, 67 sites had single- state, 11 had double-state, and 3 had triple-state transfor- mations, totaling 98 polymor- phisms and including two indels. The first indel consisted of an insertion of a pyrimidine (T or C) at position 160 and occurred in 10 individuals; the second indel consisted of a deletion of a purine (A) at position 210 and occurred in one of these individuals (Table 1). Seventy-three of the substitutions were TSs and 22 were TVs. As with control regions of other fishes (Stepien, 1995), the relative frequencies of the four nucleotide bases differed; adenine was the most prevalent (399^), fol- lowed by thymine (27*^), cytosine(23'^), and guanine (11%). The TSrTV ratio was 3.4:1 and was similar to ratios reported for marine and freshwater fishes (Stepien, 1995; but see Brown et al., 1993). We observed 134 different haplotypes in the 209 individuals sequenced (Table 1). Sequences of these haplotypes have been deposited in GenBank under the accession numbers AF054672-F054805. Twenty- nine haplotypes were shared by more than one indi- vidual. Two haplotypes, no. 56 and no. 83, were shared by nine and nineteen individuals, respectively, which were widely dispersed among seven samples. Twenty-six haplotypes occurred infrequently, 25 in two to five individuals scattered among two to five samples, and one in two members of a single sample. Of the 19 individuals with haplotype no. 83, 17 were from the Atlantic and two from the Gulf The per- centage of different haplotypes in any one sample varied from im to lOO'/f {x=Sl'7c ). Haplotype diver- sity within samples ranged from 0.95 to 1.00 (x=0.98, SE=0.00) and nucleotide diversity ranged from 0.025 to 0.037 (x=0.030, SE=0.003). i-IR 4 ML PO •-TP Atlantic 1 0,36 0.18 000 Patristic distance 100 Figure 2 Neighbor-joining tree based on mtDNA nucleotide divergence between samples of red drum from Florida and South Carolina nearshore waters. Sample location and abbrevi- ates are defined in Figure 1. Percent sequence divergences between pairs of dif- ferent haplotypes ranged from 0.3% to 7.1% (x- 3.2%; SE=0.oi7). Between any two different haplo- types, the number of nucleotide differences varied from one to 26 (x=\2). The topology of the unrooted tree neighbor-joining (not shown) revealed that the 134 haplotypes were not phylogeographically structured. Haplotypes observed in Gulf and Atlantic samples were scattered throughout the tree; with the exception of two terminal groupings, nodes on the tree had low sta- tistical support. Internal branch lengths of the tree were generally short; however, one interior branch was relatively long, and it defined the only well-supported major clade (bootstrap value=85). This clade consisted of 23 haplotypes, including the 10 haplotypes that had the insertion at position 160. The 10 insertion-bearing haplotypes were found in nine Atlantic individuals but in only one Gulf individual. The D values between pairs of samples ranged from -0.08% to 0.10%. In the neighbor-joining clus- ter analysis, cohesion of the samples within geo- graphic regions was generally observed (Fig. 2). All Atlantic samples formed a distinct clade which was separated by the longest branch of the tree from the clade formed by the Gulf samples. Less cohesion was observed among the Gulf samples, although the geo- graphically proximal SA, TB. and CH samples clus- tered closely together. 132 Fishery Bulletin 98(1) Table 1 Distribution of red dnam mitochondrial DNA control region haplotypes from 11 locations. Abbreviations for sample locations are defined in Figure 1. Gulf ^ Atlantic Haplotype AP TB OF SA CH FB IR ML PO TP SC 1 1 2 3 1 3 1 111 4 1 5 1 6 1 1 7 1111 1 8 1 9 1 10 1 11 1 12 1 13 1 14 1 15 1 16 1 1 17 1 18 1 : ' 19 1 20 1 21' 1 22 1 23 1 24 1 25 2 26 1 27 1 28 1 29 1 30 1 31 1 32 1 1 33 1 1 34 1 35 1 36 1 37 1 1 38 1 39 1 40 1 41 1 42 1 43 1 44 1 45 2 46 1 47 1 48 1 49 1 50 1 51 1 52 1 53 12 11 54 1 continued Seyoum et a\ Genetic population structure in Saaenops ocellatus 133 Table 1 (continued) Gulf Atlantic Haplotype AP TB OF SA CH FB IR ML PO TP SC 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84' 85 86 87 88 89' 90 91 92 93' 94 95' 96 97 98 99 100 101' 102 103 104 105 106 107 108 109 110 continued 134 Fishery Bulletin 98(1) Table 1 (continued) Gulf Atlantic Haplotype AP TB OF SA CH FB IR ML PO TP SC 111 112 113 114 115 116 117''' 118 119 120 121 122' 123 124 125 126 127 128 129 130' 131 132 133 134 Total 20 22 14 20 20 17 21 22 23 22 ' Haplotypes with an insertion at position 160. - Haplotype with an insertion at position 210. The D value between the ML sample and the remaining pooled Atlantic samples was -0.05%. In contrast, the D value between the ML sample and the Gulf samples ranged from -0.02% to 0.076%, The analysis of molecular variance ( AMOVA) for all samples yielded a (^^.j- value of -0.001, indicating that no significant heterogeneity was detected between any two samples. For the geographic analysis, in which samples were divided into shifting regional sub- sets, the variance components among samples within groups and among individuals within samples were not significant for any grouping. Significant values of (|)f.j.were observed in five of the 10 groupings (Table 2). In four of the five significant groupings, the division occurred in peninsular south Florida. Overall, results of the AMOVA suggest that a genetic transition occurs in red drum along the Florida coast between Sara- sota Bay in the Gulf and Mosquito Lagoon in the Atlantic. However, the highest (/)(^.y, value and between- group variance component were observed when sam- ples were grouped according to their actual Atlantic and Gulf locations (Table 2, grouping 1). The signifi- cant 0f,j. value for the AP sample versus all other sam- ples suggests that an additional genetic discontinuity occurs in Gulf waters off northwest Florida. In the Mantel test between interpopulational D and geographic distance, no association was observed among the Atlantic samples (P=0.20) or among the Gulf samples (P=0.053). However, a significant association was observed for all red drum samples (P<0.01 ), reflecting the genetic transition that occurs in south Florida. Discussion Genetic population structure Because there are few absolute barriers to gene flow in the ocean, it is generally expected that marine spe- cies with continuous distributions, large populations, and high levels of larval and adult dispersal should have very little intraspecific population structure over large geographic areas (Avise, 1987; Palumbi, 1992). Seyoum et al : Genetic population structure in Saaenops ocellatus 135 Table 2 Geographic analysis of molecular variance in the mtDNA control region of red drum. The table lists the i^,..,. values and between- group variance components for the ten possible geogiaphical groupings and the probability P of find mg a more extreme variance component by chance (500 permutations). I n each group, the letter G represents > samples from the Gulf of Mexico (Gj =AP, G,,=TB, G,=OF. Gj= =SA. G,=CH, Gk=FBi and the letter A represents samples from the Atlantic Ocean (A| = =IR, A2= ML, A3 =PO, A4=TP, A5=SC ). Abbreviations for sample locations are defined in Figure 1. Variance between Grouping First group Second group groups (%) 0,.., P 1 G1G2G3G4G5G6 AjA^AjA^Aj 1.96 0.020 <0.002 2 GiGoG-jG^GjOg Aj A2A3A4A5 1.40 0.014 0.012 3 Gfi.fi,Gfi,G,K,A., ■'^3-'^4A5 0.39 0.004 NS 4 Gfifi,Gfifi,^,\.,k, A4A5 -0.17 -0.002 NS 5 Gfi,fi,Gfifi,k,K\,K, A5 -0.54 -0.005 NS 6 '^1^2^3^4*-'5 GeAiA^AjA^A^ 1.48 0.015 0.004 7 G1G2G3G4 G5G6A1A2A3A4A5 0.78 0.008 0.044 8 G1G2G3 G^GgGgAiA^AiA^A, 0.73 0.007 NS 9 Gfi, G3G4G,G,AjA,A3A,A4 0.74 0.007 NS 10 G: G2G3G4G,G,A,A.A3A4A, 1.50 0.015 <0.002 Red drum is a pelagic marine fish with these demo- graphic and Hfe history characteristics. However, although some genetic exchange may occur between red drum from distant locations, populations differ from the expectation of genetic homogeneity. Hierar- chical analysis of the structuring of genetic variance supports the hypothesis that the species is weakly subdivided between the Atlantic Ocean and the Gulf of Mexico. The existence of the subdivision is also supported by cluster analysis of sequence divergence values. The geographic coverage of our samples, par- ticularly the inclusion of samples from Florida's east- ern seaboard, allowed us to infer that the genetic break separating these two populations occurs in south Florida. Another genetic discontinuity appar- ently occurs in Gulf waters off northwest Florida. The Atlantic-Gulf subdivision in red drum may result from a combination of extrinsic and behavioral factors. Gold et al. ( 1993, 1994) summarized a number of potentially important oceanographic and geogi-aphic factors. Because the southernmost portion of the east Florida shelf is extremely nairow and provides little of the neritic habitat (Jones et al., 1985) generally occu- pied by adult red drum, it may represent a significant barrier to adult migration. Biotic factors largely pre- clude large-scale passive dispersal of eggs and larvae (Peters and McMichael, 1987), and widespread disper- sal at the juvenile stage is rare (Murphy and Taylor^). If partitioning of genetic variation in red drum results from adult migration or vagrant movement between spawning locations, it is not evident from studies of fish movement. Although very few tagged adult red drum have been recaptured after spending significant periods at large, the evidence suggests that some are highly mobile and may disperse to distances of up to 320 km (Woodward and Nicholson, 1997; Crabtree^). However, movement of red drum between the Atlan- tic and Gulf regions has not been documented. In apparent contrast to the mark-recapture data, recent ultrasonic tracking studies of adults provide limited evidence for spawning fidelity to certain Atlantic estu- aries over two-to-three-year periods, and potentially longer (Nicholson and Jordan'"). Overall, the available movement data for adult red drum are not adequate to draw conclusions relating to regional recruitment processes and patterns of dispersal. Nevertheless, it is clear from the genetic data that reproductive exchange between spawning populations in the Atlantic and Gulf regions is limited. Because red drum from Mosquito Lagoon were used to produce hatchery populations for at least two stock enhancement programs along the Atlan- tic seaboard, it was important to determine whether the spawning aggregation within that system repre- sented a self-contained, genetically divergent popu- *• Murphy, M. D., and R. G. Taylor 1989. Tag/recapture and age validation of red drum in Florida. Final report, NOAA grant NA86-WC-H-06136, National Marine Fisheries Service, Pascagoula, MS, 27 p. ■'Crabtree, R. 1997. Florida Marme Research Institute. Department of Environmental Protection, St. Petersburg, FL. Unpublished data. '"Nicholson, N., and S. R. Jordan. 1994. Biotelemetry study of red drum in Georgia, November 1989^June 1993. Coastal Resources Division, Georgia Department of Natural Resources, Brunswick, GA, 65 p. 136 Fishery Bulletin 98(1) lation. Contrary to the study of Gold et al. (1994), we found no evidence to support the hjrpothesis that red drum in Mosquito Lagoon are reproductively iso- lated from other Atlantic red drum. In our survey of the mtDNA control region, there was no nucleo- tide divergence between Mosquito Lagoon and other Atlantic samples, whereas divergence values between the Mosquito Lagoon and the Gulf samples were among the highest. Furthermore, other investigators also found no significant differences at the allozyme loci that putatively distinguish Mosquito Lagoon red drum from other red drum (Campton*; Crawford and Bert'M. This lack of difference suggests that the allo- zjrme frequency differences observed by Gold and Rich- ardson ( 1994) may not be temporally stable. A lack of samples geographically proximal to Mosquito Lagoon may have also influenced the outcome of that study. Considering all the evidence, it seems more likely that red drum from Mosquito Lagoon belong to the larger genetic population occupying nearshore Atlan- tic waters of the southern United States. Although it is generally better to obtain hatchery broodstock from locations within or near the intended release site (Utter, 1998), the use of Mosquito Lagoon red drum as a source of broodstock in Atlantic coast stock enhancement programs should not produce a nega- tive genetic impact on wild red drum. Implications for fishery and hatchery management Because red drum support valuable fisheries through- out the Gulf of Mexico and Atlantic seaboard, the species is of special concern to state and regional fish- ery management agencies. Our results support the hypothesis that a genetic transition in red drum pop- ulation structure occurs in south Florida. In theory, it requires the regular exchange of only a few indi- viduals between breeding populations to homoge- nize their genetic composition (Slatkin, 1987). Thus, genetic exchange between Atlantic and Gulf stocks by any recruitment process must be sufficiently low to allow genetic differences to accumulate or, if the dif- ferences reflect a historical disassociation, for them to be maintained. The two red drum stocks can best be described genetically as demes, separate semi-iso- lated groups between which the structuring of het- erogeneity differs from the assumption of panmixia (Hartl and Clark, 1989). Therefore, the Atlantic and Gulf stocks are likely to respond independently to harvest regulations and these fisheries should con- tinue to be managed separately. " Crawford, C, and Bert, T. 1997. Florida Marine Research Institute, Department of Environmental Protection, St. Peters- burg, FL. Unpublished data. Gold et al. (1993) observed a pattern of isolation by distance in the distribution of mtDNA haplotypes among Gulf samples that ranged from the south- east coast of Texas to the southwest coast of Florida. Although we did ^ot observe a similar pattern for Gulf samples ranging from Apalachicola Bay to Flor- ida Bay, the probability value for the Mantel coeffi- cient was nearly significant at the 0.05 level and the AMOVA value for the Apalachicola sample versus all other samples was highly significant. This indicates that the minimum geographic scale at which the isolation-by-distance mechanism operates is greater than the distance between Apalachicola Bay and Florida Bay (approximately 670 km) or that genetic discontinuity also exists between Florida Gulf red drum and red drum inhabiting the northern and western Gulf of Mexico. Accordingly, cooperative man- agement of the Gulf fishery on a regional basis is appropriate. No pattern of isolation-by-distance was evident for red drum along the southern Atlantic sea- board; the fishery between South Carolina and south- east Florida should be managed as a single unit. Our principal objective for undertaking this study was to improve upon available genetic information relating to red drum population structure for fish- ery management purposes. Our most informative statistical tools were those that assessed relation- ships among the samples. Genotype frequency differ- ences accumulate quickly in subdivided populations compared with the rate at which distinct phyletic lineages emerge and sort geographically. Moreover, mitochondrial DNA restriction fragment length poly- morphism and sequence data for marine populations are typically characterized by haplotype distributions which consist of a few numerically and geographi- cally prevalent haplotypes and many rare, geograph- ically restricted haplotypes that may be important with respect to population structure. Therefore, as our results and the results of Tringali and Bert ( 1996) demonstrate, genetically-based management units (sensu Moritz, 1994), especially in marine fishery stocks, may be more easily identified by applying pop- ulation-level analyses that take full advantage of both differences in genotype frequencies among samples and phylogenetic relatedness of individual genotypes. Statistical tests of association when applied to skewed haplotype distributions often lack the power to detect the low levels of population divergence that may char- acterize marine populations. Moreover, these tests ignore the interrelatedness of haplotypes in terms of sequence similarity or difference. The two principal analytical methods we employed, clustering of inters- ample nucleotide divergence values and AMOVA, are based on both the occurrence of haplotypes at particu- lar locations and their sequence similarity to other Seyoum et al : Genetic population structure in Saaenops ocellatus 137 haplotypes at those locations. This approach may be particularly important in studies of pelagic marine species because, like red drum, genetic divergence separating populations is often buried within a high background of overall genetic diversity. Finally, Moritz (1994) described the potential of using mtDNA as a marker for evaluating the suc- cess of stock enhancement programs. For this appli- cation, the mtDNA haplotypes borne by hatchery fish should be sufficiently rare in wild populations. The portion of control region we examined provides an excellent source of naturally occurring genetic markers. Because the percentage of wild red drum individuals with different haplotypes in any given sample averaged 87% and ranged up to 1001, a genetic monitoring program using this mtDNA frag- ment to track haplotypes borne by hatchery-released red drum after release should allow for assessment of the survival and reproductive output of these fish in the natural environment. In addition, nucleotide substitutions in this portion of the control region could be used to estimate the contribution of each female parent to the broods. This information will be important in the evaluation of breeding protocols designed to optimize levels of genetic variability in hatchery broods, in the assessment of genetic risk posed to wild red drum populations by hatchery stock- ing programs (e.g. Tringali and Bert, 1998), and in the evaluation of stock supplementation programs. Acknowledgments We thank M. Murphy and R. Taylor for valuable dis- cussions regarding red drum life history, R. Muller for advice on statistical analyses, and L. Barbieri for providing information on biotelemetry studies of red drum in Georgia. We also thank C. Crawford for technical assistance and R. Crabtree, B. Falls, A. McMillen-Jackson, S. Lotz, A. Redford, R. Ruiz- Carus, and T. Thompson for assistance with field col- lections. We thank the editorial staff of the Florida Marine Research Institute, J. Gold, D. O'Foighil, and D. Winkelman for valuable comments that improved the manuscript. Financial support was provided by the State of Florida and the U.S. Fish and Wildlife Service, Department of the Interior, Federal Aid for Sportfish Restoration Project Grant F-69 to TMB. Literature cited Avise, J. C. 1987. Identification and interpretation of mitochondrial DNAstocksin marine species. In H. Kumpfled.l, Proceed- ings of the stock identification workshop, p. 105-136. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFC-199. Bentzen, P., C. T. Taggart, D. E. Ruzzante, and D. Cook. 1996. Microsatellite polymorphism and the population struc- ture of Atlantic cod ( Gadus morhua i in the northwest Atlan- tic. Can. J. Fish. Aquat. Sci. 49:2,527-2,537. Bohlmeyer, D. A., and J. R. Gold. 1991. Genetic studies in marine fishes. II. A protein electro- phoretic analysis of population structure in the red drum, Sciaenops ocellatus. Mar. Biol. 108:197-206. Brown, A. H. D., and T. M. Clegg. 1983. Analysis of variation in related DNA sequences. In M. Dekker (ed.), Statistical analysis of DNAsequence data, p. 107-132. Marcel Dekker, New York, NY. Brown, J. R., A.T. Beckenbach, and M. J. Smith. 1993. Intraspecific DNA sequence variation of the mito- chondrial control region of white sturgeon lAcipenser trans- montanus). Mol. Biol. Evol. 10(21:326-341. Cabot, E. L., and T. Beckenbach. 1989. Simultaneous editing of multiple nucleic acid and protein sequences with ESEE. Comput. Appl. Biosci. 5: 233-234. Excoffier, L., P. E. Smousc, and J. M. Quattro. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479- 491. Fajen, A., and F. Breden. 1992. Mitochondrial DNA sequence variation among natu- ral populations of the Trinidad guppy, Poecilia reticulata. Evolution 46(51:1,457-1.465. Gold, J. R., T. L. King, L. R. Richardson, D. A. Bohlmeyer, and G. C. Matlock. 1994. Allozyme differentiation within and between red drum iSciaenops ocellatus) from the Gulf of Mexico and Atlantic ocean. -J. Fish Biol. 44:,567-590. Gold, J. R., and L. R. Richardson. 1991. Genetic studies in marine fishes. FV.An analysis of pop- ulation structure in the reddrum (Sciaenops ocellatus) using mitochondrial DNA. Fish. Res. (Amst.) 12:213-241. 1994. Genetic distinctness of red drum {Sciaenops ocellatus) from Mosquito Lagoon, eastern Florida. Fish. Bull. 92: 58-66. Gold, J. R., L. R. Richardson, C. Furman, and T. L. King. 1993. Mitochondrial DNA differentiation and population structure in red drum (Sciaenops ocellatus) from the Gulf of Mexico and Atlantic Ocean. Mar. Biol. (Berl.) 116:175- 185. Graves, J. E. 1996. Conservation genetics of fishes in the pelagic marine realm. In J. C. Avise and J. L. Hamrick (eds.). Conserva- tion genetics: case histories from nature, p. 335-366. Chap- man and Hall. New York, NY. Hartl, D. L., and A. G. Clark. 1989. Principles of population genetics. 2nd ed. Sinaeur Associates, Sunderland. MA. Hattori, M., and Y. Sakaki. 1986. Dideoxy sequencing method using denatured plasmid templates. Ann. Rev Biochem. 152:232-238. Hindar, K., N. Ryman, and F, Utter. 1991. Genetic effects of cultured fishes on natural popula- tions. Can. J. Fish. Aquat. Sci. 48:94.5-957. Jin, L., and J. W. H. Ferguson. 1990. Neighbor-joining tree and UPGMA tree software, vers. 2.0. Center for Demographic and Population Genet- ics, Univ. Texas Health Science Center, Houston, 16 p. 138 Fishery Bulletin 98(1) Johnson, D. R., and N. A. Fiinicelli. 1991. Spawning of the red drum in Mosquito Lagoon, east central Florida. Estuaries 14:74-79. Jones, A. C, S. A. Berkeley, J. A. Bohnsack, S. A. Bortone, D. K. Camp, G. H. Darcey, J. C. Davis, K. D. Haddad, M. Y. Hedgepeth, E. W. Irby, Jr., W. C. Jaap, F. S. Kennedy, Jr., W. G. Lyons, E. L. Nakamura, T. H. Perkins, J. K. Reed, K. A. Steidinger, J. T. Tilmant, and R. O. Williams. 1985. Ocean habitat and fishery resources of Florida. In W. Seaman Jr. (ed.), Florida aquatic habitat and fishery resources, p. 437-543. Special publication of the Florida Chapter. Am. Fish. Soc, Kissimee, FL. Kocher, T. D., W. K. Thomas, A. Meyer, S. V. Edwards, S. Paabo, F. X. Villablanca, and A. C. Wilson. 1989. Dynamics of mitochondrial DNA evolution in ani- mals: amplification and sequencing with conserved prim- ers. Proc. Nat. Acad. Sci. USA 86:6, 196-6,200. Kumar, S., K. Tamura, and M. Nei. 1993. MEGA: molecular evolutionary genetic analysis, vers. 1.01. Pennsylvania State Univ., University Park, PA 412 p. Marchuk, D., M. Drumm, A. Saulino, and F. S. Collins. 1990. Construction of T-vectors, a rapid and general system for direct cloning of unmodified PCR products. Nucleic Acids Res. 19(51:1,154. McEachron, L. W., C. E. McCarty, and R. P. Vega. 1995. Beneficial uses of marine fish hatcheries: enhance- ment of red drum in Texas coastal waters. Am. Fish. Soc. Symp. 15:161-166. McElroy, D., P. Moran, E. Bermingham, and L Komfield. 1992. REAP: an integrated environment for the manipu- lation and phylogenetic analysis of restriction data. J. Hered. 83:157-158. Mercer, L. 1984. A biological and fisheries profile of red drum, Sci- aenops ocellatus. N. Carolina Dep. Natl. Res., Div. Mar Fish. Spec. Sci. Rep. 41:1-89. Meyer, A., T. D. Kocher, P. Basasibwaki, and A. C. Wilson. 1990. Monophyletic origin of Victoria cichlid fishes suggested by mitochondrial DNA sequences. Nature 347:550-553. Moritz, C. 1994. Application of mitochondrial DNA analysis in conser- vation: a critical review. Mol. Ecol. 3:401-411. Murphy, M. D., and R. G. Taylor. 1990. Reproduction, growth, and mortality of red drum, Sci- aenops ocellatus in Florida. Fish. Bull. 88:531-542. Nei, M. 1987. Molecular evolutionary genetics. Columbia Univ. Press, New York, NY. Nei, M., and F. Tajima. 1981. DNA polymorphisms detectable by restriction endo- nucleases. Genetics 97:145-163. O'Foighil, D., T. J. Hilbish, and R. M. Showman. 1996. Mitochondrial gene variation in Merceneria clam sib- ling species reveals a relict secondary contact zone in the western Gulf of Mexico. Mar. Biol. 126:67.5-683. Palumbi, S. R. 1992. Marine speciation on a small planet. Trends Ecol. Evol. 7:114-118. Peters, K. M., and R. H. McMichael. 1987. Early life history iii Sciaenops ocellatus (Pisces: Sci- aenidae) in Tampa, Florida. Estuaries 10(2):92-107. Ramsey, P. R., and J. M. Wakeman. 1987. Population structure of Sciaenops ocellatus and Cyno- scion nebulosus (Pisces: Sciaenidae): Biochemical vari- ation, genetic subdivision and dispersal. Copeia 3:682-695. Rohlf, F. J., and D. E. Slice. 1995. BIOMstat for windows, statistical software for biolo- gists, vers. 3.0. ( manual ). Applied Biostatistics, Setauket, NY, 57 p. Ross, J. L., J. S. Pavela, and M. E. Chittenden Jr. 1983. Seasonal occurrences of black drum Pogonias cromis. and red drum, Sciaenops ocellatus, off Texas. Northeast Gulf Sci. 6(1 ):67-70. Saiki, R. K., D. H. Gelfand, S. Stoffel, S. Scharf, R. H. Higuchi, G. T. Horn, K. B. Mullis, and H. A. Eriich. 1988. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science (Wash. D.C.) 239:487-491. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for recon- structing phylogenetic trees. Mol. Biol. Evol. 4:406—425. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sanger, F., S. Nicklon, and A. R. Coulson. 1977. DNA sequencing with chain termination inhibitors. Proc. Natl. Acad. Sci. USA 74:5,463-5,467. Slatkin, M. 1987. Gene flow and the geographic structure of natural populations. Science (Wash. D.C.) 236:782-792. Sokal, R. R., and F. J. Rohlf. 1995. Biometry, 3rd ed. W. H. Freeman, New York, NY, 887 p. Stabile, J., J. R. Waldman, J. Hart, and I. Wirgin. 1996. Stock structure and homing fidelity is high in Gulf of Mexico sturgeon iAcipenser oxyrinchus desotoi) based on restriction fragment length polymorphism and sequence analyses of mitochondrial DNA. Genetics 144:767-775. Stepien, C. 1995. Population genetic divergence and geographic pat- terns from DNA sequences: examples from marine and freshwater fishes. Am. Fish. Soc. Symp. 17:263-287. Tringali, M. D., and T. M. Bert. 1996. The genetic stock structure of common snook {Centra- pornus undecimalis). Can. J. Fish. Aquat. Sci. 53:974-984. 1998. Risk to effective population size should be an impor- tant consideration in fish stock-enhancement programs. In Proceedings of marine stock enhancement: a new perspec- tive, p. 641-660. Bull. Mar Sci. 62(2). Utter, F. 1998. Genetic problems of hatchery-reared progeny released into the wild, and how to deal with them. In Proceedings of marine stock enhancement: a new perspective, p. 623—640. Bull. Mar. Sci. 62(2). Woodward, A. G., and N. Nicholson. 1997. Movement of adult red drum in the central south Atlantic Bight (abstract). In 1997 Southern Div, Am. Fish. Soc, February 13-16, San Antonio, TX, 28 p. Wright, S. 1943. Isolation by distance. Genetics 28:114-138. 139 Abstract.— On the eastern seaboard of the United States, populations of the blue crab, Callinectes aapidus, experi- ence recurring outbreaks of a parasitic dinoflagellate, Hematodinium perezi. Epizootics fulminate in summer and autumn causing mortalities in high- salinity embayments and estuaries. In laboratory studies, we experimentally investigated host mortality due to the disease, assessed differential hemato- logical changes in infected crabs, and examined proliferation of the parasite. Mature, overwintering, nonovigerous female crabs were injected with lO'^ or 10^ cells of//, perezi. Mortalities began 14 d after infection, with a median time to death of 30.3 ±1.5 d (SE). Sub- sequent mortality rates were greater than SG'/r in infected crabs. A relative risk model indicated that infected crabs were seven to eight times more likely to die than controls and that decreases in total hemocvte densities covaried signif- icantly with mortality. Hemocyte densi- ties declined precipitously (mean=48'^ ) within 3 d of infection and exhibited differential changes in subpopulations of gi-anulocytes and hyalinoc)d;es that lasted throughout the course of the infection. Crabs that did not present infections after injection (i.e. "immune" hosts) did not show hemocytopenia and exhibited significant long-term (21-27 d) granulocvtemia. Detection of the para- site in the hemolymph of infected crabs increased from approximately 30'^i after 14 d to 60'7< after 21 d to 100% after 35 d. Plasmodial stages were, however, detectable in histological preparations of the heart within 3 days of infection and increased in number over 5 and 7 days. Sporulation of the parasite occurred over a short time (at least 4 d, after 43 d of infection I and did not culminate in the immediate death of the host. Hematodinium perezi repre- sents a significant threat to the blue crab fisheries in high-salinity estuar- ies. Although the parasite infects male and female crabs, it may have a greater impact on mature females as they move to higher salinities to breed. Mortality and hematology of blue crabs, Callinectes sapidus, experimentally infected with the parasitic dinoflagellate Hematodinium perezi* Jeffrey D. Shields Christopher M. Squyars Department of Environmental Sciences Virginia Institute of Marine Science The College of William and Mary P.O. Box 1346, Gloucester Point, VA 23602 E-mail address (for J D Shields): |eff(givims edu Manuscript accepted 23 August 1999. Fish. Bull. 98:139-152 (2000). Hematodinium perezi is a parasitic dinoflagellate that proliferates in the hemolymph of several crab spe- cies. In the blue crab, Callinectes sapidus, H. perezi is highly patho- genic and usually kills the host. The main symptom of the infection is lethargy. Heavy infections are characterized by discolored (brown, yellow, milky or chalky ) hemolymph that does not clot. The disease oc- curs in blue crabs in high-salinity i>\Yi() waters from Delaware to Florida, and in the Gulf of Mexico (Newman and Johnson, 1975; Mes- sickandSinderman, 1992). In 1975, Newman and Johnson (1975) re- ported a prevalence of 309r in blue crabs from Florida; the effect of this disease on the blue crab population was thought to be high. In 1991 and 1992, prevalences of infection up to IQO'^/r were found in blue crabs (mean prevalence=43'^, several locations from 709f to 100*^, /7=971) from coastal bays in Mary- land and Virginia (Messick, 1994). Commercial watermen reported re- duced catches, lethargic and mor- ibund crabs in pots and shedding facilities, and crabs that died soon after capture (Rux, Oesterling'). In 1996 and 1997, IQ'A to 40% of adult crabs from the eastern portions of Chesapeake Bay in Virginia were infected.- The disease has a low prevalence or does not occur in the larger, riverine ("bayside") fishery; it appears most detrimental to the coastal ("seaside") crab fisheries. Outbreaks of infestation by Hema- todinium spp. have caused concerns to several major crustacean fisher- ies. Significant population declines and economic losses have been reported for the Tanner iChion- oecetes bairdi) and snow (C. opilio) crab fisheries of Alaska and New- foundland (Meyers et al.. 1987, 1990; Taylor and Khan, 1995),'^ the Norway lobster (Nephrops norveg- icus) fishery of western Scotland (Field et al., 1992), and the velvet crab iNecora puber) fishery of west- ern France (Wilhelm and Miahle, 1996). The parasite causes a condi- Contribution 2241 from the Virginia Insti- tute of Marine Science, The College of William and Marv, Gloucester Point, VA 23602. Rux, S. 1993. Red Bank Seafood Co., Box 37 Marionville, VA 23408. Personal common. ; Oesteriing, M. 1993. VASG, Virginia Inst. Marine Science, Gloucester Point. VA 23062. Personal commun. Shields. J. D. 1997. An investigation into the epizootiology of Hematodinium perezi, a parasitic dinoflagellate in the blue crab, Callinectes sapidus. Saltonstall-Kennedy Progi'am, National Marine Fisheries Ser- vice, NOAA. Final Report. Prevalences in Newfoundland are now at 1-15"^; in the northern bays. Taylor, D. 1998. DFO. CP 5567, White Hills, St. Johns, Newfoundland, Canada, AlC 5X1. Personal commun. 140 Fishery Bulletin 98(1) tion known as bitter crab disease in snow and Tanner crabs (Meyers et al., 1987). Low prevalences (1-4%) of another species, H. australis, have been reported in sand (Portunus pelagicus) and mud (Scylla ser- rata) crabs from Australia (Shields, 1992; Hudson and Shields, 1994). Infections of Hematodinium spp. or Hematodin- jum-like species have been reported from a variety of different hosts (see Shields, 1994, for review). There are, however, only two described species oi Hemato- dinium: H. perezi Chatton and Poisson, 1931, and H. australis Hudson and Shields, 1994. By convention (Newman and Johnson, 1975; MacLean and Rud- dell, 1978) and from its distinct morphological fea- tures, we concur that Hematodinium perezi is the infectious species in the American blue crab. Blue crabs sustain one of the largest fisheries in Chesapeake Bay. Current management plans and state regulations are based on population assess- ments that include numbers of juvenile and adult crabs found during winter, spring, and summer sur- veys (Lipcius and Van Engel, 1990; Abbe and Stagg, 1996; Rugolo et al., 1998). Although these projections include estimates of natural mortalities, they do not account for the potential epizootics and mortali- ties caused by Hematodinium perezi. In this study, we examined host mortality in controlled laboratory experiments and documented changes in the hemo- grams (total cell counts, and differential counts) of inoculated crabs versus uninfected crabs. We also examined proliferative growth of//, perezi at approx- imately weekly intervals and made observations on the biology and life history of the parasite. Materials and methods Blue crabs were collected from Chesapeake Bay and several of its subestuaries during the annual VIMS Winter Dredge Survey (part of the Chesapeake Bay Stock Assessment Program) with a 1.83-m-wide Vir- ginia crab dredge fitted with 0.5-inch ( 1.25-cm ) Vexar mesh dragged on the bottom for one minute at three knots. Crabs were also taken with commercial crab pots from two reference locations on the Delmarva Peninsula, Red Bank and Hungars Creeks, Virginia. Uninfected crabs were housed together for three to seven days prior to treatment to ensure acclimation and absence of overt bacterial or protozoal diseases (as assayed below). During the experiments, crabs were fed fish and squid semiweekly and held individ- ually in aquaria (5 gal., 19 liter) at 20° to 21°C, and 24 ppt salinity. Although H. perezi infects both sexes, only mature, nonovigerous female crabs (healthy, orange maturing gonads, little to no shel] damage, 120-160 mm carapace width including epibranchial spines) were used in the experiments. Females were used to limit the number of treatment effects (e.g. poten- tial differences between sexes) and to improve sample sizes given the laborious nature of the experiments. Hematodinium perezi was maintained in the lab- oratory by serial passage of infected hemolymph. Hemolymph from naturally infected crabs was in- jected directly into uninfected crabs. Naive (unex- posed) crabs and crabs used for inoculation experi- ments were obtained from low-salinity non-enzootic locations. Infected and inoculated crabs were housed separately and used as hemolymph donors to inject naive hosts (10^-10^ parasites per host). Injections were given in the arthrodial membrane of the fifth leg at the juncture of the basis with the cara- pace. We have maintained //. perezi for over seven months using this method with no apparent loss from pathogenicity. Two mortality experiments and one early life his- tory experiment were undertaken. The mortality-I experiment used raw, infected hemolymph as the inoc- ulant. Although appropriate for maintaining infec- tions in the laboratory, raw hemolymph cannot be adjusted to manipulate parasite densities without the use of physiological buffers, nor can it be guar- anteed as sterile without appropriate assessment (see Welsh and Sizemore, 1985). Preliminary experi- ments with sterile sea water, physiological buffers, and infected hemolymph indicated that buffer- washed parasites remained infectious, and could, therefore, be adjusted to consistent densities appropriate to controlled experiments. The mortality-II experiment used buffer-washed parasites adjusted to a density similar to that used in the mortality-I experiment. Mortality-II experiment closely resembled mortality-I experiment except for 1 ) handling (buffer washes with centrifugation) and 2) the use of plasmodial versus uninucleate stages of the parasite. Uninfected crabs served as controls in both experiments. Controls were used to assess handling effects and to establish base- line densities of hemocytes. The early infection exper- iment was designed to examine the effects of early infections on the hematology of the host and the early life history of the parasite. Experimental den- sities in the early infection experiment were four times higher than those in the previous experiments (4.1 X 10'' vs. approx. 1.0 x 10'^ parasites/crab, respec- tively) and were arbitrarily higher to insure obser- vation of parasites prior to their proliferation. In the mortality-I and mortality-II experiments different proportions of trophonts and plasmodia were used (for definitions see below). The mortal- ity-I experiment consisted of a control group of unin- fected crabs (n=22) injected individually with 100 pL Shields and Squyars: Mortality and hematology of Call/nectes sopidus infected with Hematodinium perezi 141 of hemolymph from an uninfected donor crab and an experimental gi'oup («=20) injected individually with 100 ]iL of infected hemolymph from a donor crab containing an estimated 1.3 x 10'' trophonts/mL (1.3 X 10^ trophonts per crab). The mortality-II experiment consisted of a con- trol group (n=8) injected individually with 100 ]iL of physiological saline buffer (modified from Apple- ton and Vickerman, 1998; NaCl, 19.31 g/L; KCl 0.65 g/L; CaCl.,-2H.,0 1.38 g/L; MgSO^-TH^O 1.73 g/L; Na2S04 0."'38 g/L; HEPES 0.82 g/L;) adjusted to pH 7.8, with added glucose (1.0 mg/mL) and two exper- imental treatments (high dose=1.0 x 10^ parasites/ crab; low dose=1.0 x lO"' parasites per/crab, /i = 10, 10 respectively). To prepare the inoculum for the exper- imental treatments, 2.0 mL of infected hemolymph were drawn from a donor crab infected with 6.15 x 10" parasites/mL (comprising 97% plasmodia; 3% trophonts). The infected hemolymph was diluted 1:1 with buffer, centrifuged at 4000 rpm for 10 minutes, the supernatant was decanted, and the cells were resuspended in buffer. The cells were then adjusted to 1.0 X 10'' parasites/mL, centrifuged through two more washes, and serially diluted to attain densities of 1.0 X 10*' parasites/mL and 1.0 x lO'* parasites/mL (for inoculum of 100 pL, 1.0 x 10^ parasites/crab and 1.0 X 10^ parasites/crab, respectively). In both experiments, crabs were monitored daily for mortalities. Deaths within the first nine days of each experiment were excluded because of handling stress arising from infrequent, bacterial infections (e.g. John- son, 1976). None of the crabs in the experiments were infected with amoebae, microsporans, or overt bacte- rial infections (but see Welsh and Sizemore, 1985 for background levels of Vibrio spp. in hemolymph of C. sapidus). Ten crabs from each treatment in the mor- tality-! experiment, and all of the crabs in the mortal- ity-II experiment were bled approximately weekly to assess infection status. In the mortality-I experiment, the same ten crabs were bled approximately weekly until they died; other crabs from within the experi- ment were added as replacements. Crab hemolymph was taken by using a tubercu- lin syringe (1 mL) with a 25.5-ga. needle from the arthrodial membrane at the juncture of the basis and the ischium of the 5th pereopod (swimming leg). Ethanol (70*7^^) was used to sterilize the site of inoculation and blood letting. Total and differential counts of host hemocytes and estimates of parasite density were obtained from individual crabs with a hemocytometer (Neubauer improved, Bright Line, two counts per crab) with phase contrast microscopy at 400x. Host hemocytes were identified as granu- locytes, semigranulocytes (intermediate cells with relatively few granules, Bodammer, 1978; Johnson, 1980) and hyalinocytes (cell types defined in Soder- hall and Cerenius, 1992). Hemocyte and parasite densities higher than 1.0 x IC cells/mL were diluted 1:5 with buffer and recounted to provide better esti- mates of cell density. For comparative purposes, total hemocyte densities and differential counts from naturally infected male and female crabs were also obtained. Parasites were easily distinguished from host cells by using phase contrast microscopy (Fig. 1): uni- nucleate trophonts (9-15 pm) possessed few small, refractile vacuoles and were rounded or amoeboid, without filopodia; multinucleate plasmodia (20-100 pm) were slender, vermiform, and motile. The den- sity of infection refers to the number of parasites per mL of hemolymph. Total hemocyte density refers to the number of hemocytes per mL of hemolymph. Mean intensity refers to the mean number of para- sites per quantity of infected host tissue (Margolis et al., 1982). Permanent preparations of hemolymph were pro- cessed and stained as described in Messick ( 1994). Briefly, acid-cleaned, poly-1-lysine-coated microslides were smeared with fresh hemolymph, allowed to stand for 2-3 minutes, and fixed in Bouin's fixative. The smears were processed through a routine Harris hematoxylin and eosin-Y procedure ( Humason, 1979, p. 123 without acid destain). The early infection experiment consisted of a con- trol group (n=5 crabs) injected individually with 100 pL of hemolymph from an uninfected donor crab and an experimental group {n=20) injected with 100 pL of hemolymph from a donor crab containing an estimated 4.1 x 10^ parasites/ml (4.1 x 10^ parasites per crab; comprising 79% plasmodia, 21% trophonts). Three days prior to infection, cell counts were con- ducted on all crabs to serve as a benchmark (presa- mple) for before-after comparisons. On days 3, 5, and 7 after inoculation, five infected crabs were bled and dissected. Differential cell counts were conducted and tissue samples taken for histological analysis. Tissue samples were processed through a routine hematox- ylin and eosin procedure and included muscle, hepa- topancreas, heart, and, in some cases, foregut. The control crabs were bled and tissue samples taken 10 days after injection. For statistical analyses, the proportional hazards model with the Weibull distribution was used to examine survival data and associated variables (Cox and Oakes, 1984 ). The Tarone-Ware log-rank test was used to examine differences between survival curves (Wilkinson, 1997). ANOVA was used to analyze rela- tionships in hemocyte densities and proportion of cell type (cell type density divided by total hemocyte den- sity) between inoculated and uninfected crabs. Simi- 142 Fishery Bulletin 98(1) B P .».♦ c*^ S' D O 4 r - • r • O V k - _ 1 « .1 / Figure 1 Hemalodinium perezi from the blue crab, Callinectes sapidus. (A and B) Vermiform Plasmodia (p) in hemolymph (granulocytes, star). Bar = 10 pm. (C) Amoeboid trophonts (arrows) with few refractile granules. Bar = 10 pm. (D) Round trophonts (arrows) with many refractile granules. Bar = 10 pm. lar densities and proportions of cell types were noted in hematology and survival between the mortality-! and mortality-II experiments; hence, data were com- bined a posteriori for the analyses. Where similar trends were noted between statistics for injection dosage ( lO-^ vs. 10''), data were also combined for the analysis (i.e. survivorship, hematology). SYSTAT (Wilkinson, 1997) and SAS (SAS, 1988) were used for the analyses. A probability level of P < 0.05 was accepted as significant. Results Inoculated crabs that became infected with Hema- todinium perezi began dying two weeks after inoc- ulation (Fig. 2). Mortalities peaked at three weeks after injection and continued to accumulate from weeks 3 through 5. The mortality rate of the infected crabs was 86%, whereas less than 20% of the con- trols died. Crab mortalities were similar over the time course of infection between mortality-I (infected Shields and Squyars: Mortality and hematology of Callinectes sapidus Infected with Hematodinium perezi 143 100 80 - 60 40 - 20 Expt 1, control Expt n, control 10 15 20 25 30 35 40 C/5 uu \ 1 t \ \ Expt 1 \\\ 80 Expt 11, high dose I 1 v. 60 Expt 11, low dose \ l 1 \ * 40 - ""*»". *- — ', \ 20 - \ \ A 1 r 1 . 1 1 . 1 1 1 1 . 1 . . 1 , , , , 1 . , , , 1 , . . , 1 . . . 1 . , \, 10 15 20 25 30 Days after inoculation 35 40 Figure 2 Survivorship curves for uninfected crabs and crabs infected with H. perezi (high dose=10' parasites/crab, low=10^ parasites/crab). Sample sizes were 22 and 8 uninfected controls; 20 (mortality-I experiment), 10 and 10 (high and low dose, mortality-II experiment) respectively. hemolymph, uninucleate trophonts) and mortal- ity-II (buffer-washed parasites, vermiform Plasmo- dia) experiments (Tarone-Ware, y}=\.2\ with 1 df, P=0.27), even between different initial doses of the parasite (Fig. 2; Tarone-Ware, x^=0.74 with 1 df , P= 0.39). Uninfected crabs (controls) experienced signif- icantly fewer mortalities than did infected hosts ( Fig. 3; Tarone-Ware, x^=, 19.27 with 1 df , P<0.001). The controls for the mortality-II experiment did, how- ever, exhibit background mortalities (Fig. 2); but the mortality rate was not significantly different from controls in the mortality- 1 experiment (Tarone-Ware, X2=0.65 with 1 df, P= 0.42). None of the control crabs developed infections with H. perezi. Because mortal- ities within treatments were similar between experi- ments, data were grouped for further analysis. The median time to death for infected crabs was 30.3 ±1.5 (SE) days. Because the controls exhibited few mortalities, the median time to death for the uninfected controls could not be calculated. Infected crabs had a significantly higher mortality rate, seven to eight times greater than that of the uninfected con- 144 Fishery Bulletin 98(1) 100 80 \ '% ~-~-~^ • ••-.^ ■•••.... -- _..... ^\ or fraction o \ x^ \^ Inoculated > \ " *■■- E 40 '•%. \ '•\ 3 C/3 'Cr^x 20 . . , . 1 . . , , 1 . , , , 1 . , , , 1 , . , . 1 . . , . 1 . . . . 1 , , , . 5 10 15 20 25 30 35 40 Days after inoculation VARIABLE ESTIMATE SE X^ P Intercept 2 933 106 763 33 OOOOI Uninfected 1055 287 13 50 0002 Injected Scale 517 083 Figure 3 Survival function (Kaplan-Meier estimation) for uninfected crabs and crabs infected with H. perezi. Dashed lines are upper and lower standard errors. Data include all crabs from mortality-I and mortality-II experiments. Sample sizes are given in Figure 1. trols (Fig. 3; proportional hazards, X"=13.50,P<0.001: ). Hemocyte and parasite den- relative risk=ei055''05i74 sity were jointly analyzed as covariates in the propor- tional hazards model. For injected crabs, the decline in ln( total hemocyte density) was significantly asso- ciated with mortality (In day of death = 0.875 + 0.145 In total hemocyte density - 0.017 In Parasite density + 0.409 W; X"=4.47 with 1 df, P<0.05). Hemocyte density (untransformed), and parasite density (In, and untransformed) were not associated with mor- tality (x'-, P=0.07, 0.61, and 0.47, respectively); thus, decreases in hemocyte density (In), not parasite den- sity, were associated with imminent death. Direct observations from crabs used to maintain infections and experimental results indicated that the parasite was detectable in the hemolymph approxi- mately two weeks after injection (Fig. 4). Although the parasite could be detected as early as one week after inoculation, detectability (the percentage of infected crabs exhibiting detectable parasites in the hemolymph) was relatively low (3,0-35%) after 14 to 18 days, reaching 80-85% after 26 to 32 days, and 100% after 35 days. (Detectability was based solely on inoculated animals that developed infections. The four crabs from the mortality-II experiment that did not present infections, hereafter refeiTed to as "immune" crabs, were excluded from the analysis of detectabil- ity. ) Proliferation and gi'owth of the parasite followed a similar pattern as detectability, and the two vari- ables were clearly related (Table 1; Fig. 4). Growth of the parasite showed a marked increase in the mean density of vermiform plasmodia over days 18 to 26 (Table 1). The mean density of trophonts increased markedly over days 32 to 35. Note, how- ever, that to avoid mortalities from other causes (e.g. secondary infections), sampling could not be done on a daily basis. Plasmodia were found within the hearts of 93% (n=14/15) of the injected crabs in the early infection experiment. Plasmodia were found in 4 of 5 crabs as early as day 3 (Table 2l. Uninucleate trophonts were observed in the heart on and after day 7. Rel- Shields and Squyars Mortality and hematology of Ca/linectes sapidus infected with Hematodinium perezi 145 Days alter injection Figure 4 Detectability of parasites in hemolymph of infected blue crabs over the course of infec- tion. Data were combined from mortality-I and mortality-II experiments and include only infected crabs. Samples sizes were 21, 11, 10, 10, 16, 4, and 4 crabs on days 7, 14, 18, 21, 26, 32. and 35, respectively. Table 1 Parasite intensity (vlO^ parasites/mLper infected hosti in the hemolymph in for mortality-I and mortality-II experiments. N^,^, = crabs with plasmodia, A^, exhibiting parasites. See Table 3 for sample sizes for hemocytometry. relation to days after roph ~ crabs with troph noculation. Counts combined onts, A^,„/,.t.,„; = infected crabs Days Np,., Plasmodia log( Plasmodia '^Iroph Trophont log( trophont) N Mean ±SE Mean ±SE Mean ±SE Mean ±SE 7 1 0.50 ±0.00 4.70 ±0.00 1 0.25 ±0.00 4.40 ±0.00 1 14 4 1.44 ±0.53 5.09 ±0.16 3 1.25 ±0.29 5.06 ±0.11 4 18 1 1.25 ±0.00 5.10 ±0.00 2 1.00 ±0.25 4.99 ±0.11 3 21 5 7.70 ±4.61 5.61 ±0.24 5 8.65 ±4.17 5.87 ±0.18 5 26 11 14.98 ±6.07 5.85 ±0.17 12 8.19 ±3.86 5.40 ±0.22 14 32 2 8.38 ±1.63 5.92 ±0.09 3 4.25 ±1.32 5.59 ±0.14 3 3.5 4 7.88 ±3.89 5,69 ±0.25 4 49.81 ±36.82 6.35 ±0.31 4 atively more parasites were observed in the heart tissue over time (Table 2); but no effort was made to standardize area in the histological preparations. Growth of the parasite was rapid in the heart. The dosage in the early infection experiment was, how- ever, four times higher than that in the mortality-I and mortality-II experiments; thus, results between experiments were not directly comparable. Sporulation from the trophont stage to the dino- spore stage was observed only in crabs that were used to maintain infections. Parasites in one crab sporulated at least twice and each event lasted less than 4 d. Parasite density was extraordinarily high (1.6x10^ dinospores/mL) during sporulation, and dropped to moderate levels (3.3 x 10^ trophonts/mL) thereafter. Dinospores were observed five times over the course of 26 d, beginning 43 d after injection. Additionally, some crabs injected with only the tro- phont (vegetative) stage were observed with Plasmo- dia after 3 to 4 weeks of infection. Hemograms of infected crabs were significantly different from those of uninfected controls (Tables 146 Fishery Bulletin 98(1) 3 and 4, Fig. 5). Total hemocyte density was signifi- cantly depressed in infected crabs (Fig. 5A; 2-way ANOVA by group and day, i^=5.03, P<0.001). Total Table 2 Relative intensity of Plasmodia in histological prepara- tions of heart sections of mature, nonovigerous female blue crabs from the early infection experiment. Mean intensi- ties represent direct counts of Plasmodia and are not stan- dardized by tissue area. Day N,„f^c,ed/N!,j,,t,d Mean intensity (±SD) Range Control 0/4 3 4/5 5 5/5 7 5/5 0.0 ±0.0 — 3.6 ±3.9 1-10 12.0 ±11.0 1-26 55.8 ±26.1 15-74 hemocyte density was not significantly different between crabs inoculated with different initial doses (2-way ANOVA, 10'^ vs. 10'^ parasites per crab and day F=3.19, df=l, 64, P=0.079). Crabs that were in- jected and did not^acquire the infection ("immune" hosts) did not have significant decreases in hemocyte densities (Table 3, Fig. 5; 2-way ANOVA, F=IA6. df = 13, 105, P=0.145). In the early infection experiment, the decrease in total hemocyte densities occurred within three days of inoculation (Table 5). In addition to a decrease in cell density, the pro- portions of different host cell types (density of cell type divided by total hemocyte density) in infected crabs shifted to those with proportionally more gran- ulocytes than hyalinocytes (Table 6, Fig. 5, B and D) (2-way ANOVA, i^=1.83; df=20, 149, P<0.05). Sig- nificant shifts in the population of semigranulo- cytes were also noted (F=2.5l, df =20, 149, P<0.001). Table 3 Total mortality-hemocyte densities (xlO*^ hemocytes/mLl in relation to days after inoculation. Hemocyte counts were combined from mortality-1 and mortality-II experiments. ( - = not done, no infected crabs survived to day 40 1. Days Uninfected control crabs Inoculated infected crabs Inoculated, immune crabs N Hemocyte density Hemocyte density N Hemocyte density Mean ±SE N Mean ±SE Mean ±SE 7 18 29.17 ±3.09 22 14.39 ±2.05 4 22.99 ±2.32 14 8 32.81 ±3.01 11 16.10 ±3.19 4 33.25 ±2.76 18 9 32.28 ±6.72 10 12.24 ±1.51 - 21 8 23.83 ±2.20 10 17.68 ±4.39 4 32.72 ±1.04 26 18 26.65 ±2.15 16 7.64 ±1.49 4 26.29 ±8.00 32 12 20.97 ±3.41 4 4.21 ±2.04 ~ - 35 8 23.37 ±4.48 4 10.86 ±6.73 4 23.61 ±11.72 40 10 20.35 ±2.12 - .All . I.M.I 4 22.53 ±5.69 Table 4 Total and differential hemocyte densities (mean ±SE; x 10^ hemocytes/mL) in relation to severity of infection (light < 4.0 x lO""' parasites/mL; moderate = 4.0 parasites/mL; - = not done I. natu X lO"^ ■ally infected male and female blue crabs in to 2.0 X 10>* parasite.s/mL; heavy > 2.0 x 10'' Severity n Hemocyte density Granulocyte density Semigranulocyte density Hyalinocjfte density Mature males Light 4 16.16 ±2.67 5.46 ±1.41 7.69 ±2.07 3.02 ±0.61 Moderate 6 7.46 ±1.74 1,92 ±0.63 4.10 ±1.10 1.45 ±0.46 Heavy 16 6.66 ±2. .53 1..34±0..53 3.39 ±1.35 1.93 ±0.76 Mature females Light Moderate 5 22.41 ±11.25 9.32 ±5.15 5.44 ±2.85 7.65 ±3.66 Heavy 5 14.37 ±7.37 5.98 ±3.09 4.01 ±1.63 4.. 39 ±2.89 Shields and Squyars Mortality and hematology of Callinectes sapidus Infected with Hematodlnium perezi 147 "Immune" crabs exhibited a fluctuation in cell types with significantly higher proportions of granulocytes to semigranulocytes during the first five weeks after inoculation (F=4.35, df =5, 18, P<0.01). By day 40, the hemograms of "immune" hosts were virtually identical to those of the uninfected controls (Table 7, Fig. 5, C and D). In the early infection experiment, hemocyte popu- lations shifted within the first three days of infec- tion (Tables 5 and 6; ANOVA, log hemocytes, F=9.16; df =3, 31, P<0.01); the proportion of granulocytes in infected crabs increased significantly compared with the proportion of semigranulocytes (ANOVA, 7^=4.39, P<0.05). Uninfected crabs exhibited minor fluctua- tions in the proportion of granulocytes to that of hya- linocytes but the proportions were similar to those observed in the mortality-I and mortality-II experi- ments (Tables 6 and 7). Discussion In laboratory experiments, Hematodinium perezi caused significant mortality to infected mature, nonovigerous blue crabs. Infections were not always fatal (four crabs survived inoculation without devel- oping infections), but the overall mortality to labo- oratory-inoculated crabs was high at 86% over 40 days. The proportional hazards model indicated that infected crabs were seven to eight times more likely to die than uninfected crabs. Infections in Tanner crab, Chionoecetes bairdi, and Norway lobster, Nephrops norvegiciis, are frequently fatal to the host (Meyers et al., 1987; Field et al., 1992). The mortality of natu- urally infected Tanner crabs held in aquaria for 97 days was 67% (;? = 11) and hosts survived from 20 to 158 days in the laboratory. Uninfected Tanner crabs experienced no mortality during the course of the experiment (Meyers et al., 1987). Naturally infected Norway lobsters suffered mortality rates of 86% to 100% over 27 d and 75 d, respectively, and had mor- tality rates 2-4 times higher than uninfected lob- sters, and most of the deaths occurred early in the course of the experiment (Field et al., 1992). During epizootics, juvenile blue crabs have a higher prevalence of//, perezi than do mature hosts (Messick, 1994). Male blue crabs have a prevalence of infection similar to that for females along the Total hemocyte density S 7 5 o. g ■^ = 55 5 10 15 20 25 30 35 |o Immune" hosts 6 •S 4 c o I 0.2 o u. a. 40 granule semigranulo hyalino 5 10 15 20 25 30 35 40 Infected hosts O-0.6 04 |02 ^ / granulo semigranulo hyalino 10 15 20 25 30 35 40 Uninfected hosts g.06 04 o |02 p granulo semigranulo hyalino 5 10 15 20 25 30 35 40 Days after inoculation Days after moculation Figure S Total hemocyte densities and proportions of host cell types in uninfected, infected and "immune" crabs. Data combined from mor- tality-I and mortality-II e.xperiments. Bars = SE. Standard errors (not shown) for proportion of host cell types were low (0.02-0.05). "Immune" crabs were survivors from mortality-II e.\periment that never developed infections. Sample sizes given in Table 3. 148 Fishery Bulletin 98(1) Table 5 Differential densities of hemocytes (xlOi^ hemocytes/ml ) in relation to days after inoculation for crabs in the early infection experi- ment. Sample sizes are given in Table 2. Days Uninfected control crabs Presample 10 Inoculated, infected crabs Presample 3 5 7 Granulocyte density Mean ±SE 3.72 ±0.31 7.48 ±1.60 6.88 ±0.92 4.95 ±0.40 2.51 ±0.91 2.93 ±0.68 Semigranulocyte density Mean ±SE 7.66 ±0.50 11.20 ±1.94 11.43 ±1.00 4.36 ±0.31 5.40 ±1.60 6.92 ±0.80 Hyalinocyte density Mean ±SE 4.53 ±0.79 5.30 ±1.47 4.97 ±0.40 2.80+0.15 1.49 ±0.51 1.56 ±0.31 Delmarva Peninsula (Messick, 1994; Messick and Shields^) and show significantly gi-eater changes than females in certain blood parameters.'' Infected blue crabs apparently die before acquiring the bitter flavor found in infected Tanner and snow crabs. Survival analysis indicated that parasite density was not associated with mortality. Similarly, survival time of Norway lobsters did not show a significant relationship with severity of infection, but host mor- tality did increase with the progression of the disease (Field et al., 1992). In blue crabs, absolute declines in ln( total hemocyte density) were associated with host mortality. Hence, the cellular defensive response of the host appeared seriously compromised by infec- tion. Anecdotal evidence from hosts used to maintain the parasite suggests that infections established with Plasmodia are more pathogenic than those estab- lished with trophonts; this may explain the similar mortality curves for the high and low doses of Plas- modia in the mortality-II experiment. Observations on naturally and experimentally infected crabs indi- cate three possible outcomes to the disease: 1 Crabs with acute infections, such as those reported here, show rapid mortalities, typically dying within 40 d. Acute infections rarely lead to heavy infections (10'^"^ parasites/mL), and may not lead to the development of dinospores. •• Messick, G. and J. D. Shields. 1999. The cpizootiology of the parasitic dinoflagellate HcmatiHiiniuin perezi in the blue crab, Callinectes sapidus. Oxford Cooperative Laboratory, 904 S. Morris St., Oxford, MD 21654. Unpubl. data. ^ Shields, J. D. 1999. Mortality and pathophysiology studies of blue crabs infected with the parasitic dinoflagellate Hcmnlo- dinium perczi. Saltonstall-Kenncdy Program. NOAA/National Marine Fisheries Service. Final Report. Table 6 Total hemocyte densities (xlO'' hemocytes/mL) in relation to days after inoculation (-- = not done) in the early infec- tion experiment. Uninfected control crabs Inoculated, infected crabs Hemocyte density Days n Mean ±SE n Hemocyte density Mean ±SE Presample 5 15.91 ±1.25 3 5 7 10 5 23.98 ±4.87 20 5 5 5 23.28 ±2.06 12.11 ±0.. 50 9..39 ±2.94 11.40 ±0.96 2 Crabs with chronic infections (observed in very cases, n=4) endure the acute stage, survive for longer periods (up to 90 days), and develop infec- tions that lead to massive numbers of dinospores (Fig. 6). 3 Some crabs successfully resist the parasite or are refractory to the infection. Preliminary experi- ments (not shown) suggest that resistant crabs (/! = 10) may become refractory to further inocula- tions with H. perezi. Blue crab catches fluctuate yearly in Chesapeake Bay but causes for these fluctuations are not well understood. Since salinity appears to limit the dis- tribution of H. perezi (Newman and Johnson, 1975), the dinoflagellate could feasibly infect and cause significant mortalities to juvenile and adult crabs Shields and Squyars: Mortality and hematology of Callinectes sapldus infected with Hematodinium perezi 149 Table 7 Differentia densities of hemocytes (xlO'' hemocytes/mL) in relation U days after inoculation. Hemocyte counts were combined from morl; lity-1 and mortality-II experiments (- = no data, no infected crabs survived to day 40) Sample sizes are given in Table 3. Days Granulocyte density Semigranulocyte density Hyalinocyte density Mean ±SE Mean ±SE Mean ±SE Uninfected control crabs t 8.13 ±0.97 13.88 ±1.28 7.16 ±1.10 14 8.34 ±1.17 15.13 ±1.66 9.34 ±0.94 18 8.99 ±1.42 15.37 ±3.33 7.83 ±2.18 21 5.88 ±0.93 10.86 ±0.93 7.09 ±0.91 26 6.56 ±0.73 13.06 ±1.04 7.04 ±0.76 32 4.10+0.81 12.14 ±2.21 4.73 ±0.53 35 5.64 ±1.15 11.67 ±2.34 6.05 ±1.37 40 4.66 ±0.61 9.29 ±1.13 6.39 ±0.68 Inoculated, infected crabs 7 5.28 ±1.04 6.18 ±0.71 2.92 ±0.56 14 5.59 ±1.34 6.90 ±1.26 3.62 ±0.94 18 3.29 ±0.67 6.76 ±0.65 2.20 ±0.47 21 4.77 ±1.23 8.37 ±1.75 4.60 ±1.54 26 2.10 ±0.45 4.25 ±0.83 1.28 ±0.36 32 1.11 ±0.57 2.19 ±1.14 0.90 ±0.41 35 3.11 ±2.13 6.11 ±3.58 1.64 ±1.06 40 - - -- Inoculated. immune crabs 7 9.73 ±0.65 7.08 ±1.58 6.19 ±0.93 14 12.31 ±0.83 11.91 ±1.34 9.03 ±1.00 18 -- -- 21 11.34 ±2.51 11.94 ±3.44 9.44 ±4.75 26 8.16 ±3.73 10.00 ±2. .55 8.14 ±2.80 32 -- -- - 35 7.98 ±4.03 8.03 ±3.89 7.61 ±3.82 40 4.59 ±1.35 10.00 ±2.30 7.94 ±2.71 where salinities are greater than W^h; i.e. much of the mainstem of Chesapeake Bay. Current models for blue crab populations in Chesapeake Bay are based on population assessments from various sur- veys (Lipcius and Van Engel, 1990; Abbe and Stagg, 1996; Rugolo et al., 1998). These models project crab abundance for the fishery as a whole but do not sep- arate the larger, low-salinity "bayside" fishery from the smaller, high-salinity "seaside" fishery where mortalities due to H. perezi occur. Such projections include estimates of natural mortalities but do not account for the potential epizootics and resulting mortalities caused by H. perezi. Differential models of exploitation by region may be warranted, espe- cially during or immediately following epizootics. Disease estimates must, however, account for the variation in the prevalence of detection because the prevalence in field samples may be significantly underreported (see Fig. 3). The life cycle of H. perezi from C. sapid us has not been fully documented. Several morphological and life history differences, however, distinctly separate Hematodinium sp. ex A'', norvegicus (Appleton and Vickerman, 1998) from H. perezi ex C. sapidus. For example, the syncytial and network forms of Hema- todinium sp. ex N. norvegicus (Field and Appleton, 1995) have not been observed with H. perezi (Mes- sick, 1994; present study); nor do the plasmodia (cf filamentous trophonts of Appleton and Vickerman, 1998) of H. perezi develop as "gorgonlocks"; rather, they undergo budding to produce additional Plas- modia, and schizogony (cf. segmentation in malaria life cycles) to produce uninucleate trophonts (senior author, unpubl. data). The trophonts then undergo further fission. Such differences may warrant generic separation between the two parasites. Hematodinium perezi was successfully transmitted to blue crabs by injection. Transmission experiments 150 Fishery Bulletin 98(1) 1,000,000,000 i -•- Plasmodia E 100,000,000 ;5 —k- Trophonts ! Dmospores / T 3 10,000,000 \ y^ te densi 8 i/^^-V w^ 1 100,000 •--■■-• 1.1,1 J 1 8 15 22 29 36 43 50 57 64 71 78 Days after injection Figure 6 Crab no. 3977 presenting a chronic, experimentally induced infection of H. perezi (plasmodia, circles; trophonts. triangles; dinospores. boxes). The crab died after 80 days. with the parasite in Tanner crabs and Austrahan sand crabs (Portunus pelagicus) have been partially successful. Parasites from primary cell culture (with sterile hemolymph) were successful in establishing infections in Tanner crabs, but inoculation with vege- tative stages did not produce infections (Meyers et al., 1987). Injections of trophonts (vegetative stages) were successful in producing infections in the sand crab, but other stages were not investigated (Hudson and Shields, 1994). Infection experiments with Norway lobster have not been reported. Transmission with cultured dinospores has yet to be achieved (Appleton and Vickerman, 1998). Sporulation is a rapid event with H. perezi, presumably occurring over several hours instead of several days or weeks as reported for Hematodinium sp. from Tanner crabs (Meyers et al., 1987, 1990). At lower temperatures and salinities, H. perezi apparently ceases to grow or slows its prolif- eration in naturally infected blue crabs. ^ Densities of circulating hemocytes declined rapidly in infected blue crabs. The decline occurred within the first three days and progressed to a 48% decrease in total hemocyte densities within the first week of infection. After three weeks, absolute declines of up to 809^ were noted for total hemocyte densities. The loss of cells was evident early in the infection even though the parasites were not detectable in hemolymph. Declines in hemocyte densities have been reported for starved lobsters (33-60% loss) (Stewart et al., 1967), Aeromo/ms-infected lobsters ( 80-84% loss) ( Stewart et al., 1983), Fusarium -miected brown shrimp, Penaeus californiensis (approximately 887c loss) (Hose et al., " Messick, G. 1998. Oxford Cooperative Laboratory, 904 S. Morris St., Oxford, MD 21654. Personal, commun. 1984), and V;'6r/o-infected Cancer irroratus (95% loss) (Newman and Feng, 1982). Reductions in hemocytes were noted for Norway lobster, N. norvegicus, infected with Hematodinium sp. (Field and Appleton, 1995), and for blue crabs, C. sapidus, infected with Par- amoeba perniciosa (Sawyer et al., 1970), but the degree of loss, and differential counts were not quan- tified. Declines in hemocyte counts occur quickly in crayfish, Pacifastacus leniusculus, (10 min) and are associated with loss of resistance to Aphanomyces infections; the declines are dependent upon the stim- uli (yeast vs. zymosan vs. buffers), and are evident over the course of several days (Perrson et al., 1987). Crustacean cell types probably represent matura- tion of a single lineage (e.g. Bodammer, 1978; Bachau, 1981; Hose et al., 1990). Hyalinocytes represent younger cells that become semigranulocytes (inter- mediate hemocytes), then granulocytes. Infected crabs exhibited marked shifts in subpopulations of different hemocyte stages (cell types). Because there was an absolute decline in the total number of cir- culating hemocytes and relative declines in hya- linocytes and semigranulocytes, we suggest that cell death and differential sequestration occur in response to the disease. General declines in hemocyte density in A^. norvegicus infected with Hematodin- ium sp. may occur from sequestration, other defense reactions, and hydrostatic effects of heavy infections or clogging of hemal sinuses (Field and Appleton, 1995). In our study, the rapid decline in total hemo- cyte density (within three to seven days) argues against hydrostatic effects and clogged sinuses. The shift towards proportionally more granulocytes than hyalinocytes may result from mobilization of tissue- dwelling reserves, differential cell death (Mix and Shields and Squyars: Mortality and hematology of Callinectes sapidus infected with Hematodinium perezi 151 Sparks, 1980), changes in mitotic stimuli of hemo- poetic tissue (Hose et al., 1984). or sequestration of specific cell types in defense of the infection (e.g. nodule formation, Johnson, 1976, 1977). Hyalino- cytes and semigranulocytes are the major phagocytic hemocytes in crustaceans (Bachau, 1981; Hose et al., 1990). Such hemocytes form nodules in bacte- rial, amoebic, and Hematodinium infections in blue crabs, Hematodinium infections in A^. norvegicus, and gaffkemia infections in Homarus americanus (Johnson, 1976, 1977; Johnson et al., 1981; Messick, 1994; Field and Appleton, 1995) are thus removed from circulation, and may account, in part, for the observed declines. In fungal infections of crayfish, P. leniusculus, hemocytic nodules do not dissociate in the presence of zymosan, a yeast derivative, and may last several days (Perrson et al., 1987). In Aeromo- ?jas-infected lobsters, the hyalinocytes increase in proportion to the other cell types, presumably by increased mitotic activity in hemopoeitic centers, but there is a significant decline in hemocytes after five days of infection (Stewart et al., 1983). Lastly, several blue crabs (n =4) successfully fought off the infection. These "immune" crabs exhibited significant sustained levels of granulocytes, did not suffer hemocytopenia, their hemolymph clotted nor- mally, and they did not exhibit gross changes in morbidity. Histological preparations of heart, hepa- topancreas, muscle, and hemopoeitic tissues were negative for latent infections in the "immune" ani- mals. The relative increase in semigranulocytes and sustained densities of hyalinocytes over time (Fig. 5) suggests an increase in mitotic activity in hemopo- etic tissue in response to the infection. This increase may not be sufficient to counter the parasite in sus- ceptible hosts. In A^. norvegicus infected with Hema- todinium sp., the hemopoeitic tissues show marked increases in activity. Although changes in host cell densities were not quantified, apparent stem cells were present in the active nodes (Tables 1 and 2 in Field and Appleton, 1995). The role of the granulo- cytes in the defense against Hematodinium infec- tions and the underlying mechanisms leading to refractory hosts warrant further study. Acknowledgments We wish to thank Seth Rux and Mike Seebo for their generous help. Seth Rux and members of the VIMS Dredge Surveys ably provided crabs. We thank Mary Anne Vogelbein, Romuald Lipcius, and Morris Rob- erts for technical advice, as well as Mike Newman for assistance with the survival analyses. Gretchen Messick and John Pearce improved the manuscript. This work was supported by NOAA, Saltonstall-Ken- nedy Grant NA76FD0148. Literature cited Abbe, G. R., and C. Stagg. 1996. Trends in blue crab {Callinectes sapidus Rathbun) catches near Calvert Cliffs, Maryland from 1968 to 1995 and their relationship to the Maryland commercial fishery. J. Shellfish Res. 15:751-758. Appleton, P. L., and K. Vickerman. 1998. In vitro cultivation and developmental cycle of a par- asitic dincflagellate (Hematodinium sp.) associated with mortality of the Norway lobster (Nephrops norvegicus) in British waters. Parasitology 116:115-130. Bachau, A. G. 1981. Crustaceans. In N, A. Ratcliffe and A. F. Rowley (eds. ), Invertebrate blood cells, vol. 2. Arthropods to urochor- dates, invertebrates and vertebrates compared, p. 385-420. Academic Press, London, 641 p. Bodammer, J. E. 1978. Cytologica! observations on the blood and hemopoi- etic tissue in the crab. Callinectes sapidus. Cell Tiss. Res. 187:79-96. Cox, D. R., and D. Oakes. 1984. Analysis of survival data. Chapman and Hall, NY, 201 p. Field, R. H., and P. L. Appleton. 1995. A Hematodinium-hke dinoflagellate infection of the Norway lobster Nephrops norvegicus: observations on path- ology and progression of infection. Dis. Aquat. Org. 22: 115-128. Field, R. H., C. J. Chapman, A. C. Taylor, D. M. Neil, and K. Vickerman. 1992. Infection of the Norway lobster Nephrops norvegicus by a Hemalodinnim-M'ke species of dinoflagellate on the west coast of Scotland. Dis. Aquat. Org. 13:1-15. Hose, J. E., D. V. Lightner, R. M. Redman, and D. A. Danald. 1984. Observations on the pathogenesis of the imperfect fungus, Fusanum solani, in the California brown shrimp, Penaeus californiensis. J. Invertebr. Pathol. 44:292-303. Hose, J. E., G. G. Martin, and A. S. Gerard. 1990. A decapod hemocyte classification scheme integrating morphology, cytochemistry, and function. Biol. Bull. 178: 33-45. Hudson, D., and J. D. Shields. 1994. Hematodinium australis n. sp., a parasitic dinoflagel- late of the sand crab, Portunus pelagicus, and mud crab, Scylla serrata. from Moreton Bay, Australia. Dis. Aquat. Orgs., 19:109-119. Humason, G. L. 1979. Animal tissue techniques, 4th ed. Freeman and Co., San Francisco. Johnson, P. T. 1976. Bacterial infection in the blue crab, Callinectes sapi- dus: course of infection and histopathology. J. Invertebr. Pathol. 28:25-36. 1977. Paramoebiasis in the blue crab, Callinectes sapidus. J. Invertebr Pathol. 29: 308-320. 1980. Histology of the blue crab, Callinectes sapidus: a model for the Decapoda. Praeger Scientific, NY, 440 p. Johnson, P. T., J. E. Stewart and B. Arie. 1981. Histopathology of Aerococcus viridans var. humari infection (Gaffkemia) in the lobster, Homarus americanus, 152 Fishery Bulletin 98(1) and a comparison with histological reactions to a Gram-neg- ative species Pseudomonas perolens. J. Invertebr. Pathol. 38:127-148. Lipcius, R. N., and W. A. Van Engel. 1990. Blue crab population dynamics in Chesapeake Bay: variation in abundance (York River, 1972-1988) and stock- recruitment functions. Bull. Mar Sci. 46:180-194. Margolis, L., G. W. Esch, J. C. Holmes, A. M. Kuris, and G. A. Schad. 1982. The use of ecological terms in parasitology. J. Para- sitol. 68:131-133. MacLean, S. A-, and C. L. Ruddell. 1978. Three new crustacean hosts for the parasitic dinofla- gellate Hematoduuum perezi (Dinofiagellata: Syndinidae). J. Parasitol. 64:158-160. Messick, G. A. 1994. Hematodinium perezi infections in adult and juvenile blue crabs Callinectes sapidus from coastal bays of Mary- land and Virginia, USA. Dis. Aqua. Org., 19:77-82. Messick, G. A., and C. J. Sinderman. 1992. Synopsis of principal diseases of the blue crab, Cal- linectes sapidus. U.S. Dep. Commer., NOAATech. Memo. NMFS-F/NEC-88. Meyers, T. R., C. Botelho, T. M. Koeneman, S. Short, and K. Imamura. 1990. Distribution of bitter crab dinoflagellate syndrome in southeast Alaskan Tanner crabs Chionoecetes bairdi. Dis. Aquat. Org. 9:37-43. Meyers, T. R., T. M. Koeneman, C. Botelho, and S. Short. 1987. Bitter crab disease: a fatal dinoflagellate infection and marketing problem for Alaskan Tanner crabs Chion- oecetes bairdi. Dis. Aquat. Org. 3:195-216. Mix, M. C, and A. K. Sparks. 1980. Tanner crab Chionoecetes bairdi Rathbun haemocyte classification and an evaluation of using differential counts to measure infection with a fungal disease. J. Fish Dis. 3:285-293. Newman, M. C, and S. Y. Feng. 1982. Susceptibility and resistance of the rock crab. Cancer irroratus, to natural and experimental bacterial infection. J. Invertebr Pathol. 40:75-88. Newman, M. W., and C. A. Johnson. 1975. A disease of blue crabs iCallinectes sapidus) caused by a parasitic dinoflagellate, Hematodinium sp. J. Para- sitol. 61:554-557. Persson, M., L. Cerenius, and K. Soderhall. 1987. The influence of haemocyte number on the resistance of the freshwater crayfish, Pacifastacus leniusculus Dana, to the parasitic fungus Aphanomyces astaci. J. Fish Dis. 10:471-477. Rugolo, L. J., K. S. Knotts, A. M. Lange, and V. A. Crecco. 1998. Stock assessments of Chesapeake Bay blue crab {Cal- linectes sap/rfus Rathbun). J. Shellfish Res. 17:493-517. SAS Institute, Inc. 1988. SAS, STAT user's guide, release 6.03. SAS Institute, Inc.. Cary, NC, 1028 p. Sawyer, T. K., R. Cox, and M. Higginbottom. 1970. Hemocyte values in healthy blue crabs, Callinectes sapidus. and crabs infected with the amoeba, Paramoeba perniciosa. J. Invertebr. Pathol. 15:440-446. Shields, J. D. 1992. The parasites and symbionts of the blue sand crab, Portunus pelagicus. from Moreton Bay, Australia. J. Crust. Biol. 12:94-100. 1994. The parasitic dinoflagellates of marine Crustacea. Ann. Rev Fish Dis. 4:241-271. Soderhall, K., and L. Cerenius. 1992. Crustacean immunity. Ann. Rev. Fish Dis. 2:3-23. Stewart, J. E., B. Arie, and L. Marks. 1983. Hemocyte patterns during gaffkemia infections and induction of resistance in Homarus americanus. Rapp. P-V. Reun. Cons. Int. Explor. Mer. 182:126-129. Stewart, J. E., J. W. Comick, D. M. Foley, M. F. Li, and C. M. Bishop. 1967. Muscle weight relationship to serum proteins, hemo- cytes, and hepatopancreas in the lobster, Homarus ameri- canus. J. Fish. Res. Board Can. 24:2339-2354. Taylor, D. M., and R. A. Khan. 1995. Observations on the occurrence o{ Hematodinium sp. (Dinofiagellata: Syndinidae), the causative agent of bitter crab disease in Newfoundland snow crab ^Chionoecetes opilio). J. Invertebr Pathol. 65:283-288. Wilhelm, G., E. Mialhe. 1996. Dinoflagellate infection associated with the decline o{ Necora puber crab populations in France. Dis. Aquat. Orgs. 26:213-219. Wilkinson, L. 1997. SYSTAT 7.0: new statistics. SPSS, Inc., Chicago, IL, 303 p. Welsh, P. C, and R. K. Sizemore. 1985. Incidence of bacteremia in stressed and unstressed populations of the blue crab, Callinectes sapidus. Applied Environ. Microbiol. .50:420-425. 153 Abstract.— Atlantic sturgeon ^Aapcnscr oxynnchtis) are a large anadromous fish which is especially vulnerable to overharvesting owing to its late age of maturity and low rate of reproduction. Age determination methods and growth rates are poorly described for this spe- cies. Pectoral-fin spine sections and sag- ittal otolith sections were examined to determine whether one of these struc- tures would be u.seful in estimating the age and growth of Atlantic sturgeon. Otoliths have been difficult to collect, process, and interpret. Interpretation of annuli in sectioned pectoral spines has proven to be an unbiased method for aging juvenile and adult Hudson River Atlantic sturgeon. Marginal incre- ment analysis has indicated an annual cycle of annulus deposition. Microchem- ical analysis with an electron micro- probe of the periphery of fin spines has shown seasonal patterns of calcium and phosphorus concentrations related to the translucent and opaque zones of the annuli. Formation of yearly annuli was verified in 4-year-old laboratory-reared sturgeon. \'on BertalanflTy growth models (based upon fin-spine interpretations) were fitted for the Hudson River popu- lation. Models predicted a more rapid growth rate for males than for females (females: K=0.07, L„=251 cm; males: A'=0.25,Z,. = 180cm). Females, however, attained a greater maximum age 1 42 yr I and size (TL=277 cm). We believe that exploitation has had a large but unquan- tifiable bias on gi'owth estimates for male and female Atlantic sturgeon. As the Hudson River population recovers, age structure and growth rate estimates should be refined to predict population recovery rates more accurately in the absence of a directed fishery. Age determination and growth of Hudson River Atlantic sturgeon, Aa'penser oxyrinchus* Jill T. Stevenson David H. Secor Center lor Environmental Science Chesapeake Biological Laboratory University ol Maryland PO Box 38 Solomons, Maryland 20688-0038 Present address (lor J Stevenson): Highly Migratory Species Management Division, F/SFl National Marine Fishenes Service 1315 East-West Highway Silver Spring, MD 20910 Email address (for J Stevenson) jill stevenson(g'noaa gov Manu-script accepted 13 January 1999. Fish. Bull. 97: 153-166 ( 1999). " Atlantic sturgeon iAcipeiiser oxyrin- chus) are a large anadromous fish that ranges the East Coast of North America and spawns in rivers from Florida to Canada. Population levels throughout the range of the species declined appreciably in the late 19th century owing to increased harvest of sturgeon for caviar following the Civil War (Murawski and Pacheco, 1977; Secor and Waldman, in press). Overfishing and deterioration of habitat, predominantly the blockage of spawning runs, have contributed to the extirpation of several Atlantic sturgeon populations (Taub, 1990; Waldman and Wirgin' ). The life his- tory strategy of the anadromous Atlantic sturgeon indicates that age structure and vital rates are espe- cially critical to conservation . Atlan- tic sturgeon exhibit high maximum age, late maturation (females 14-17, males 10-12; Van Eenennaam et al., 1996), and probable low mor- tality rates; growth, however, is rapid. These traits, as well as low rel- ative fecundity and less-than-annual spawning frequency, make sturgeon especially susceptible to overexploi- tation (Boreman, 1997). Therefore, models of Atlantic sturgeon popula- tion dynamics may be expected to be sensitive to biases in estimated vital rates and reproductive schedules. Atlantic sturgeon growth rates have been estimated in several studies, but results are divergent (Table 1). Poorly validated techniques have been employed to estimate age; and rates of growth, reproduction, and mortality have not been developed sufficiently to support resource man- agement models (Taub, 1990). Studies of acipenserid age have employed annuli in calcified struc- tures including scutes, pectoral-fin spines, otoliths, operculi, and other skeletal parts ( Harkness, 1923; Gree- ley, 1937; Brennan and Cailliet, 1989, 1991; Guenette et al., 1992; Rien and Beamesderfer, 1994). The term "fin ray," previously used to describe the leading ( primary) ray of the pectoral fin supporting element, was revised by Feindeis ( 1997 ) because this ele- ment becomes fully ensheathed with dermal bone early in ontogeny, and therefore should be termed a spine. Pectoral-fin spine sections have been preferred for aging because annuli in sections can be consistently inter- preted, and fin spines are easily col- ' Contribution 3272 of the University of Maryland Center for Environmental Sci- ence, Solomons, MD 20688-0038. Waldman, J. R., and I. I, Wirgin. 1998. Status and restoration options for Atlantic sturgeon in North America. ICES Coun- cil Meeting/T 16. 154 Fishery Bulletin 98(1) Table 1 Imprecision in aging studies of various long-lived species, reported as coefficient of variation (CV). Species Study CV Maximum age observed Atlantic sturgeon (Acipenser oxyrinchus) (fin spines) This study 4.8 42 White sturgeon (Acipenser transmontanus) Rien and Beamesderfer (1994) 7.8 104 Yellowfin sole (Pleuronectus asper) Kimura and Lyons (1991) 3.2 26 Sablefish Kimura and Lyons ( 1991 1 12.9 29 Pacific ocean perch (Sebastes alutus) Kimura and Lyons ( 1991 ) 4.9 78 Tarpon (Megalops atlanticusl Cyr ( 1991 ) 12.0 50 lected and processed without sacrifice of fish. They have been used in the past to age Atlantic sturgeon, but authors have described difficulty in aging older fish (>20 yr; Magnin, 1964; Huff, 1975; Dovel and Berggren, 1983). Aging based upon fin spines has been validated for white sturgeon, Acipenser trans- montanus (Brennan, 1988; Brennan and Cailliet, 1991) and lake sturgeon, A. fulvescens (Rossiter et al., 1995); however, no studies have validated the period- icity of annulus deposition in Atlantic sturgeon. Otoliths are often preferred for estimating fish age, but rate of otolith annulus formation has not been evaluated in sturgeons. Greeley ( 1937) enumer- ated ridges on the external surface of otoliths with- out examination of an internal section. Subsequent studies have indicated that annuli on an internal section of otoliths do not provide age estimates as precise as those from fin-spine sections (Schneberger and Woodbury, 1944; Brennan and Cailliet, 1991). Otolithic material, however, does not resorb, which is desirable for accurate aging, especially of long- lived fishes. The objectives of this study were to identify an appropriate calcified structure and develop a precise and unbiased method for determining age of Atlan- tic sturgeon and to model growth rates of Hudson River Atlantic sturgeon on the basis of juvenile and adult fish collected in 1992-96. Materials and methods Adult Atlantic sturgeon were collected during 1992- 95 from fishery harvests in the Hudson River and New York Bight, in cooperation with New York State Department of Environmental Conservation and New Jersey Department of Environmental Protec- tion. Fish were collected by using drift and anchored gill nets (25-36 cm stretched mesh) and otter trawls. During the period of collection, a minimum size limit of 152 cm total length (TL) was imposed on the New York fishery. In New Jersey, the minimum size limit of 107 cm TL was replaced with a 152-cm size limit in 1993. We collected pectoral-fin spines from sturgeon smaller than the minimum commercial size limit (<149 cm TL) in the Hudson River during 1993-95, using monofilament anchored gill nets (3—13 cm stretched mesh). Because few fish less than 152 cm were collected, other mid-Atlantic Bight regions were sampled. Fin spines were obtained from sturgeon collected from Chesapeake Bay commercial pound nets and gill nets in 1996 (n = ll), from National Marine Fisheries Service trawl surveys in the Mid- Atlantic Bight (1994 and 1996; n=4), and from U.S. Fish and Wildlife Service gillnet surveys of the Dela- ware Bay (/z=8). Because the Hudson River stock is the dominant reproducing stock in the Mid-Atlantic area (Waldman et al., 1996), subadults from areas other than the Hudson River were assumed to be predominantly of Hudson River origin. Removal and preparation of hard parts Pectoral-fin spines ( n =634 ) were removed at the point of articulation, air-dried, and sectioned less than one centimeter distal to the articulation point. Soft tissue adhering to the fin spines was allowed to decom- pose through microbial decay. A one-centimeter- wide section of each fin spine was then embedded in a block of Spurr epoxy, sectioned with an Isomet saw (Buehler, Lake Bluff, IL), and mounted on glass slides (see Secor et al., 1991). Some fin spines (64%) were not embedded but were sectioned with a jew- eller's saw. All sections were mounted with thermo- plastic glue on glass slides and polished with an automated poHshing wheel ( MINIMET 1000, Buehler, Lake Bluff, IL) with fine grit carborundum paper and a 0.3-)im alumina slurry on a polishing cloth. Final sections were 1-2 mm thick. Sagittal otoliths were removed from the severed heads of Atlantic sturgeon collected in 1994-95 Stevenson and Secor; Growth of Aapenser oxynnchus 155 (« = 114). Sagittae were cleaned in 107c bleach, rinsed several times with deion- ized water, and air-dried. One sagittal oto- lith from each pair was embedded in Spurr epoxy and sectioned as described in Secor et al. (1991). Owing to their fragility, oto- lith sections were polished by hand with a variety of fine sandpapers and a 0.03-pm alumina slurry. Annuli in thin sections of fin spines and otoliths were viewed under reflected light at 15x magnification by two experi- enced readers. An annulus was defined as a bipartite zone comprising an opaque and a translucent zone (Fig. 1 ). The first trans- lucent zone was counted as the beginning of the first year of life. In some instances, a secondary fin spine was embedded within the primary fin spine (Feindeis, 1997), in which case care was taken to enumerate annuli only in the primary spine. False annuli were consistently observed in fin- spine sections of older individuals and were excluded from annulus counts. These structures were not continuous around the entire circumference of the section and were thus distinguishable from annuli to be counted (Prince et al., 1985). Precision and bias Readers counted annuli without knowing collection date, fish size, or previous age determination. They were trained with the aid of an imaging system that permit- ted simultaneous observation of annular growth zones. Paired difference tests were used to statistically evaluate bias and pre- cision among readers in a single blind test. The coefficient of variation (CV) was used to measure precision; bias was assessed visually using age-bias plots (Fig. 2). An age-bias plot showed paired estimates of age for the same fish (Campana et al., 1995), with the estimates of reader 2 rep- resented as mean age, and 95';'f confidence intervals corresponding to each of the age classes estimated by reader 1. For exam- ple, if reader 1 estimated five fish to be 15 years old, the age-bias plot indicated the mean age of those five fish as estimated by reader 2. Divergence from the equiva- lence line, where A^e^,,,,/,, , = Age„„,^„. ^_, indicates a systematic difference between readers. Paired age estimates for either fin- translucent zones Figure 1 Backscatter electron micrographs show (A) a sagittal otolith section; IB) a pectoral-fin spine (lower lobe only) from a Hudson River Atlantic sturgeon; and (C) a fin-spine section from a hatchery-reared Atlantic sturgeon. White arrows (B) indicate annuli. Organic deposits in the fin- spine section are indicated with dark arrows (B). Pits resulting from chemical microanalysis are visible in the otolith section. Dashed circle in (A) indicates annuli that are difficult to interpret. 156 Fishery Bulletin 98(1) spine sections or otolith sections made by two readers were contrasted. In addition, for each reader, ages estimated from fin spine sections were compared with ages estimated from oto- hth sections from the same fish. Validation of annulus formation The periodicity of annulus formation in Atlantic sturgeon fin spines and otoliths was studied by using measurements and chemical microanal- ysis of marginal increments, and observation of annuli in juvenile hatchery-reared sturgeon marked with oxytetracycline (OTC) at a known age. Efforts were concentrated on fin-spine age validation because annuli were difficult to con- sistently interpret in otolith sections. Marginal increment analysis measures the opaque zone deposited after the last identifi- able translucent zone at the margin of a struc- ture used for age estimation (Kalish, 1995). Seasonal growth of that opaque zone is used to determine the timing of annulus formation (Cailliet et al., 1986; Beamish and McFarlane, 1987; Brennan and Cailliet, 1989). In our anal- ysis, marginal increments were measured with image analysis software (Optimas, Inc., 1994) to the nearest 0.001 mm. Marginal increment ratio (MIR) was calculated as MIR = MIx 1/A, where MI = the width of the outermost opaque zone (marginal increment); and A = the mean width of the three annuli deposited previous to the marginal increment. Mean MIR was computed for each month sepa- rately, with ages and sexes combined. Because very few fish were collected for several months, monthly samples were pooled for four three- month seasons. The winter season (December- February) was eliminated owing to very low sample size. Microchemical analysis of calcified struc- tures can verify the periodicity of annuli (Jones and Geen, 1977; Casselman, 1983; Radtke and Targett, 1984; Cailliet and Radtke, 1987). Calcium concen- tration is correlated with optically defined growth zones in the hard part (e.g. calcium is increased in the opaque zone and decreased in the translucent zone; Cailliet and Radtke, 1987; Lai et al., 1996). For instance, the most recently formed material in fin o O 41 A ^ 31 "^ / \ .^^^''' ""I - ■X CV=4.8% :,^ *i4^-^ 11 f**^^ 14 19 24 29 34 Fin spine age estimated by Reader 1 (years) 5 10 15 20 25 30 35 Otolith age estimated by Reader 1 (years) 24 16 - 10 20 30 40 Age estimated by fin spine section (years) Figure 2 Bias m age estimates for Hudson River Atlantic sturgeon are eval- uated by age-bias plots that compare estimates made by two read- ers interpretmg the same (Al fin-spine sections and (Bl otolith sections; and (Cl made by one reader interpreting fin-spine and otolith sections taken from the same fish. Dashed lines are "equiv- alence lines" indicating the hypothetical case where both estimates are identical. Error bars represent the 95'7f confidence interval about the mean age (reader 2) for all fish assigned a given age by reader 1. spines might contain high or low concentrations of cal- cium in fish collected during summer (rapid growth) or winter (slower growth), respectively (Cailliet and Radtke, 1987). In this instance, fin-spine microanaly- sis of calcium across annuli might show nadirs in cal- cium associated with translucent zones. Calcium concentration in fin-spine sections was measured using a JEOL JXA-840A wavelength-dis- Stevenson and Secor Growth of Aapenser oxynnchus 157 persive (WDS) electron microprobe at the Center for Microanalysis, College Park, MD. Phosphorus was also measured because it makes up a large fraction of the fin spine's hydroxyapatite structure ( Steven- son, 1997). Accelerating voltage was 25 kV and cup current was 20 nA. Each measurement represented rastering or scanning by the probe over a 5 x 5 jam area of the hardpart section. Molar weights of cal- cium and phosphorus were measured in peripheral regions of fin spines for 20 fish aged 3-18 years (fin- spine estimate) collected in March, June, September, and November. The mean of three peripheral points was calculated for each individual, and means were contrasted by month of collection with ANOVA. Microprobe measurements were used to construct elemental chronologies of 15 fin spine sections taken from sturgeon aged 12-36 years (fin-spine estimate) and 164-236 cm TL. For each chronology, a series of point measurements was taken along an axis that traversed several annuli, with five points per annu- lus; points were assumed to sample seasons in linear proportion. A solution of OTC (25 milligrams per kilogram of body weight) was injected into the dorsal muscula- ture of five juvenile laboratory-reared fish of known age (55—65 cm TL) from Hudson River broodstock. Juveniles had been reared in circular fiberglass tanks in a recirculating system and fed a mixture of commercial pellet feeds. They were subjected to a twelve-hour photoperiod; water temperature ranged from 15° to 20°C. Three months after injection, the first fin spine section was removed from three fish with a jeweller's saw, and silver nitrate was applied to encourage clotting of the wound. Fifteen months after injection, a second section was removed from the opposite pectoral-fin spine of one fish only. Fin- spine segments were dried, sectioned, mounted on glass slides, and polished. Thin sections were viewed with epifluorescent microscopy to identify OTC marks. The position of OTC marks with respect to opaque and translucent zones was recorded and a micro- graph was taken. In all cases, the reader was aware that the fish had received an injection of OTC but had no knowledge of the date of section collection. Growth Reported measurements of dressed carcasses were converted to total length based upon conversion met- rics derived for Hudson River Atlantic sturgeon (Ste- venson, 1997). Sturgeon included in the conversion metric sample (n=235l were collected during spawn- ing season in the Hudson River; all but five pos- sessed mature gonads. Ages derived from fin-spine and otolith sections were used to fit von Bertalanffy growth models by using a Marquardt iterative esti- mation procedure for the three model parameters. Length-at-age relationships were first examined to determine variance structure and progression of modal length with age. Growth parameters were estimated iteratively for males and females with a least-squares method. Because of high variance at the point of growth inflection, it was unlikely that a single growth model would fit all portions of the growth curve. Therefore, the juvenile portion (42-152 cm TL) was modeled separately with a power func- tion based upon the best fit of residuals. PC-SAS (SAS Institute, Inc., 1994) and Statgraph- ics Plus (STSC, Inc., 1992) were used for all statisti- cal tests. Data that did not satisfy the assumption of heteroscedasticity (Bartlett's test, a=0.05) were transformed to satisfy this assumption. Transformed data that did not satisfy this assumption were ana- lyzed with a Kruskal-Wallis nonparametric test to examine differences among groups. Results Comparison of hard parts Otoliths were irregularly shaped and their annuli were difficult to interpret. In contrasting several sec- tioning planes, we observed that annuli on transverse sections yielded the most consistent interpretations. The first three to nine growth zones showed a clear alternation of opaque and translucent zones. There- after, translucent zones were irregularly spaced and often appeared to overlap (Fig. lA). Low optical con- trast between opaque and translucent zones reduced the readers' confidence in assigning annuli, espe- cially in sections with more than twenty annuli. In contrast, fin-spine sections exhibited concentric narrow translucent zones and wide opaque zones when viewed with transmitted light. Fin spines con- tained a vascularized core and deposits of organic material in lobe regions (Fig. IB). Interspersed were fibrils that we interpreted as collagen or some other structural protein. Annuli became narrower toward the outer edge in larger (and presumably older) fish. Secondary fin spines (84% of fin-spine sections) and false annuli were observed but were simple to iden- tify and disregard. Belts of two to five narrow annuli were apparent in most female fin-spine sections (96% of a subsample of 48). These belts were not apparent in the juvenile sturgeon examined. Imprecision (CV) in age estimates was 4.8% between two readers of the same fin-spine section (Fig. 2A). Mean imprecision was 1.2 years (^=-1.97, P>0.05); estimates by the two readers were not significantly 158 Fishery Bulletin 98(1) different. Age estimates from otolith sec- tions were less precise than from fin-spine sections ( Fig. 2B). Absolute imprecision was 3.3 years and the CV on paired differences was 14.8%. No significant bias occurred in otolith interpretations between readers. In a comparison of ages estimated by a single reader from otolith sections and from cor- responding fin-spine sections (i.e. from the same fish), the former were significantly lower than the latter (Fig. 2C; mean dif- ference=5 yr, ^=9.01, P<0.05). The bias was most apparent for presumed by older fish. Validation of the fin-spine aging method Marginal increment ratios for fish collected in late winter and spring (February-April) were significantly lower than those for summer and fall (Fig. 3; Kruskal-Wallis test;P=0.04). Mean MIR was 18% in spring, 33% in summer, and 36%. in fall. The high- est rate of marginal increment completion was observed for winter months (Decem- ber-February), although sample size was very small for these months (n=3). Readings of annuli from hatchery-reared sturgeon resulted in exact age estimates. The sturgeon were 4-i- years old and annuli were clearly defined. There was a distinct difference in the shape of fin spines from hatchery-reared and wild juvenile stur- geon (Fig. IC). Hatchery-reared fish exhib- ited irregularly shaped, compressed annuli which could indicate erosion or injury, but were nonetheless easily recognized. In OTC-injected fish, a distinct OTC mark fol- lowed by an opaque zone was apparent in all fin spines examined three months after injection. The fish did not show any symp- toms of stress following fin-spine removal. The only sample examined at fifteen months after injection exhibited a clear OTC mark that was followed by opaque and translucent zones. Microchemical analysis The concentrations of calcium and phosphorus in peripheral regions of fin spines showed significant seasonality (Kruskal-Wallis test, P<0.04; Fig. 4). Fin spines collected in November were significantly lower in calcium and phosphorus than those collected in March, June, and September. Calcium-to-phosphorus ratios increased significantly from March through November (Kruskal-Wallis test, P=0.04). Plots of cal- 100 80 60 40 20 o e ^ 1 1 ^ 9 5 122 85 ^ i • • I 1 + 1 1 1 1 56 4 Feb. April June Aug. Month Oct. Dec. 50 40 30 B -r^ -p 20 - 10 Spring (Mar-May) Summer (June-Aug.) Season Fail (Sept. -Nov.) Figure 3 (A) Marginal increment ratios (MIR) ± one standard error for sectioned Hudson River Atlantic sturgeon pectoral fin spines evaluate current season's growth. Numbers represent sample size. (B) Marginal incre- ment ± ^b'7( confidence interval for data grouped by season. MIR = Ml X 1/A, where Ml = the width of the outermost opaque zone (mar- ginal increment) and A = the mean width of the three annuli deposited previous to the marginal increment. cium and phosphorus in fin spines revealed cyclical trends in both elements, with peaks associated with translucent zones (Fig. 5). Growth The size-age relationship for prespawning fish, as judged by interpretations of annuli in fin spines, was best fitted by a power regression (r-= 0.75; Fig. 6). Von Bertalanffy models for both males and females considerably underestimated length at age for sub- adults. Over all life history portions, the fit of the von Bertalanffy model was better for females (ages 2-42; r^=0.56) than for males (ages 4-36; r-=0.33) owing to the broader range of ages and lengths in our sample Stevenson and Secor: Growth of Acipenser oxyr/nchus 159 of females (Fig. 7A). Males grew faster but reached a smaller asymptotic length at a younger age than did females. Females grew more slowly (K=0.07) toward a significantly larger maximum length (251 ±12.8 cm). The asymptotic growth phase for females and males was 12-42 and 11-28 years, respec- tively. Estimates of growth coefficients, K, were significantly different between sexes (/=73.2, df=431, P<0.05). Log transformation of the data did not correct for the error struc- ture. Residuals from the models were desig- nated as corresponding either to fish of the size at entry into the fisheries ( 152 cm TL in New York and after 1993 in New Jersey), or to larger or smaller fish. The resulting resid- ual plots indicated that the faster-growing males were harvested just as they reached size at entry ( positive residuals) and that the slower-growing females entered the fishery at much older ages (negative residuals; Fig. 7B). The modeled growth pattern for females appeared more biased than that for males owing to a sharp shift in residuals from posi- tive to negative at about 15 years. At older ages, the von Bertalanffy model underesti- mated the size of females. These patterns in residuals were also apparent in males, albeit less pronounced. Discussion Fin spines or otoliths? Owing to ease of collection and processing, as well as precision and accuracy in aging, we rec- ommend the use of fin spines rather than otoliths for demographic analysis of Atlantic sturgeon. Prep- aration techniques used in this study (embedding, sectioning, and polishing) may have improved the visual resolution of annuli in pectoral-fin spines over that reported in earlier investigations. Annuli in sturgeon otoliths and, as a result, growth rates, have been grossly misinterpreted in the past owing to examination of the external surface of hard parts only (Greeley, 1937). Especially in presumed older individuals, otoliths should not be used to verify age estimates based on fin spines. Annular clarity diminishes towards the distal end of the oto- lith (more recent growth), which may result in lower age estimates. In addition, individuals must be sacri- ficed to collect otoliths. Concern over declining stur- geon populations should restrict the size of otolith samples. 31 ^ A (4) (6) (J^' ^ 30 r -] - i 29 i- . (6) 1 28 _ 1 - O 27 - 26 14.9 L B " 13.9 _ 1 ^ £ 12.9 1. _ _ 00 1 11.9 L -^ ^ - ~ r ^ 10.9 r 9.9 2.9 1 L c ? 2.7 ~ :|j 2.5 - T . T t 2.3 k -^ 1 a. T3 2.1 1 U 1 1.9 ^ Mairch June Sept. Nov. Month Figure 4 Microchemical analysis of marginal increments of fin spine sections taken from Hudson River Atlantic sturgeon show significant season- ality in (A) calcmm; (B) phosphorus; and (C) calcium-to-phosphorus ratio. Means and 95"^!: confidence intervals are depicted. Three points were analyzed per section. Numbers in parentheses indicate how many fin-spine sections were analyzed for fish collected in each of the four months (these apply to all three graphs). Precision and accuracy Age estimates based on fin spines were unbiased and precise. Aging imprecision for Atlantic sturgeon was similar or better than imprecision reported for other long-lived (>20 yv) species, for which precision is often affected by narrow annuli, which result from reduced growth rates in older fish (Rien and Beames- derfer, 1994; Table 1 ). The annual patterns of mar- ginal increment ratio (MIR), observations on fish of known age, and microchemistry have supported the hypothesis that a pair of opaque and translucent zones forms annually in Atlantic sturgeon fin spines. A small sample of OTC-marked juveniles also indicated annual annulus growth (Stevenson, 1997). Results for Atlantic sturgeon were similar to Huff's (1975) anal- ysis of Gulf sturgeon (Acipenser oxyrinchus desotoi) fin-spine sections, in which a higher percentage of fin-spine sections showed that a completely formed 160 Fishery Bulletin 98(1) U a. 142 132 12.2 112 10.2 9.2 36 h- 33 30 1- 27 24 21 18.1 - 16.1 - 14.1 12.1 10.1 8.1 Female- 18 vrs -224 cm 9 10 11 12 13 14 15 16 17 18 _1 L_ 9 10 11 12 13 14 15 16 17 18 Male- 20yrs - 187cm 8 9 10 11 12 13 14 15 16 16 Figure 5 Series of microprobe analyses for calcium and phosphorus across annuli in fin-spine sections were used to construct elemental chronologies for Hudson River Atlantic sturgeon. In these two representative chronolo- gies, dashed lines correspond with translucent zones. Stevenson and Secor: Growth of Aapenser oxyrinchus 161 160 140 • • 1 • \X' ' "Z - /■ • / / m / Total length (cm) 00 o to o o o / / / 5" / ../■■ • :/ : / " ' /■■■ y= 44.7 * xO''^ r2=0.75 « = 92 L r * 60 f / 40 / 1 , . . , 1 , , , . 1 , , , , 1 , . . , 1 . . , , 1 , , , . 1 5 10 15 20 25 30 Age (years) Figure 6 A power regression best describes the growth of juvenile and sub- adult Hudson River Atlantic sturgeon (TL<152 cm), based on ages derived from fin spine sections. marginal annulus was completely formed in the fall rather than in the spring. Ideally, marginal incre- ment analysis should be performed on separate age classes to ensure that annulus formation occurs for all ages (Casselman, 1983). However, owing to the rela- tive scarcity of older individuals and the unavailability of fish in some seasons and because marginal incre- ments are often difficult to discern in fin-spine sec- tions taken from older individuals, age classes were pooled in our study. Accuracy of age estimation at older ages remains untenable. An attempt to validate longevity estimates of Hudson River Atlantic sturgeon using radiometric ->0Pb/'--^Ra dating was unsuccessful (Burton et al., 1999). Recapture of hatchery-released Atlantic sturgeon (Secor-) has provided an opportu- nity to verify age determinations in older or mature individuals. Fin spines taken from juvenile sturgeon did not exhibit belts of annuli; such belts may provide information about spawning behavior in females. ^ Secor, D. H. 1998. Habitat utilization patterns of mid-Atlan- tic Bight juvenile Atlantic sturgeon. Final report on project 1445-CT-09-0189 to National Biological Survey, 30 p. Ref. no. [UMCES] CBL 98-019. lAvailable from David Secor, Chesa- peake Biological Laboratory, 1 Williams St., Solomons. MD 20688-0038.1 Chemical microanalysis of hard parts is a promis- ing tool for age verification (Jones and Geen, 1977; Casselman, 1983; Cailliet and Radtke, 1987). How- ever, seasonal cycles were not consistently observed in elemental chronologies of Atlantic sturgeon fin spines. The technique assumes that the hard part is a closed system, and there is no resorption or remod- eling. Burton et al. (1999) using radiometric tracers, found that Atlantic sturgeon fin spines appear to be open systems. In our study, we may have observed a corrupted seasonal signal. Growth Sexually dimorphic growth patterns in Atlantic stur- geon may be a result of differential reproductive schedules and migration patterns. The age-length relationship shows substantial variability in both males and females. Males mature earlier ( 12 yr) and spawn annually. Females mature later at a much larger size and are thought to spawn every 3-5 years ( Smith, 1985 1. Lower growth rates and larger achieved sizes are typical for large, long-lived fish that broad- cast numerous offspring (Adams, 1980; Moreau, 1987; Roff, 1988, Winemiller and Rose, 1992). 162 Fishery Bulletin 98(1) 3 -o Males L-inf= 180 + 2.3 cm K = 0.25 + 0.030 t-0 = 2.37 + 0.486 r-sq. =0.33 « = 301 Females L-inf=251 + 12.8 K = 0.07 + 0.011 t-0 = -3.23 + 1.33 r-sq. = 0.56 ^ n=255 cm A 300 - 250 ^.^ 1. 200 ^M ^ S r 3^M3~^r ^ 150 •■Ms i™^ In J-^ Males o 100 r' n=301 50 . 1 .... 1 .... 1 .... 1 . Females «=255 50 - ^ 30 D "^l - 10 3 ^ E^S Se,' ' - -10 - ^ -30 - ^y^^ „ ^ ^ -50 - - -70 " 80 D E) 50 - 20 KI t'J^U i- u ' -10 -40 -70 , 10 20 30 40 50 10 20 30 40 50 Fin spine age (years) Figure 7 lA) Predictions of length at age from von Bertalanffy models for male and female Atlantic sturgeon (black curves) fitted observed length at age (grey squares) for females better. (B) Plots of residuals from the growth models reveal a difference between fish of size-at-entry (152 ±5 cm TL; black squares); and fish of TL above or below size-at-entry (gray squares). Note different y-axis scales for males and for females. Ages derived from fin-spine sections. Growth is difficult to model accurately in long- lived species (Mulligan et al., 1987). Because little bias was detected in our readings (aging precision was unaffected by age), discrepancies in age esti- mates should not introduce systematic error in esti- mates of population parameters such as mortality rate, K, and L^. High variability in observed length at age indicates that length may be a poor predictor of age for Atlantic sturgeon. This high variability may result from amplification of early growth dif- ferences over a long life span. Variance in size at age may also have resulted from the nature of the Hudson River Atlantic sturgeon fishery. The imposed minimum size limit on Atlantic sturgeon commercial fisheries (>152 cm TL) may have biased the sample used for growth estimates and could have caused K and L^ to be over- and under-estimated, respec- tively (Fig. 8). The von Bertalanffy curve may have been driven upwards at younger ages by the poten- tial maximum growth rate of the population. For instance, as soon as the fastest growing members of an age class exceeded the minimum size limit, they were harvested. At later ages, the curve is pulled downward as slower growing individuals enter the Stevenson and Secor; Growth of Aapenser oxyrinchus 163 Males Females mean = 1 75 + 1 .30 cm mode = 152 cm mean = 217 + 3.49 cm mode= 2 1 8 cm nJ cm 240 280 120 Total length (cm) 160 200 240 280 B mean =17 + 0.278 yrs mode= 1 6 years an = 23 + 0.913 yrs mode= 19 years 50 - sz 40 S 30 o - % 20 - |,0 A . 12 10 8 6 41- 2 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45 Age (years) Figure 8 (A) Total length and (B) age of male and female Atlantic sturgeon caught in the Hudson River gillnet fishery (1992-95). Sublegal fish (<152 cm) were a result of a minimum dressed carcass length of 91 cm. Conversion from dressed length to total length may have resulted in "undersized" fish. fishery (as evidenced by residual patterns), increas- ing K and decreasing L„. Growth parameter estimates reported here for females are consistent with another study of the same population (Doroshov et al.'^) but do not agree with an earlier study (with sexes combined) of the popu- lation (Dovel and Berggren, 1983). Recent reduction of the accumulated biomass by fishing, and resul- tant age and size truncation of the population, would cause lower apparent L _^ and higher K values than in an unexploited population. Maximum length (L^) determined for females was substantially smaller than historical records of maximum size for this spe- cies (427 cm; Murawski and Pacheco, 1977), which may be a result of increased fishing pressure on the * Doroshov. S., J. Van Eenennaam, G. Moberg, and G. Waton. 1994. Reproductive conditions of the Atlantic Sturgeon ^Acipen- ser oxyrinchus) stock in the Hudson River. Report for year two to Hudson River Foundation. Animal Science Department, Uni- versity of California, Davis, 65 p. largest (female) component of the population during the past ten years. The historical longevity of females in the Hudson River also probably exceeded our esti- mate of 42 years; Magnin ( 1964) reported a 60-year- old female in the St. Lawrence estuary. Because the largest male sampled ( 72 kg) was only 22 years of age, yet much smaller males were older than 30 years, male longevity in the unexploited population was also probably higher than in the pop- ulation we sampled. Consideration of possible bias in von Bertalanffy growth parameters should direct managers to undertake sensitivity analyses in devel- oping demographic-based stock assessments. Because L„ and K are inversely related (Kimura, 1980), age and size truncation would result in an over-estimated K (a lower L^^ is approached more rapidly). We found L^ values that were relatively low com- pared with estimates in previous studies of Hudson River Atlantic sturgeons (Greeley, 1937; Dovel and Berggren, 1983). These low values may indicate 164 Fishery Bulletin 98(1) age truncation of the population by directed fishing effort. However, Dovel and Berggren ( 1983) reported difficulty in aging fish greater than 153 cm TL and may have misidentified annuli in both juveniles and adults. Greeley's (1937) study was based on enumer- ating annuli in otoliths. Likewise, K estimates from our study are higher than those estimated previ- ously. Our growth estimates are consistent, however, with those of Doroshov et al.,^ who aged the same population in the early 1990s. Comparative studies of fish populations along a latitudinal gradient have shown an inverse relation between latitude and rates of growth and mortality (e.g. Leggett and Carscadden, 1978; Jennings and Beverton, 1991). Among Atlantic sturgeon popu- lations, the most rapid growth was exhibited by fish sampled in southern latitudes. Maximum size increased with increasing latitude, which may indi- cate postponement of maturation (Magnin, 1964). Values of Jf*L for the Hudson River population did not differ significantly from those for any other popula- tion (a=0.05), whereas L„ values estimated for the Hudson River population were statistically differ- ent from those estimated for all other populations (a=0.05) along a latitudinal gradient (Table 2). Implications for stock restoration Growth rates of acipenserid species are apparently affected by migratory behavior (Roff, 1988). Anad- romous species of sturgeons exhibit higher growth coefficients than do semi-anadromous or freshwater resident species, although the semi-anadromous Table 2 Growth parameters have been calculated in various stud- ies of Atlantic sturgeon populations (combined sexes). For Smith (1985), n=number of age classes. Smith's (1985) data were converted from fork length to total length (FL= 111- TL; Beamesderfer, 1993 ) and were presented as mean length by age class in the literature. Location Study K ijcml St. Lawrence River Magnm, 1964 582 0.03 315 Kennebec. ME Hudson River, NY Winyah Bay, SC Suwannee River, FLA Smith, 1985 This study Smith, 1985 Smith, 1985 7 0.06 236 634 0.08 225 24 0.12 242 17 0.14 184 white sturgeon achieves a higher maximum length than do anadromous sturgeons (Table 3). Atlantic sturgeon undergo an ontogenetic habitat shift from estuarine nursery grounds to marine waters (Gil- bert, 1989). We believe these species grow rapidly as juveniles (as shown by their high /f coefficient) to outgrow predation in marine waters. Unfortunately, rapid gi-owth may affect the timing of juvenile migra- tion and increase their susceptibility to coastal fish- eries at an early age. Estimates of growth traits for Hudson River Atlan- tic sturgeon indicate that they are extremely vulner- able to overfishing. Slow growth following matura- tion, high longevity, and presumed low natural mor- tality indicate that year classes in this population reach their maximum biomass at relatively old ages (Alverson and Carney, 1975). Therefore, maximum fishing yield from sturgeon would be achieved at low - Table 3 Growth parameters (von Bertalanffy parameters unless otherwise indicated) estimated for North American sturgeons . Parameters were estimated for both sexes combined except where otherwise noted. Species Study Location n L, (cmTL) K 'o Atlantic sturgeon This study' Hudson River, NY 634 256 ( female 1 0.07 -3.2 (A. oxvrinchus) 180 (male) 0.25 2.4 White sturgeon DeVore et al., 1995 Unimpounded Columbia River, OR 783 307 0.03 -1.13 (A. transmontanus) Green sturgeon USFWS' Klamath River, CA 173 238 0.05 2.0 (A. medirostris) Shortnose sturgeon Dadswell, 1979 St John's River, Canada 446 144-' 0.04 2.0 (A. hrevirostris) Lake sturgeon Thuemler, 1985^ Menominee River, WI-MI 1464 156 0.07 -0.02 (A. fulvescens) ' Growth parameters estim ated for males and females separately. - United States Fish and Wildhfe Service. 1982. Klamath River Fisheries Investigation Program annual report, p. 123-141. * Converted from fork length (F'L= 1.11 x TL; Beamesderfer, 1993). ■* Growth parameters derived from plot of mean TL at age. Stevenson and Secor: Growth ol Acipenser oxynnchus 165 rates of fishing or harvest of large, mature individu- als (Bullock et al., 1992; Stevenson, 1997). A mora- torium on commercial sturgeon fishing in the New York Bight was instituted in 1996; however, Atlantic sturgeon continue to be caught incidentally in many coastal fisheries. Continued research and monitor- ing of recruitment and bycatch mortality will be nec- essary to ensure the restoration of this population. Acknowledgments We would like to thank I. Burliuk, D. Bush, E. Nack, and S. Nack, commercial sturgeon fishermen, as well as M. Mangold of the U.S. Fish and Wildlife Service (USFWS), J. Fields (National Marine Fisheries Ser- vice), and J. Mohler (USFWS), for their help in the collection of sturgeon fin spine samples. J. Van Eenen- naam (University of California, Davis), J. Mohler and M. Mangold (USFWS), K. McKown and B. Young (New York State Department of Environmental Consei-va- tion), and B. Andrews (New Jersey Department of Environmental Protection) provided additional previ- ously collected fin spine samples. M. Bain (Cornell University) provided resources for sampling subadult sturgeon in the Hudson River. J. Van Eenennaam pro- vided valuable insight on aging techniques for white sturgeon and Atlantic sturgeon. Anne Henderson-Arza- palo reared and sampled juvenile known-age sturgeon for validation. The Hudson Riverkeeper provided on- site laboratory facilities. This research was funded by a grant to D. Secor by the Hudson River Foundation. Literature cited Adams, P. B. 1980. Life history patterns in marine fishes and their con- sequences for fisheries management. Fish. Bull. 78:1-11. Alverson, D. L., and M. J. Carney. 1975. A graphic review of the growth and decay of popula- tion cohorts. .J. Cons. Int. Explor. Mer 36;133-143. Beamesderfer, R. C. 1993. A standard weight (Ws) equation for white sturgeon. Calif Dep. Fish Game 79:6.3-69. Beamish, R. J., and G. A. MacFarlane. 1987. Current trends in age determination methodology. In P. C. Summerfelt and G. E. Hall leds.). The age and growth offish, p. 35-42. low-a State Univ. Press, Ames. lA. Boreman, J. 1997. Sensitivity of North American sturgeons and paddlefish to fishing mortality. Environ. Biol. Fishes 48:399—405. Brennan, J. S. 1988. Age determination and verification of California white sturgeon [Acipenser transmontanus): a comparative analy- sis. M.S. thesis. San .Jose State Univ. CA, 70 p. Brennan, J. S., and G. M. Caiiliet. 1989. Comparative age-determination techniques for white sturgeon in California. Trans. Am. Fish. Soc. 118:296-310. 1991. Age determination and validation studies of white stur- geon in California. In P. Williot (ed. ), Acipenser: sectes du premier colloque international sur I'esturgeon, p. 209-234. French Institute of Agriculture and Environmen- tal Engineering Research (CEMAGREF), Gazinet, F'rance, Bullock, L, H., M. D. Murphy, M. E. Mitchell. 1992. Age, growth, and reproduction of jewfish, Epincph- elus itajara, in the eastern Gulf of Mexico. Fish. Bull. 90:243-249. Burton, E. J., A. H. Andrews, K. H. Coale, G. M. Caiiliet. 1999. Application of radiometric age determination to three long-lived fishes using -"'Pb;--''Ra disequilibria in calcified structures: a review. /;; J. A. Musick (ed. I, Life in the slow lane: ecology and conservation of long-lived marine ani- mals, p. 77-S7. Am. Fish. Soc. Symp. 23. Caiiliet, G. M., and R. L. Radtke. 1987. A progress report on the electron microprobe analysis technique for age determination and verification in elas- mobranches. In R. C. Summerfelt and G. E. Hall (eds.), The age and growth offish, p. 359-369. Iowa State Univ. Press, Ames, lA. Caiiliet, G. M., R. L. Radtke, and B. A. Welden. 1986. Elasmobranch age determination and verification, a review. In T. Uyeno, R. Arai. T. Taniuchi, and K. Matsu- ura (eds.). Proceedings of the second international confer- ence of Indo-Pacific Fishes, p. 345-360. Campana, S. E., M. C. Annand, and J. I. McMillan. 1995. Graphical and statistical methods for determining the consistency of age determinations. Trans. Am. Fish. Soc. 124:131-138. Casselman, J. M. 1983. Age and growth assessment offish from their calcifi- cation tissue-techniques and tools. In E. D. Prince and L. M. Pulos leds.l. Proceedings of the international workshop on age determination of oceanic pelagic fishes, tunas, bill- fishes, and sharks, p. 1-17. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 58. Cyr, E. C. 1991. Aspects of the Ufe history of the tarpon. Megalops atlantuus, from South Florida. Ph.D. diss., Univ. S. Caro- lina. Columbia, SC, 60 p. Dadswell, M. J. 1979. Biology and population characteristics of the short- nose sturgeon, Acipenser 6ret'i>osfscf'«.s, in the Meno- minee River, Wisconsin-Michigan. Environ. Biol. Fishes 14:73-78. Van Eenennaam, J., S. I. Doroshov, G. P. Moberg, J. G. Watson, D. S. Moore, and J. Linares. 1996. Reproductive conditions of the Atlantic sturgeon (Acipenser oxyrinchus) in the Hudson River Estuaries 19:769-777. Waldman, J. R., J. T. Hart, I. I. Wirgin. 1996. Stock composition of the New York Bight Atlantic sturgeon fishery based on analysis of mitochondrial DNA. Trans. Am. Fi.sh. Soc. 125:364-371. Winemiller, K. O., and K. A. Rose. 1992. Patterns of life-history diversification in North Ameri- ican fishes: implication for population regulation. Can. J. Fish. Aquat. Sci. 49:2,196-2,218. v 167 Abstract.— Although juvenile fish are studied extensively in estuarine and nearshore environments, surprisingly little is known about the basic habitat requirements of juveniles for offshore settlement and nursery areas. Between June 1996 and July 1997, settlement and nursery habitats of age-0 (early juvenile) demersal fish on the conti- nental shelf of the New York Bight were investigated by using a two-meter beam trawl. Replicate tows at 21 sta- tions along three cross-shelf transects (20-95 m depth), were sampled on a near monthly basis to determine gen- eral ecology (21,309 fish collected in 659 tows). Of the 47 species collected, 33 included age-0 juveniles, and 25 included near-settlement size individuals. The two dominant species, Pleuronectes fer- rugmeus and Merluccius biliuearis, con- stituted 88.9'/f of the total catch of age-0 fish. Of all age-0 fish, 94"^ were collected during summer and fall. Comparisons of weighted means and the use of canoni- cal correspondence analysis determined that settlement and nursery habitats across the shelf are primarily delin- eated by depth, temperature, and time of year. Three zones across the shelf (inner, middle, and outer) each had dis- tinct juvenile fish assemblages. Knowl- edge gained about the distribution and quality of juvenile habitat for commer- cially important offshore species should facilitate their improved management. Settlement and nursery habitats for demersal fishes on the continental shelf of the New York Bight* Brian P. Steves Marine Sciences Research Center State University of New York at Stony Brook Stony Brook, New York 11794-5000 Present address: Manne Invasions Laboratory Smithsonian Environmental Research Center 648 Contees Wharf Road Edgewater, Maryland 21037 E-mail address Stevesg'serc siedu Robert K. Cowen Marine Sciences Research Center Stale University of New York at Stony Brook Stony Brook, New York 11794-5000 Mark H. Malchoff New York Sea Grant Program 3059 Sound Avenue Riverhead, New York 1 1901 Manuscript accepted 7 December 1998. Fish. Bull. 98:167-188 ( 1999). With the decrease in fish abun- dance in the latter half of the twen- tieth century, particularly in the Northwest Atlantic (McHugh, 1972; NEFSC, 1992), fisheries managers have been concerned with both over- fishing and habitat degradation. Much effort has been put into under- standing the abundance, distribu- tion (Colton, 1972; Colvocoresses and Musick, 1984), and environ- mental preferences (Scott, 1982; Auster et al., 1991; Felley and Vecchione, 1995) of adult ground- fishes in the northwestern Atlan- tic. Although information on adult groundfishes is useful, events during the early life history of fish may be more important in determining recruitment variability (Sissen- wine, 1984; Houde, 1987; Peterman et al., 1988; Bradford, 1992; Miller, 1994). Several ichthyoplankton sur- veys have helped to increase our understanding of egg and larval distributions of groundfishes in the Mid-Atlantic Bight (MAB; Morse et al., 1987; Cowen et al. 1993). However, less is known about the juvenile stage, which represents a dramatic change in lifestyle for groundfishes: they leave the three- dimensional environment of the plankton and settle onto the two- dimensional world of the sea floor (Chambers and Leggett, 1992). Within this two-dimensional envi- ronment, the growth, survival, and recruitment of groundfishes are affected by various factors associ- ated with the quality (value for growth) and quantity (area) of their nursery habitat (Gibson, 1994). Re- search involving the nurseries of groundfishes has been limited to estuaries and nearby coastal habi- tats where they can readily be stud- ied (Riley et al., 1981; Able et al., 1989; Bolle et al., 1994; Henderson and Seaby, 1994; Nash et al., 1994; ' Contribution 1179 of the Marine Sciences Research Center, State University of New York at Stony Brook, Stony Brook, NY 11794-5000. 168 Fishery Bulletin 98(1) but see Toole et al., 1997). However, Able et al. ( 1989) have suggested that there is a lack of studies that examine the continental shelves of the MAB as poten- tial nursery grounds. This is remarkable because the early life history stages of many species of groundfish are believed to inhabit almost the entire shelf as well as the slope (Fahay, 1983; Miller et al., 1991). A wide assortment of biotic and abiotic variables may have a role in determining the distribution of continental shelf nursery grounds. Variations of abi- otic parameters, such as temperature, salinity, and oxygen, can affect the metabolism of marine flat- fishes during the juvenile stage (Malloy and Tar- gett, 1991: Pihl et al., 1991; Gibson, 1994; Neill et al., 1994). Changes in temperature, for example, can have a marked effect on feeding rate and growth; thus lower temperatures may slow growth and con- sequently increase susceptibility to mortality due to size-selective predation (Van der Veer et al., 1994). Salinity seems to have a small influence on the growth rate of fish, but it is often effective in con- trolling their distribution (Rodgers, 1992). By adjust- ing their position in relation to local abiotic factors, fish may modify their gi'owth rate and survival with respect to average environmental parameters (Gibson, 1994). Biotic factors, particularly food resources and shel- ter related to biological activity, may also be a potent determinant of nursery habitat. Auster et al. (1991) found that several biotic microhabitat types, includ- ing amphipod tube mats, shells, and biogenic depres- sions, significantly affected the abundance of fishes and other megafauna. Other biotic factors that influ- ence the distribution of juvenile fishes include the presence of potential predators (Bailey, 1994) and the availability of potential prey species. Biotic fac- tors may be ephemeral; however their presence or absence at a given time and space may control the distribution of associated fish species. The spatial distribution of amphipod tubes in particular have been shown to have a strong influence on the abun- dance of age-0 silver hake {Meiiiicciii^ bilinearis) during fall in the MAB (Auster et al., 1997). Recent recruitment research has emphasized fac- tors that affect survival of the early life history stages of fishes. Identification of the habitats preferred by juvenile fishes in offshore waters will allow for a more complete understanding of recruitment varia- tion. Although shelf habitat is magnitudes larger in area than the nearshore habitat traditionally stud- ied, there is currently little knowledge of its role as a nursery. For many flatfish species, the size of nurs- ery habitat and year-class strength have been cor- related (Rijnsdorp et al., 1992; Gibson, 1994). Our objectives were to provide a first-order analysis of the species that use the shelf as settlement and nursery habitat during the course of a year and to address the relation of these distributions to environmental correlates. ^ Description of the study area The New York Bight (NYB), a central portion of the MAB, encompasses an area of 39,000 km^ on the East Coast of the United States. Its boundary extends out to the 200-m isobath between Montauk Point, New York, and Cape May at the southern end of New Jersey. The bathymetry of the NYB varies on sev- eral geographical scales. Although the continental shelf in this area is, to the most extent, gently slop- ing, large-scale features do exist. The Hudson River Canyon, which almost bisects the NYB. is the most obvious of these. Imposed on this bathymetry are the convoluted isobaths of the ridge and the swale topography that dominates the NYB (Freeland and Swift, 1978). Surficial sediments range from fine sand along the southern shore of Long Island to pebbly gi-avel off the coast of New Jersey ( Schlee, 1964). On a somewhat smaller scale, the sediments vary in color from yellow ochre to greenish gi'ay and in composition from biogenic calcium carbonate to quartz and feldspar (Freeland and Swift, 1978). The hydrography of the NYB changes seasonally. During the winter months, the entire water column is well mixed, but the coldest waters are found near- shore (Bowman and Wunderlich, 1977). Stratifica- tion begins in the spring and peaks in the summer. During stratification, a body of cold dense water, known as the cold pool, remains trapped on the bottom under the pycnocline on the midshelf This cold pool foi'ms a distinct band of midshelf temper- ature minima (about 4-5°C in midsummei) from Georges Bank to Cape Hatteras, persisting from late spring to early autumn (Houghton et al., 1982). Out- side the cold pool, bottom temperatures in the NYB range inshore from less than 1"C in the winter to above 21°C in the summer. Deeper bottom water, near the 100-m isobath, is less variable, ranging from 7° to 12'=C (Ketchum and Corwin, 1964). Salin- ity in the NYB varies less, with only a mild seasonal cycle. The lowest salinities (<31 psu) are found near the apex of the NYB, southeast of the Hudson River. However, salinities as high as 35 psu are found at the 200-m isobath. As with temperature, the largest seasonal fluctuations in salinity are found nearshoi-e ( Bowman and Wunderlich, 1977 ). Overall, nearshore habitats are less stable than those offshore in terms of both salinity and temperature. Steves et a\ Settlement and nursery habitats for demersal fishes 169 Materials and methods Sample collection Ten sampling cruises were conducted between June 1996 and July 1997 on board an 85-foot commercial fishing vessel, the Illusion. Summer and fall cruises were conducted monthly, and winter and spring cruises were conducted every other month (Table 1). The 21 stations sampled during each cruise were arranged in three transects (west, central, and east), each with seven stations ranging in depth from about 20 meters to 90 meters ( Fig. 1 ). Stations on each tran- sect were located according to bathymetrj' so that six stations were distributed evenly across the range of depths (one approximately every 15 m of depth). The seventh station was placed to fill any large distance between stations that was due to variability in the slope. The distance from shore of stations at a par- ticular depth varied between transects. Not all stations were sampled during every cruise (Table 1). Owing to weather, only five stations were sampled during cruise 6 (December 1996) and only two of the three transects were sampled during the February 1997 and April 1997 cruises. Because less than 25^ of the stations were sampled during cruise 6, we did not include these data in some of the analy- ses. However, cruises 7 and 8 were included because of their broader range of coverage (between the two cruises all three transects were covered at least once). Temperature and salinity data were collected after trawling at each station by using an internally record- ing conductivity-temperature-depth (CTD) probe (Applied Microsystems Inc. model AMS-STD 12). Care was taken to collect data from as close to the bottom as possible without risking damage to the CTD Because a backup CTD was unavailable and the CTD used was not always reliable, gaps in the physical data exist. Expendable bathythermograph (XBT) data collected nearby in early August by the MV Oleander as part of the NOAA Ship of Oppor- tunity Program (SOOP) were substituted for miss- ing temperature data for the central transect in that month. Juvenile groundfishes were collected with a modi- fied 2-m beam trawl with 4-mm stretch mesh net ( and 5-cm stretch mesh outer net for chaffing), towed at about 2-2.5 knots. The addition of a meter wheel allowed us to measure the area swept as the dis- tance trawled, multiplied by the width of the trawl (as in Carney and Carey, 1980). At each station, three 5-min tows were made; a fouled trawl was dis- counted and another trawl was repeated. The mean area swept during a 5-min tow was 698 m- ±35.1 m^ (95% CI). 41 5 Figure 1 Twenty-one stations ( numbers 1 on the continental shelf of the New York Bight were sampled with a two-meter beam trawl. Table 1 Dates and sampling stations in the New York Bight for each of the ten cruises in this study. For each of the three transects, numbers indicate the stations sampled on a given cruise (sample locations are depicted in Figure 1). Cruise West Center East number Dates transect transect transect 1 14-17 Jun 1996 8-13 16-21 1-7 2 9-11 Jul 1996 8-14 15-21 1-7 3 9-11 Aug 1996 8-14' 15-21- 1-7' 4 20-22 Sep 1996 8-14 15-21 1-7 5 25-27 Oct 1996 8-14 15-21 1-7 6 10 Dec 1996 — — 1-5 7 26-27 Feb 1997 8-14 — 1-7 8 16-17 Apr 1997 8-14 15-21 — 9 11-13 Jun 1997 8-14 15-21 1-7 10 15-17 Jul 1997 8-14 15-21 1-7 ' During the August cruise, no temperature or salinity data were collected. - The center transect was supplemented by XBT data. A 2-m beam trawl is the recommended standard trawl for juvenile groundfish research because its fixed width allows for ready quantification of the area trawled (Kuipers et al., 1992). Our trawl had skids that were heavier than those on most stan- 170 Fishery Bulletin 98(1) dard 2-m beam trawls in order to reduce the amount of time the trawl took to reach the bottom and to ensure that it was heavy enough to remain there. To accommodate the increased weight (about 38 kg) and prevent the trawl from digging into the sedi- ment, the skids were also longer (0.85 m) and wider (0.1 m). Typically, 2-m beam trawls are not used in depths greater than seven meters (Rodgers, 1992; Gibson, 1994; but see Pearcy, 1978); however, these modifications enabled us to trawl easily in waters as deep as 100 m. Fish samples were sorted on deck and all recently settled juvenile fishes (age-0) were preserved in 957c denatured ethanol. These fish were measured in the laboratory to the nearest tenth of a millimeter standard length (SL) with digital calipers. Older fish were mea- sured on deck to the nearest millimeter and returned to the water With the exception of Sebastes and Uro- phycis spp., all age-0 fishes were identified to species. Fish stages were identified according to morphologi- cal changes such as squamation and, for flatfish, eye migration, as well as analysis of length freo.uencies of cohorts. The relative abundance and composition of the remaining sample (shell hash, sand dollars, and shrimp) were estimated from subsamples. Length-frequency distributions of each species col- lected were calculated for each cruise. To account for differences in sampling effort among cruises, abun- dance data were standardized to the average cruise, which had 63 tows (21 stations x 3 tows each) last- ing five minutes each. Length frequencies were then used to estimate the size range of the age-0 fishes collected during each cruise, as well as to infer some growth rates of the age-0 cohort between cruises. Distribution analysis Abundance data were standardized to the average number offish per 1000 m'^. Because we were specifi- cally interested in early juveniles, and adults were not efficiently collected, we limited our analysis to age-0 fishes. All species of age-0 fishes collected were included in the analyses. The distribution of age-0 fishes was evaluated with respect to a suite of envi- ronmental data (bottom temperature, bottom salin- ity, and depth) by comparing weighted means for each species (Scott, 1982). For a given species, an environmental variable such as temperature was weighted by the abundance of that species at each sampling station. The sums of each weighted vari- able were then divided by the total abundance of that species collected to yield a mean value for that species on that environmental parameter. Cross-shelf migrations with ontogeny are a common feature of demersal fishes, particularly flatfish (Riley et al, 1981; Toole et al., 1997). To determine changes in cross-shelf distribution with size, weighted mean depths by size class were calculated for the more abundant species collected. In particular, distinct dif- ferences between th^ location of near-settlement size fishes and larger age-0 fishes might indicate migra- tion between settlement and nursery habitats. Environmental correlates Distribution and abundance of fishes in relation to underlying multivariate environmental gradients were analyzed by canonical correspondence analy- sis (CCA) by using the software program CANOCO (ter Braak, 1992). This analysis has been widely applied in the field of community ecology (ter Braak, 1995; ter Braak and Verdonshot, 1995; Rakocinski et al., 1996); it entails reciprocal averaging of species and environmental data based on the assumption of a unimodal response of species abundance to the environment (Palmer, 1993; ter Braak, 1995). In a comparison of ordination techniques. Palmer (1993) found that CCA was robust, being less susceptible to spurious results such as the "arch effect" often common to principal components analysis (PCA). Furthermore, Palmer's (1993) simulations of CCA illustrated that noisy or skewed species data could be compensated for, and a variety of data types and sampling design were possible. In total, 25 environmental variables (Table 2) were sampled during the cruises and included in the CCA analysis. Bottom temperature and salinity (from CTD casts), station depth (from an onboard depth sounder), latitude and longitude (global positioning system [GPS]), and distance from shore (nautical charts) were employed as continuous measures during the analy- sis. The relative abundances of nonfish constituents collected by the trawl were also considered to be envi- ronmental variables. Information on surficial sediment character at each station, another environmental vari- able we employed, was obtained from published data from the Marine EcoSystems Atlas (MESA) program's New York Bight Project (Freeland and Swift, 1978). Although these data are independent of this study, the MESA sampling area was the same and the spatial resolution of its sediment sampling was fine enough to obtain general information suitable for our analysis. The data on nonfish trawl constituents and surficial sed- iments were entered into the analysis as scaled values (i.e. as a number describing relative abundance among stations and tows). Because collections of both species and environmental data were made year-round, sea- sonal variables were also included and coded in the anal- ysis as a set of nominal variables: spring (March-May), summer (June- August), fall (September-November), Steves et al : Settlement and nursery habitats for demersal fishes 171 Table 2 List of environmental variables used in the canonica correspondence analysis (CCA). Environmental variable Code Type Units Spring Spr seasonal true-false Summer Sum seasonal true-false Fall Fall seasonal true-false Winter Win seasonal true-false Latitude Lat physical " latitude Longitude Long physical " longitude Bottom temperature T physical " Celsius Bottom salinity Sal physical psu Distance from shore D physical km Depth Z physical m Median grain size MePhi physical -scale % organics in sediment %Org physical percent % calcium carbonate in sediment %CaC03 physical percent Northern sea stars, Asterias forbesi noss biological relative volume of sample Sand dollars, Echinarchnius parma snd biological relative volume of sample Margined sea stars, Pontaster tenuispinus marg biological relative volume of sample Small shell fragments hash biological relat ve volume of sample Large mollusk shells and shell pieces shell biological relat ve volume of sample Sea scallops, Placopecten magellanicus scall biological relative volume of sample Sand shrimp, Crangon septemspinosa crang biological relative volume of sample Boreal shrimp, Pandalus montagui borsh biological relative volume of sample Rock crabs. Cancer spp. crab biological relative volume of sample Amphipods amph biological relative volume of sample Fig sponges, Subentes ficus spon biological relative volume of sample Large biogenic tubes tube biological relative volume of sample and winter (December-February). Each environmen- tal variable was standardized to have a mean of zero and a standard unit of variance so that all variables had equal weight despite differences in the scales of their usual units. Although seasonal patterns are likely to be impor- tant in juvenile fishes, these patterns are likely to be influenced by the spawning patterns of adult fishes rather than by habitat selection of settling fishes. These seasonal effects may dilute the variation explained as due to other factors. Some environmental factors such as temperature, salinity, and the presence of ephemeral biota that vary with season are possibly of more inter- est than simply accounting for settlement timing. To this end, a partial CCA was conducted using season as a covariable. Partial CCA allows environmental vari- ables to be treated as covariables, thus removing their influence from the rest of the data set (ter Braak and Verdonschot, 1995). A forward selection of environmental variables was used to select a minimum set of environmental variables that best explained the distribution and abundance of age-0 fishes (ter Braak, 1992). For- ward selection begins by the selection of the variable that explains the most variation in the data set. This variable is then treated as a covariable, the data set is then reevaluated, and the next most important variable is selected. This procedure continues until the desired subset of variables has been selected. To test the advisability of adding variables, Monte Carlo permutation simulations are conducted. Vari- ables are added as long as their addition continues to contribute significantly to the explained variance (P<0.05). Forward selection was performed on the data set, with season treated as a covariable. The combination of partial CCA and forward selection is not considered problematic with regard to the assump- tions of the analysis (ter Braak, 1992). Results Faunal composition and distribution A total of 21,309 fish representing 47 species and 32 families were collected over the course of the ten cruises and the 659 tows. Of these 47 species, 33 were collected as age-0 juveniles, including 25 spe- 172 Fishery Bulletin 98(1) Table 3 Minimum, mean, and maximum standard length (mm) of the 32 species of age-0 demersal fish collected. Maximum larval size (Fahay, 1983) and sample size in) are also indicated. Maximum sizes estimated from size-frequency distributions are included (see Figs. 4-6). Species are ordered by decreasing total abundance. An asterisk next to minimum standard length indicates that the species was collected at near-settlement size. Max larval size Mm. SL Avg. SL Max. SL Species Species code (mm) (mm) (mm) (mm) n Pleuronectes ferrugineus PLEFER 11-16 5.7* 17.4 34.9 10,452 Merluccius bilinearis MERBIL 17-20 12.6* 28.2 145.0 3,195 Liparis inquilinus LIPINQ 12-13 9.6* 20.4 49.0 595 Raja erinacea RAJERI 80-90 87.2* 117.6 146.4 281 Urophycis spp. UROSPP 20-25 8.8* 32.5 49.0 224 Macrozoarces americanus MACAME 30 32.0* 89.1 119.0 213 Citharichthys arctifrons CITARC 13-15 10.8* 18.8 29.7 120 Lepophidium profundorum LEPPRO 20 85.5 85.5 115.0 117 Ophichthus cruentifer OPHCRU 83.5 70.0* 111.0 146.0 80 Gadus morhua GADMOR 20-30 20.0* 36.8 49.0 62 Peprilus triacanthus PEPTRI 12-18 6.8* 15.2 45.0 61 Glyptocephalus cynoglossus GLYCYN 22-35 13.5* 30.0 47.5 40 Helicolenus dactylopterus HELDAC 20 21.0* 28.9 35.0 38 Enchelyopus cimbrius ENCCIM 20 9.1* 26.7 47.0 35 Etropus microstomus ETRMIC 10-12 14.7* 23.6 29.0 26 Prionotus carolinus PRICAR 20 11.1* 30.1 49.0 26 Myoxocephatus octodecemspinosus MYOOCT 15 15.9* 22.2 44.0 22 Scophthalmus aquosus SCOAQU 9 9.1* 59.8 143.0 19 Paralichthys oblongus PAROBL 10-12 9.4* 15.8 37.0 15 Hippoglossoides platessoides HIPPLA 18-34 20.1* 23.7 26.1 13 Ophidian marginatum OPHMAR 35-50 52.1* 64.4 75.3 10 Stenotomus chrysops STECHR 17 52 64.4 75 10 Sehastes spp. SEBSPP 24 25.5* 28.1 33.1 8 Lophius americanus LOPAME 50 56.0* 74.2 96.5 5 Melanogrammus aeglefinus MELAEG 20 39.0 44.7 48.0 4 Centropristis striata CENSTR 8.3 25.0 29.8 35.5 3 Hemitripterus americanus HEMAME 19 27.0 28.0 29.0 2 Astroscopus guttatus ASTGUT 20 9.8* 9.8 9.8 Monolene sessilicauda MONSES 33 41.0 41.0 41.0 Pholis gunnellus PHOGUN 30-35 39.0* 39.0 39.0 Tautogolabrus adspersus TAUADS 8 8.7* 8.7 8.7 Citharichthys cornutus CITCOR 13-15 53.2 53.2 53.2 Pleuronectes americanus PLEAME 10-13 81.2 81.2 81.2 cies collected at near-settlement sizes on the shelf (Table 3). Two species represented 88.9% of the age-0 fishes collected: yellowtail flounder {Pleuronectes fer- rugineus, 67.8%) and silver hake (Merluccius bilin- earis, 21.1%). Overall, densities of these dominant species on the shelf during peak settlement averaged 55.7 and 26.23 per 1000 m-, respectively (Table 4), although maximum densities of 771.4 and 660.9 per 1000 m''^, respectively, occurred during peak settle- ment in individual tows. In addition to age-0 juveniles, both larval (ri=267, <2'/, total), and adult («=2765, 13.0% total) fishes were captured with the trawl. Of the 25 species collected at or near-settlement size, 11 included individuals smaller than predicted settle- ment size, based on published maximum lai-val size (Fahay, 1983). Most ( 94% ) age-0 fishes were collected in the summer and fall (Table 4). P. ferrugineus and M. bilinearis, the two dominant species, settled almost exclusively during summer and fall, respectively (Fig. 2). However, if species richness is considered rather than species abundance, summer and fall are still the primary Steves et al : Settlement and nursery habitats for demersal fishes 173 Table 4 Mean seasonal density (fish/1000 m'~i of 21 abundant age Ofish es over the entire continental shelf of the New York B ght. Winter and spring cruises are combined. Species are grouped by their season of highest density and only species with a den sity greater than 0.1 in at least one season are included Winter 1996- Summer 1996 Fall 1996 Spring 1997 Summer 1997 Summer Plcuronectes ferrugineus 55.68 0.53 — 1.29 Citharichthys arctifrons 3.84 2.25 0.66 0.09 Macrozoarces americanus 0.94 0.36 0.16 0.20 Gadus morhua 0.40 0.03 0.14 — Helicolenus dactylopterus 0.09 0.03 — 0.26 Peprilus triacanthus 0.24 0.04 — 0.17 Glyptocephalus cynoglossus 0.21 0.03 — 0.05 Enchelyopus cimbrius 0.21 0.03 — 0.05 Myoxocephalus octodecemspinosus 0.11 0.07 0.11 0.09 Fall Merluccius bilinearis 0.07 26.23 1.04 0.22 Liparis inquilinus 1.19 1.89 0.09 1.75 Urophycis spp. 0.26 1.88 1.05 0.07 Lepophidium profunforum 0.09 0.77 — — Ophichthus cruentifer 0.02 0.61 — 0.11 Etropus microstomus 0.11 0.40 0.79 0.05 Paralichthys oblongus 0.06 0.13 — 0.05 Winter-Spring Raja ennacea 0.60 0.56 0.90 0.54 Prionotus carolinus 0.04 0.22 0.46 — Centropristis atriata — 0.15 0.20 — Ophidian marginatum — — 0.11 — Scophthalmus aquosus 0.02 0.07 0.11 0.05 time of settlement on the shelf (Fig. 2). Although many species settled between August and October, Macrozoarces americanus and Gadus morhua set- tled, and Raja erinacea hatched, on the shelf start- ing in midwinter. Juvenile fish were colleced at all 21 stations sam- pled. Mean depths of all species of age-0 fish, weighted by abundance, are shown in Figure 3. Although the depth distributions cover the entire depth range of the surveyed shelf, for convenience of discussion we sepa- rate species into three depth groups (inner, middle, and outer shelf; Table 5). Inner shelf species included two flounders (Etropus microstomus and Paralich- thys oblongus), searobin (Prionotus carolinus), and fourbeard rockling (Enchelyopus cimbrius). Little skate (Raja erinacea) was collected on the inner shelf at sizes near to hatching size. Midshelf set- tlers were dominated by Plcuronectes ferrugineus but also included large numbers of inquiline snail- fish (Liparis inquilinus) and phycid hakes (Urophy- cis spp.). The dominant fish that settled on the outer shelf were silver hake (Merluccius bilinearis). Gulf Stream Rounder (Citharichthys arctifrons), margined snake eel (Ophichthus cruentifer), fawn cusk eel (Lepophidium profundorum), and black-bellied rose fish (Helicolenus dactylopterus). Generally, distributions along depth (Fig. 3A) and salinity (Fig. 3B) gradients showed similar but inverse trends; this finding was expected because of high correlation between bottom depth and bottom salinity. Weighted average distributions with respect to bottom temperature (Fig. 3C) did not show similar trends; seasonal variation in temperature and the presence of minimum temperatures midshelf during the summer precludes temperature and depth trends from being similar. Midshelf species such as P. fer- rugineus and Liparis inquilinus, which settle in summer, had the coldest mean temperatures of col- lection. Outer-shelf species found in slope water had slightly warmer preferences, and those of inner-shelf species were even warmer. Individual year classes were distinct, based on length frequencies for the more abundant species (Fig. 4-6). For several species (e.g. P. ferrugineus and 174 Fishery Bulletin 98(1) Elropus microstomus Scophthalmus aquosus Citharichthys arctifrons Pholis gunnellus Lepophidium profundorum Centrophstis striata Prionotus carolinus Tautogolabrus adspersus Merluccius bilinearis Parallchthys oblongus Astroscopus guttatus Peprilus triacanthus Myoxocephalus octodecemspinosus Pleuronectes ferrugineus Urophycis spp Optiictitnus cruentifer Lipans inquilinus Glyptocephalus cynoglossus Sebastes spp, Hippoglossoides platessoides Hemitripterus americanus Helicolenus dactylopterus Enchelyopus cimbrius Raja erinacea Gadus morhua Macrozoarces americanus Settlement time JFMAMJJASOND Month Figure 2 Timing of settlement for age-0 fish collected. For each species, a black horizontal bar marks the months in which cruises collected fish at or near settlement size. In some months there was no cruise (Table 11, and any settlement that might have occurred was not witnessed. M. bilinearis), large decreases in abundance were observed following initial settlement, whereas abun- dance decreased at a more modest rate for other spe- cies such as C. arctifrons, Macrozoarces americanus, and L. inquilinus. For species such as P. ferrugineus and L. inqui- linus that settled during the summer (Fig. 4), the cohort could be followed for the entire first year after settlement (June 1996— July 1997). For other species that did not settle in the summer (Figs. 5 and 6), the cohort from the previous year was observed until the new cohort settled in the fall, winter, or spring. Inferred growth rates from the length-frequency dis- tributions varied from two millimeters per month for species such as P. ferrugineus and L. inquilinus (Fig. 4 1, to about 15 mm per month for Merluccius bilin- earis (Fig. 5) and Macrozoarces americanus (Fig. 6). Cross-shelf movement between and within settle- ment and nursery areas was evident for some species but not for others. Pleuronectes ferrugineus, Etropus microstomus, and Lepophidorum profundorum are examples of species with consistent mean depths of distribution during their first year after settlement (Fig. 7, A-C). However, M. americanus exhibited a gradual migration towards deeper waters, whereas C. arctifrons and Merluccius bilinearis moved rapidly after settlement to waters about 30 meters deeper for the remainder of the first year (Fig. 7, D-F). Habitat characteristics Bottom temperatures in the NYB were dynamic (Fig. 8), and seasonal shifts in temperature were more varied at nearshore stations (5-20°C at the 25-m isobath) than at offshore stations (7-ll°C at 90 m). Midshelf bottom temperatures showed a mod- erate seasonal range (4.3-14°C at 50 m). During the summer, bottom temperatures were stable, and there was only a slight increase of about 1°C per month. The highest rate of increase in bottom tem- peratures was recorded at the inshore stations: 8°C between August and September, associated with an early fall turnover in 1996. This was followed by the highest rate of cooling (-3°C) at the same near- shore locations between the September and October cruises. During fall turnover, the bottom tempera- tures increased by about 4°C per month over the mid- shelf, for a total of 8°C in two months. Following this Steves et al.: Settlement and nursery habitats for demersal fishes 175 34 psu) off- shore showed some fluctuation in their distribution on the outer shelf associated with similar fluctua- tions in warmer bottom temperatures. These warm, high-salinity bottom waters are representative of slope water intrusions (Churchill, 1985; Flagg et al., 1995). The total catch of benthic organisms from the trawl varied greatly between stations, but there were gen- erally three major groups or types of trawl samples. Groups were delineated to some extent by the depth at which they were collected. Inner-shelf stations (<40 m) typically included such organisms as the common sea star iAsterias forbesi) and fig sponge {Suberitesficus). Gammarid amphipods, sand shrimp (Crangon septemspinosa), and northern moon snail iEuspira heros) constituted much of the rest of the nonfish fauna collected there. Sand dollar (Echina- 176 Fishery Bulletin 98(1) rachnius parma ) and valves of surf clams (Spisula solidissima) dominated the midshelf group (40-70 m) of benthic organisms. Shell hash collected mid- shelf consisted of fine fragments of sand dollar tests and larger pieces of clam valves. Small, recently settled rock crabs (Cancer borealis) were found in high numbers (>100 per 5-min trawl) during the summer in this type of sample. Margined sea stars (Pontaster tenuispinus) and sea scallops iPlacopec- ten magellanicus) made up most of the trawl catch for deeper outer-shelf stations (70-90 m). Young pandalid shrimp (Pandalus montagui) constituted a large fraction (30-90%) of samples from these sta- tions in late summer and fall. Larger crustaceans such as American lobsters (Homarus americanus) and large rock crabs (C. borealis) were occasionally abundant (>10 per 5-min trawl) at these stations. Several were collected during each cruise. Table 5 Mean density (fish/1000 m-) of 21 abundant age- fishes within each of three depth strata: inner shelf l<40 m), middle shelf (41-70 m), and outer shelf (71-95 m) across all sampling periods. Species are grouped by stratum of their highest abundance; only s pecies from Table 4 are | included. Inner Middle Outer shelf shelf shelf Inner shelf Raja erinacea 1.01 0.81 0.03 Etropus microstomas 0.88 0.01 — Prionotus carolinus 0.21 0.11 0.01 Paralichthys oblongus 0.12 0.03 — Enchelyopus cimbrius 0.10 0.06 0.08 Scophthalmus aquosus 0.09 0.01 — Ophidion marginatum 0.06 — — Middle shelf Pleuronectes ferrugineus 15.31 49.81 2.25 Liparis inquilinus 0.03 1.94 1.11 Urophycis spp. 0.83 0.84 0.58 Macrozoarces americanus 0.03 0.79 0.69 Gadus morhua 0.16 0.38 0.01 Glyptocephalus cynogtossus 0.02 0.23 0.01 Peprilus triacanthus 0.05 0.19 0.05 Myoxocephalus octodecemspinosus — 0.11 0.01 Centropristis striata 0.04 0.10 — Outer shelf Merluccius bilinearis 0.40 5.42 14.69 Citharichthys arctifrons 0.29 0.64 5.63 Lepophidium profundorum — — 0.68 Ophichthus cruentifer — — 0.51 Helicotenus dactylnpterus — — 0.14 Other groups of macrobenthos were ubiquitous. Hermit crabs and cancer crabs of intermediate sizes were found throughout the stations sampled. Although they dominated the midshelf, sand dollars were also collected at inner-shelf stations. Other large benthic fauna such as horseshoe crabs (Limulus polyphemus) and spider crabs (Libinia emarginata) were collected, but these collections were sporadic. Species assemblages and environmental variables Four canonical axes, each representing a linear com- bination of the environmental data, were calculated from the data set. These four axes together accounted for 36.7% of the variation in species abundance and 82.1% of the cumulative variation, in relation to the total variation explained by the environmental vari- ables. A summary of eigenvalues and the variance accounted for by each axis is given in Table 6A. The variance explained by the entire ordination, as well as the first axis, was more significant than expected by chance, as calculated by a Monte Carlo permuta- tion test («=99 iterations, P=0.01 for both tests). Interpretation of the relationships between envi- ronmental variables and the CCA axes involves deter- mining which variables are most correlated to the axes. One intuitive and effective method of accom- plishing this is to examine an ordination plot of envi- ronmental variables (Fig. 9). Environmental variables with large components along a CCA axis have high correlations to that axis. However, the results of the ordination with 25 environmental variables were dif- ficult to interpret because of inherent covariability of variables with one another (Fig. 9, A-B). Forward selection of environmental variables resulted in selection of five environmental variables (bottom temperature, depth, the relative abundance of scallops, longitude, and the relative abundance of margined sea stars). Eigenvalues of the forward selection of these five variables (Table 6B) are pre- dictably lower than for the CCA with all 25 vari- ables included (Table 6A). However, 59% of the variability explained with all 25 variables included was explained by these five environmental variables alone. Temperature represented 23% of the entire variability explained by the environmental data, depth contributed 18%, scallop abundance explained 8%, and longitude and margined sea star abundance combined explain an additional 5%. The remaining variables, which did not add significantly to the resulting explanation, each explained less than 4% of the variation. Because only five variables were selected, the four synthetic gradients (CCA axes) are each highly correlated with one of the environmental variables (Table 6A). Thus, to some extent each axis Steves et al : Settlement and nursery habitats for demersal fishes 177 1000 500 1000 500 400 200 10 5 .^ 1 2 1 1 ' Z 0.5 I 1 = 0.5 < 1 0.5 20 10 20 10 Length fre cephalus o from top to ized to the P. ferrugineus 4 Summer settling L. inquilinus 2 fish H. dactylopterus M. octodecemspinosus 1. °:i 1 r 20 1 A : 4 A ! 4f L. ' L 40 f 2 A 20 1 2 il ! Jl A : 1 1 A 0.5 0.5 1 10 1- : i.- °i 1 1 0.5 1 1 0.5 1 4 0.5 2 10 5 1 0.5 ' n 1 0.5 1 0.5 1 0.5 1 0.5 10 4r 1 jl : k. : .i. °- i_ :i i ; I i 10 20 30 10 20 30 40 50 10 20 30 40 10 20 30 40 Standard length (mm) Figure 4 quency by cruise of four species iPteuronectes ferrugineus. Liparis inquilinus. Helicolenus dactylopterus, and Myoxo- :todecemspinosus\ that settled to the shelf between early summer and midsummer. Plots for each species are ordered bottom according to the order of the 10 cruises (see Table 1 for cruise dates). Abundance on each cruise was standard- number offish collected during 63 5-min tows (21 stations with .3 tows each). is a proxy for the environmental variable to which it is highly correlated. Examination of species-environment biplots revealed relationships between environmental variables and species distributions. The species-environment biplot for the first two axes represents the majority of the species-environment relationships that restilted from the partial CCA (Fig. 9C). The importance of depth and temperature to these two canonical axes (see Table 7) indicates that temperature and depth are the two most important determinants of the spe- cies data. Inner-shelf species such as Stenotomus chrysops, Paralichthys oblongus, Centropristis stri- ata, and Etropus microstomus are located near the upper right-hand quadrant of the ordination (posi- tive values for both of the first CCA axes). Outer- shelf species, such as the rockfishes, Sebastes spp., and Helicolenus dactylopterus and the snake and cusk eels (Ophichthus cruentifer and L. profunda- rum) are found in the lower right-hand quadrant (positive CCA 1 and negative CCA 2). Species from the middle shelf, particularly Pleuronectes ferrugin- eus and Liparis inquilinus, are located near the origin of the ordination diagram. 178 Fishery Bulletin 98(1) 1 •g s M. bilinearis 1 0.5 4 1000 500 200 100 20 10 20' 10 1 0.5 JlUi AL i JL 1 0.5 20 10 UJi I 50 100 150 200 0.5 1 0.5 20 10 1[ 0.5 1 0.5 1 0.5 1 0,5 0.5 Summer and fall settling fish L. profundorum m a HiJ- 50 100 150 C. arctifrons 150 Urophycis spp. 150 200 Standard length (mm) Figure 5 Length frequencies by cruise of four species fMerluccius bilineans. Lepophidium profundorum. Cithanchthys arctifrons, and Uro- phycis spp. ) that settled to the shelf between late summer and midfall. Plots for each species are ordered from top to bottom accord- ing to the order of the 10 cruises (see Table 1 for cruise dates). Abundance on each cruise was standardized to the number offish collected during 63 5-min tows (21 stations with 3 tows each). The origin represents the mean for each environ- mental variable, and those means are weighted by species abundance. Because of the numerical domi- nance of P ferrugineus in the samples, its mean tem- perature and depth of collection is driving the location of the origin. Consequently, the location of P. ferrugin- eus is just slightly left of the origin. The mean depth distribution off! ferrugineus, however, was midshelf and thus its abundance did not skew the species dis- tributions about the second axis, CCA 2. One of the largest effects of the forward selection of variables is that the third and fotuth CCA axes (Fig. 9D ) now have definite environmental correlates. When 25 variables were used during fiill CCA, no one variable stood out as a clear contributor to the variation along these axes (Fig. 9B). With the abtmdance of scallops being the important environmental gradient describing the third CCA axis, one important species association is made clear: the abundances of juvenile inquiline snail- fish and scallops. The fourth CCA axis, representative of longitude, accoimted for only 5% of the total environ- mental explanation of the species patterns. This pat- tern might be considered as the subtle variation among the three transects (west, central, and east). Steves et al.: Settlement and nursery habitats for demersal fishes 179 1 < E. microstomus 80 100 Fall and winter settling fish S. aquosus C. striata 1 0.5 1 0.5 1 1 0.5 2 1 Li J 1 1 0.5 10 5 1 ■1 0.5 1 I 0.5 1 1r 1 0.5 50 100 20 10 M. americanus i 40 20 A 40 20 10 5 A. 4 2 ^ 0.5 20 10 1 . ^ 4 10 5 10 5 n J. ^ 4 .ml. 100 200 300 Standard length (mm) Figure 6 Length frequencies by cruise of four species (Etropus microstomua, Scophthalmus aquosus, Centropristis striata, and Macrozoarces americanus) that settled to the shelf between midfall and winter. Plots for each species are ordered from top to bottom according to the order of the 10 cruises (see Table 1 for cruise dates). Abundance on each cruise was standardized to the number offish collected during 63 5-min tows (21 stations with 3 tows each). Discussion Many commercially and ecologically important spe- cies of demersal fishes in the NYB settle onto the continental shelf throughout the course of the year. In some cases settlement densities on the shelf may rival those of nearshore and estuarine species. For example, at some stations yellowtail flounder {P. fer- rugineus) was found to have settlement densities as high as the long-term average for a congeneric estuarine species, plaice (Pleuronectes platessa), at approximately one per m- (Modin and Phil, 1994). Yet given the large areal extent of P. ferrugineus set- tlement on the shelf, its relative total abundance is potentially much larger. Similarly dense settlement may occur within portions of the shelf environment for other species such as Merluccius bilinearis. Even for those species collected in relatively low numbers, the large areas of NYB in which they settle suggests that total numbers are substantial. Evidence that settlement does occur on the shelf is from two sources. The minimum size of collected 180 Fishery Bulletin 98(1) Pleuronectes ferrugineus 20 40 60 80 100 V0^ "^A^acame I- -0.5 _ -10 Hippla \N^wam cuarc Merbil \ X^::^ '^«'='3'= \ MoW^ Lon \ LepprbL Ophcru \ sebspk \ ^ marg 1 Z .0 1 1 1 -0.5 5 1 CCA axis 1 10 D 05_ Parobl Etrmic ^ Glycyn Phogun Stechr P™^'\ Lopam Heldac < 00_ — ___ — V-^— — — — _i:'P'"^ Ophcru fV\ .. ^ \\ Macame ^ Leppro \ \ Scall Hemame \HipRla Raieri \ marg Censtr Gadmon \ -0 5_ Jl Lon ■10 1 1 1 0.5 5 1 CCA axis 3 Figure 9 (continued) Other variables such as salinity, the abundance of benthic organisms, mean sediment size, as well as the proximate location of other essential habi- tats (e.g. estuaries), were correlated to distance off- shore and depth and are thus possible contributors to observed habitat preferences. It is doubtful that a species would prefer habitat based on distance off- shore or depth alone because both variables are spa- tial indices and as such, they can have no direct effect on habitat quality. The indirect effects of these spatial variables on habitat quality may be associ- ated with how they correlate with the distribution of 184 Fishery Bulletin 98(1) Table 6 (A) Results from the CCA including all 25 environmental variables. The sum of all canonical eigenvalues (2.187) was 44.77f of the total inertia (4.891). (B) Results of the CCA with forward selection (5 environmental variables) and season treated as a covariable. The sum of the unconstrained eigenvalues (3.825) is 78.2'7r of the total inertia, indicating that the seasonal component contributed 21.89c of the variance. The sum of all canonical eigenvalues (0.636) is 16^5^ that of the sum-6f all unconstrained eigenvalues and 13% of total inertia. See text for explanation of terms. Axes 1 2 3 4 Total inertia A Eigenvalues 0.809 0.464 0.284 0.238 4.891 Species-environment correlations 0.961 0.873 0.704 0.691 Cumulative percentage variance of species data 16.5 26.0 31.9 36.7 of species-environment relation 37.0 58.2 71.2 82.1 Sum of all unconstrained eigenvalues 4.891 Sum of all canonical eigenvalues 2.187 B Eigenvalues 0.269 0.230 0.078 0.047 4.891 Species-environment correlations 0.721 0.735 0.518 0.365 Cumulative percentage variance of species data 7.0 13.0 15.1 16.3 of species-environment relation 42.3 78.5 90.8 98.3 Sum of all unconstrained eigenvalues (after fittmg covariables) 3.825 Sum of all canonical eigenvalues (after fitting covariables) 0.636 important factors or combinations of factors, or with both. For example, age-0 inquihne snailfish (L. inqui- linus) are found almost exclusively in association with the presence of sea scallops, with which they have their inquiline relationship (Able and Musick, 1976). The distribution of these fishes on the mid to outer shelf is therefore not necessarily directly affected by any physiological response to depth or distance offshore; rather, they are probably most affected by their dependence on sea scallops for shel- ter. The distribution of scallops in turn may be affected by such variables as temperature, salinity, circulation pattern, and the production of phyto- plankton (Stewart and Arnold, 1994). Temperature, however, may have a direct effect on the quality of habitat for a given species because many physiologi- cal functions such as respiration, metabolism, and growth are directly controlled by changes in temper- ature (Neill etal., 1994). Temperature and depth not only seem to provide the best description of the overall variance within this data set, but, along with salinity and time of year, form the basis for the hydrographic descrip- tion of the NYB. In general, the bottom waters of Table 7 Weighted correlations between the five environmental vari- ables selected during the CCA with forward selection and the axes of from the analysis. Numbers in bold indicate the strongest correlation for each of the four axes. See Table 2 for meaning of environmental variable codes Axis Temp Scall Long Marg CCA 1 0.8262 CCA 2 0.4932 CCA 3 -0.0819 CCA 4 0.0863 0.4257 -0.7895 0.1485 -0.3690 -0.0796 1636 0.9711 -0.1399 0.3723 0.6508 -0.2067 -0.5973 -0.0620 0.1028 -0.8822 -0.0934 the NYB can be broken down by location into three regions: inner-shelf waters, middle shelf (cold pool), and outer shelf. This breakdown of the continental shelf for the NYB is consistent with that of other shelf systems (McRoy et al., 1986; Werner et al., 1997). During the summer, waters overlaying these regions of the shelf are divided by two frontal fea- tures: the inner front, dividing the inner and middle Steves et al : Settlement and nursery habitats for demersal fishes 185 shelves, and the middle front, dividing the middle and outer shelves. A third front, the shelf-break front, separates the outer shelf from the continental slope (McRoy et al., 1986). These three hydrographic features can be utilized as an effective means for describing the general nursery areas for gi-ound- fishes within the NYB. Differences in environmental conditions occur not only along the shelf, but also temporally. Patterns in surficial sediments and bathymetry change on the order of hundreds to thousands of years. These characteristics are relatively stable in their distri- butions on the shelf Other factors such as benthic faunal distributions may change from year to year but should remain unvarying during the first year of a fish's life. Hydrographic variables and the presence of ephemeral predator and prey species, however, may change on a seasonal or even daily basis. The scale and extent to which these variables change are important in understanding the nature of the settle- ment and nursery areas. The turnover of the cold pool in the fall represents an abrupt change in bottom temperature that sub- sequently may have biological significance. In our study, higher mean catch of age-0 P. ferrugineus cor- responded to the distribution of the cold pool, and catch dropped dramatically in the fall as bottom temperature increased during cold pool turnover. For a species with a preference for low tempera- tures, increased temperature may be a density-inde- pendent source of mortality. In at least one case, an increase in mortality of larval P. ferrugineus on Georges Bank has been associated with an increase in temperature from a warm-core ring ( Colton, 1959 ). Although such abrupt temperature fluctuations at greater depths are not common, they may occa- sionally occur near the edge of the shelf in some areas, perhaps associated with Gulf Stream mean- ders (Able et al., 1993). Able et al. (1993) found a temperature increase of 6°C over a two-day period, and they suggested that such temperature fluctua- tions may cause a cessation of feeding for tilefish, Lopholatilus chamaeleonticeps. In contrast, some species may benefit from increased bottom temper- atures associated with turnover. For example, in our study, settlement of M. bilinearis did not peak until after turnover, when bottom temperature was greater than 9'"C. Likewise, spawning distribution for Atlantic croaker (Micropogonias undulatus) in the southern portions of the MAB has a positive association with the presence and location of warmer waters coincident with the cold pool turnover (Nor- cross and Austin, 1988). This study and another from Georges Bank (Frank et al., 1992) indicate that the distribution of P. fer- rugineus corresponds well with bottom temperature. The presence of P. ferrugineus is negatively corre- lated with bottom temperature in the NYB, with highest abundance in the coldest waters (<8°C). However, on the Grand Banks the correlation with temperature was positive, with most fish collected above the warmest bottom waters (>3°C). Juveniles from the Grand Banks, however, were collected off the bottom, and highest abundances were found just below the thermocline, in waters closer to 6°C. Tem- peratures between 4° and 8°C have been noted as the preferred temperatures for older P. ferrugineus (Scott, 1982; Ross and Nelson, 1992; Walsh, 1992), and the evidence from these studies suggests that this may be true for age-0 P. ferrugineus as well. Besides P. ferrugineus, several other boreal spe- cies, American plaice iHippoglossoides platessoides), haddock (Melanogrammus aeglefinus), and Atlantic cod {Gadus morhua), were collected in our samples from the cold pool. Although not as numerically dominant as P. ferrugineus, the presence of these species this far down the coast suggests that the cold pool may act as a temporary refuge from the warmer waters normally associated with lower lati- tude. Metapopulations of such species, wherein local populations occupy small areas of suitable habitat outside the main population, are a potential confound- ing factor to the management of these species (Bailey, 1997). In the extreme, these juveniles may not be viable additions to the population if environmental conditions are not suitable for them to reach matu- rity. Just as warm-core eddies and eddy streamers may bring expatriates into this area from the south in the summer (Hare and Cowen, 1991), the cold pool may allow boreal species to temporarily extend their range southward during the summer. It is possible that conditions may occasionally enhance survival in such potentially marginal habitats to the extent that success of the population year class is facilitated. In our study, we used trawl data to describe the general distribution and large-scale habitat associa- tions of age-0 fishes. The resolution obtainable from trawl data is limited to the area swept (about 700 m^). Such resolution is sufficient to determine differ- ences between stations that are 10 km apart but not to deterine the heterogeneity in habitat within a sta- tion. In situ methods, including the use of manned submersibles, are required for these types of small- scale studies (see Auster et al., 1991, 1997; Adams et al., 1995). One other drawback of using a trawl to quantify fish abundance is that trawls are inherently poor at collecting all fishes present in a given area; gear selection is size- and species-specific. However, if gear selection is constant between sites, trawl sam- ples should be comparable (Kuipers et al., 1992). The 186 Fishery Bulletin 98(1 relative efficiency of our 2-m beam trawl on offshore nursery areas for juvenile flatfish and other demer- sal species was unknown, although it was likely that our trawl was relatively inefficient but consistent in collecting demersal fishes. Consistent inefficiency among trawl samples would not affect the general patterns that we found but could have led to under- estimation of abundances. For a trawl to be efficient, demersal fishes must re- main near the bottom. Recently settled juveniles of some species, such as M. bilinearis, have been shown to make nightly excursions away from the bottom to feed (Fahay, 1974). For a fusiform juvenile like M. bilinearis this ability is not surprising. How- ever, the pleuronectiform P. ferrugineus has been col- lected well off the bottom on the Grand Banks, not only as larvae but also as juveniles (Frank et al., 1992). These "pelagic juveniles" were of the same size classes (10-34 mm) as the postmetamorphic flounder we collected on the bottom with a 2-m beam trawl in our study. In extensive studies in the MAB with comparable or larger midwater sampling gear, few P. ferrugineus larger than 15 mm were collected (Morse, 1989; Cowen et al., 1993). For the NYE, how- ever, juvenile P. ferrugineus do not appear to exhibit such trawl-avoiding behavior. This regional differ- ence in vertical distribution of >20-mm postmeta- morphic young may be due either to differences in the temperature structure of the water column (i.e. depth of optimal temperature, as discussed above) or to some other unknown latitudinal difference in developmental rate, settlement size, or behavior (Miller et al., 1991; Fuiman and Higgs, 1997; Osse and Boogaart, 1997). In summary, age-0 demersal fishes utilize the con- tinental shelf of the NYB as both settlement and nursery habitat. According to our findings, the shelf of the NYB can be divided into three broad nursery areas (inner, middle, and outer shelves) and can be described by species assemblage as well as by hydrography. There is a need for more research con- cerning the quality of habitats for age-0 fishes in the NYB. Information on smaller-scale variation in habitat for age-0 fishes may be just as revealing as that for larger macrobenthos (see Auster et al., 1991). Manned submersibles may be used to deter- mine such small-scale habitat associations, even for small juveniles. Differences in growth rates of given species among habitats should also shed light on habitat quality within the NYB. Nursery habitats play an important role in the life history of marine fishes, and knowledge gained about the distribution and quality of these areas for commercially impor- tant offshore species should help to improve man- agement of these areas.. Acknowledgments We thank all those who have contributed to the vari- ous aspects of this work. The fieldwork was made possible with the help of numerous volunteers, par- ticularly Mark Sullivan and Teresa Rotunno. Cap- tain Mark Phillips and the crew of the FV Illusion gave much to the practical aspects of the cruises. This manuscript benefited from critical reviews from Tom Grothues, Ken Able, and Robert Cerrato, as well as from valuable discussions with Michael Fahay, Al Stoner, and Brenda Norcross. Supplemental tempera- ture data for August 1996 was kindly provided by Jack Jossi. This paper resulted from research sponsored by the National Oceanic and Atmospheric Administra- tion's Saltonstall-Kennedy program, award number NA66FD0012. The views expressed herein are those of the authors and do not necessarily reflect the views of NOAAor its subagencies. Literature cited Able, K. W., C. B. Grimes, R. S. Jones, and D. C. Twichell. 1993. Temporal and spatial variation in the habitat charac- teristics of tilefish iLopholatilus chamaeleonticeps) off the east coast of Florida. Bull. Mar. Sci. 53(3):1,01.3-1,026. Able, K. W., R. E. Matheson, W. W. Morse, M. P. Fahay, and G. Shepard. 1989. Patterns of summer flounder Paralichthys dentatus early life history in the Mid-Atlantic Bight and New Jersey estuaries. Fish. Bull. 88:1-12. Able, K. W., and J. A. Musick. 1976. Life history, ecology, and behavior of Liparis mquili- nus (Pisces: Cyclopteridael associated with the sea scallop Placopecten mageUanicus. Fish. Bull. 74(21:409-412. Adams, P. B., J. L. Butler, C. H. Baxter, T. E. Laidig, K. A. Dahlin, and W. W. Wakefield. 1995. Population estimates of Pacific coast groundfishes from video transects and swept-area trawls. Fish. Bull. 93:446-455. Auster, P. J., R. J. Malatesta, and C. L. S. Donaldson. 1997. Distributional responses to small-scale habitat van- ability by early juvenile silver hake. Merluccius bilinearis. Environ. Biol. Fi.sh. 50:195-200. Auster, P. J., R. J. Malatesta, S. C. LaRosa, R. A- Cooper, and L. L. Stewart. 1991. Microhabitat utilization by the megafaunal assem- blage at a low relief outer continental shelf site — Middle Atlantic Bight, USA. J. Northwest Atl. Fish Sci. 11:59- 69. Bailey, K. M. 1994. Predation on juvenile flatfish and recruitment vari- ability. Neth. J. Sea. Res. 32(2): 175-189. 1997. Structural dynamics and ecology of flatfish popula- ti(ms. Neth. J. Sea Res. 37:269-280. BoUe, L. J., R. Dapper, J. I. Witte, and H. W. Van der Veer. 1994. Nursery grounds of dab iLimanda limanda L.) in the southern North Sea. Neth. J. Sea Res. 32:299-307. Bowman, M. J., and L. D. Wunderlich. 1977. Hydrographic properties. MESANew York. Bight Atlas Steves et a\- Settlement and nursery habitats for demersal fisties 187 Monograph 10. New York Sea Grant Institute, Albany, NY. Bradford, M. J. 1992. Precision of recruitment predictions from early life stages of marine fishes. Fish. Bull. 90:439-453. Carney, R. S., and A. G. Carey Jr. 1980. Effectiveness of metering wheels for measurement of area sampled by beam trawls. Fish. Bull. 78:791-796. Chambers, R. C, and W. C. Leggett. 1992. Possible causes and consequences of variation in age and size at metamorphosis in flatfishes (Pleuronecti- formes): an analysis at the individual, population, and spe- cies levels. Neth. J. Sea Res. 29:7-24. Cho, R. D. 1996. Vertical distributions of ichthyoplankton in the New York Bight with relation to potential cross-shelf transport and retention mechanisms. M.S. thesis, State University of New York at Stony Brook, Stony Brook, m', 173 p. Churchill, J. H. 1985. Intrusions of outer shelf and slope water within the nearshore zone off Long Island, New York. Limnol. Ocean- ogr. 30(5):972-986. Colton, J. B. Jr. 1959. A field observation of mortality of marine fish larvae due to warming. Limnol, Oceanogr. 4:219-222. 1972. Temperature trends and the distribution of ground- fish in continental shelf waters, Nova Scotia to Long Island. Fish. Bull. 70:637-657. Colvocoresses, J. A., and J. A. Musick. 1984. Species associations and community composition of Middle Atlantic Bight continental shelf demersal fishes. Fish. Bull. 82:295-313. Cowen, R. K., J. A. Hare, and M. P. Fahay. 1993. Beyond hydrography: can physical processes explain larval fish assemblages within the Middle Atlantic Bight. Bull. Mar Sci. 53:567-587. Fahay, M. P. 1974. Occurrence of silver hake, Merlucciuf: hilineans, eggs and larvae along the middle Atlantic continental shelf during 1966. Fish. Bull. 72(31:813-834. 1983. Guide to the early stages of marine fishes occurring in the western North Atlantic Ocean, Cape Hatteras to the south- em Scotian Shelf J. Northwest Atl. Fish. Sci. 4:1-423. Felley, J. D., and M. Vecchione. 1995. Assessing habitat use by nekton on the continental slope using archived videotapes from submersibles. Fish. Bull. 93:262-273. Flagg, C. N., R. W. Houghton, and L. J. Pietrafesa. 1995. Summertime thermocline salinity maximum inti-usions in the Mid-Atlantic Bight. Deep-Sea Res. 41(2/31:325-340. Frank, K. T., J. W. Loder, J. E. Carscadden, W. C. Leggett, and C. T. Taggart. 1992. Larval flatfish distributions and drift on the southern Grand Bank. Can. J. Fish, Aquat. Sci. 49:467-483. Freeland, G. L., and D. J. P. Swift. 1978. Surficial sediments. MESA New York Bight Atlas Monograph 10. New York Sea Grant Institute, Albany, NY. 93 p. Fuiman, L. A., and D. M. Higgs. 1997. Ontogeny, growth, and the recruitment process. In R. C. Chambers and E. A. Tripple (eds.), Early life history and recruitment in fish populations, p. 225-250. Chap- man and Hall, London. Gibson, R. N. 1994. Impact of habitat quality and quantity on the recruit- ment of juvenile fiatfishes Neth. J. Sea Res. 32:191-206. Hare, J. A., and R. K. Cowen. 1991. Expatriation ofXynchtys nnvacuta (Pisces: Labridae) larvae: evidence of rapid cross-shelf exchange. J. Mar. Res. 49:801-923. Henderson, P. A., and R. M. H. Seaby. 1994. On the factors influencing juvenile flatfish abundance in the lower Severn Estuary, England. Neth. J. Sea Res. 32:321-330. Houde, E. D. 1987. Fish early life dynamics and recruitment variability. Am. Fish. Soc. Symp. 2:17-29. Houghton, R. W., R. Schlitz, R. C. Beardsley, B. Butman, and J. L. Chamberlin. 1982. The Middle Atlantic Bight cold pool: evolution of the temperature structure during the summer of 1979. J. Phys. Ocean. 12:1.019-1,029. Jenkins, G. P., M. Shaw, and B. D. Stewart. 1993. Spatial variation in food-limited growth of juvenile greenback flounder. Rhombosolea tapirina: evidence from otolith daily increments and otolith scaling. Can. J. Fish. Aquat. Sci. 50:2.558-2,567. Kendall, A. W., and N. A. Naplin. 1981. Diel-depth distribution of summer ichthyoplankton in the Middle Atlantic Bight. Fish. Bull. 79(41:705-726. Ketchum, B. H., and N. Corwin. 1964. The persistence of "winter" water on the continental shelf south of Long Island, New York. Limno. Oceanogr. 9(4):467-475. Kuipers, B. R., B. MacCurrin, J. M. Miller, H. W. Van der Veer, and J. I. Witte. 1992. Small trawls in juvenile flatfish research: their devel- opment and efficiency. Neth. J. Sea Res. 29:109-117. Malloy, K. D., and T. E. Targett. 1991. Feeding, growth, and survival of juvenile summer flounder Paralichthys dentatus: Experimental analysis of the effects of temperature and salinity. Mar Ecol. Prog. Ser. 72:213-223. McHugh, J. L. 1972. Marine fisheries ofNew York State. Fish. BuU. 3:585- 610. McRoy, C. P., D. W. Hood, L. K. Coachman, J. J. Walsh, and J. J. Goering. 1986. Processes and resources of the Bering Sea shelf (PROBES): the development and accomplishments of the project. Continental Shelf Res. 5:5-21. Miller, J. M. 1994. An overview of the second flatfish symposium: recruit- ment in flatfish. Neth. J. Sea Res. 32:10.3-106. Miller, J. M., J. S. Burke, and G. R. Fitzhugh. 1991. Early life history patterns of Atlantic North Ameri- can flatfish: likely (and unlikely l factors controlling recruit- ment. Neth, J, Sea Res, 27:261-275. Modin, J., and L. Pihl. 1 994. DifTerences in growth and mortality of juvenile plaice, Pleuronectes platessa L. follow-ing normal and extremely high settlement. Neth. J. Sea Res. 32 ( 3/4 ):33 1-341. Morse, W. W. 1989. Catchability growth, and mortality of larval fishes. Fish. Bull, 87(3 ):417-446. Morse, W. W., M. P. Fahay, and W. G. Smith. 1 987. MARMAP surveys of the continental shelf from Cape Hatteras, North Carolina, to Cape Sable, Nova Scotia ( 1977- 1984), Atlas No. 2. Annual distribution patterns of fish larvae. US, Dep, Commer, NOAA Tech. Memo. NMFS- F/NEC-47, 215p. 188 Fishery Bulletin 98(1) Nash, R. D. M., A. J. Geffen, and G. Hughes. 1994. Individual growth of juvenile plaice iPleuronectespla- tessa L.) on a small Irish Sea nursery ground (Port Erin Bay, Isle of Man, UK). Neth. J. Sea Res. 32:.369-378. NEFSC (Northeast Fisheries Science Center). 1992. Status of fishery resources off the northeastern United States for 1992. U.S. Dep. NOAA Tech. Memo. NMFS-F/ NEC-95, 133 p. Neill, W. H., J. M. Miller, H. W. Van der Veer, and K. O. Winemiller. 1994. Ecophysiology of marine fish recruitment: a concep- tual framework for understanding interannual variability. Neth. J. Sea Res. 32(2):135-152. Norcross, B. L., and H. M. Austin. 1988. Middle Atlantic Bight meridional wind component effect on bottom water temperatures and spawning distribu- tion of Atlantic croaker Continental Shelf Res. 8( 1 ):69-88. Osse, J. W. M., and J. G. M. Boogaart. 1997. Size of flatfish larvae at transformation, functional demands and historical constraints. J. Sea Res. 37:229- 239. Palmer, M. W. 1993. Putting thmgs in even better order: the advantages of canonical correspondence analysis. Ecology 74(31:2,215- 2,230. Pearcy, W. G. 1978. Distribution and abundance of small flatfish and other demersal fishes in a region of diverse sediments and bathymetry off Oregon. Fish. Bull. 76l 3 1:629-640. Peterman, R. M., M. Bradford, N. C. Lo, and R. Methot. 1988. Contribution of early life stages to interannual vari- ability in recruitment of northern anchovy iEngraulis mordax). Can. J. Fish. Aquat. Sci. 45:8-16. Pihl, L., S. P. Baden, and R. J. Diaz. 1991. Effects of periodic hypoxia on distribution of demer- sal fish and crustaceans. Mar. Biol. 108:34-360. Rakocinski, C. F., J. Lyczkowski-Schultz, and S. L. Richardson. 1996. Ichthyoplankton assemblage structure in Mississippi Sound as revealed by canonical correspondence analysis. Est. Coast. Shelf Sci. 43:237-257. Riley, J. D., D. J. Symonds, and L. Woolner. 1981. On the factors influencing the distribution of age-0 demersal fish in coastal waters. Rapp. P.-V. Reun. Cons, int. Explor Mer 178:223-228. Rijnsdorp, A. D., F. A. Van Beek, S. Flatman, R. M. Millner, J. D. Riley, M. Giret, and R. De Clerck. 1992. Recruitment of sole stocks, Solea solea (L.i, in the Northeast Atlantic. Neth. J. Sea Res. 29:173-192. Rodgers, S. I. 1992. Environmental factors affecting the distribution of sole (Solea solea (L.)) within a nursery area. Neth. J. Sea Res. 29(1-3):153-161. Ross, M. R., and G. A. Nelson. 1992. Influences of stock abundance and bottom-water tem- perature on gi'owth dynamics of haddock and yellowtail flounder on Georges Bank. Trans. Am. Fish. Soc. 121:578- 587. Schlee, J. 1964. New Jersey ofTshore gravel deposit. Pit and Quarry 1964:13-21. Scott, J. S. 1982. Selection of bottom type by groundfish of the Scotian Shelf. Can. J. Fish. Aquat. Sci. 39:943-947. Sissenwine, M. P. 1984. Why do fish populations vary? In R. M. May (ed.). Exploitation of marine communities, p. 59-94. Dhlem Kon- ferenzen. Springer-Verlag, NY. Sogard, S. M. 1992. Variability in growth rates of juvenile fishes in differ- ent estuarine habitats. Mar Ecol. Prog. Ser 85:3.5-53. Stein, D. L., B. N. Tissot, M. A. Hixon, and W. Barss. 1992. Fish-habitat association on a deep reef at the edge of the Oregon continental shelf Fish. Bull. 90:540-551. Stewart, P. L., and S. H. Arnold. 1994. Environmental requirements of the sea scallop (Placo- pecten magellanicus) in eastern Canada and its response to human impacts. Can. Tech. Rep. Fish. Aquat. Sci. 2,005:ix -f36p. ter Braak, C. J. F. 1992. CANOCO— a FORTRAN program for canonical com- munity ordination. Microcomputer Power, Ithaca, New York, 95 p. 1995. Ordination. In R. H. G. Jongman, C. J. F ter Braak, and O. F. R. van Tongeren (eds.). Data analysis in commu- nity and landscape ecology, p. 91-174. Cambridge Uni- versity Press, New York, NY. ter Braak, C. J. F., and P. F. M. Verdonshot. 1995. Canonical correspondence analysis and related multi- variate methods in aquatic ecology. Aquat. Sci. 57(3):255- 289. Toole, C. L., D. F. Markle, and C. J. Donohoe. 1997. Settlement timing, distribution, and abundance of Dover sole t Microstomas pacificus) on an outer continental shelf nursery area. Can. J. Fish. Aquat. Sci. 54:531—542. Van der Veer, H. W., R. Berghahn, and A. D. Rijnsdorp. 1994. Impact of juvenile growth on recruitment in flatfish. Neth. J. Sea Res. 32:153-173. Walsh, S. J. 1992. Factors influencing the distribution of juvenile yel- lowtail flounder iLimanda ferrugmea I on the Grand Bank of Newfoundland. Neth. J. Sea Res. 29l 1-3): 193-203. Werner, F. E., J. A. Quinlan, B. O. Blanton, and R. A. Luettich Jr. 1997. The role of hydrodynamics in explaining variability in fish populations. J. Sea Res. 27:195-212. 189 Abstract.— The distribution and abun- dance of deep-sea sharks on Chatham Rise, New Zealand, are described. Sharks were collected as bycatch in two deep-water trawl fisheries at a total of 390 stations, which ranged in depth from 740 to 1503 m. Sixteen species of shark were caught: Deania calcea, Centroscymnus crepidaten Etmopteriis granulosus, and Centroscymnus owstoni accounted for the largest portion of the shark catch. Species that would provide the highest yield of commer- cially important liver lipids were not abundant in trawls. All sharks com- bined formed only 4.2^^^ of overall bio- mass captured in trawls. Depth is a major determinant of the composition of the shark assemblage; both density of sharks (kg/km-) and species diversity were inversely proportional to depth. Distributional patterns of the shark community varied with location on Cha- tham Rise, and species composition of the shark catch varied with the spe- cies of teleost targeted in deep-water trawls. Assemblage of deep-sea sharks on Chatham Rise, New Zealand Bradley M. Wetherbee Zoology Department 2538 The Mall University of Hawaii Honolulu, Hawaii 96822 Present address Northeast Fisheries Science Center National Marine Fisheries Service, NOAA 28 Tarzwell Dr, Narragansett, Rhode Island 02882 E-mail address: brad wetherbeea'noaa gov Manuscript accepted 12 July 1999. Fish. Bull. 98:189-198 (20001. Sharks are common bycatch in deep water fisheries around the world, forming as much as 509?^ of the catch in deep-sea trawls in areas such as New Zealand and Austra- lia (Deprez et al., 1990; Clark and King' ). Most sharks captured in the New Zealand and Australian deep- water fisheries are dead by the time they are brought to the surface and are discarded, but some sharks are retained for their liver oil. In Japan and Australia, several species of deep-sea shark in the family Squal- idae are targeted in fisheries and their liver oil is utilized. Although the short-term potential of fisheries directed towards deep-sea sharks has been investigated for a few spe- cies (Summers, 1987; Davenport and Deprez, 1989), little informa- tion on even basic biology is avail- able for the species captured in these fisheries. Thus, the effects that deep-water fisheries have on shark populations that are either targeted directly or caught inciden- tally are unknown. Information on abundance, distribution, commu- nity structure, reproduction, and age and growth of deep-sea sharks would improve understanding of these effects. Shark liver oil contains commer- cially important lipids, such as squalene and diacyl glycerol ether, which are used in cosmetic, phar- maceutical, and other industries (Deprez et al., 1990; Bakes and Nichols, 1995). The lipid composi- tion of liver oil is quite variable among and within species, and consequently the most desirable sharks are those individuals and species that have the highest poten- tial as a source for these valuable lipids ( Davenport and Deprez, 1989; Bakes and Nichols, 1995). There- fore, understanding the distribution and abundance of different species of deep-sea shark, in conjunction with knowledge of the lipid compo- sition of their liver oil, is important for optimal use of these resources. Some deep-sea sharks prey upon commercially important teleosts (Clark et al., 1989; Clark and King' ), but the impacts of shark predation on fish populations in terms of the overall economic impact on the fish- ery are unknown. Diet varies consid- erably among even closely related species of deep-sea shark (Com- pagno et al., 1991; Ebert et al., 1992), and the level of predation on commercially important species of teleost by sharks also varies among species (author's unpubl. data). Information on the distri- bution and abundance of deep-sea sharks, in conjunction with knowl- 1 Clark, M.R., and K.J. King. 1989. Deep water fish resources off the North Island, New Zealand: results of a trawl survey. May 1985 to June 1986. New Zealand Fisheries Technical Report 11, 56 p. MAF Fisheries Research Center, RO. Box 297, Wellington, New Zealand. 190 Fishery Bulletin 98(1) o 1990 A 1993 600in 1200m 1500m edge of their feeding habits, would improve our under- standing of interactions be- tween sharks and commer- cially important teleosts. A variety of species of shark inhabit the deep wa- ter off New Zealand, where they form part of the by- catch of deep-sea fisheries that target teleosts such as orange roughy (Hoploste- thus atlanticus ) and smooth oreo (Pseudocyttus macula- tus) (Clark and Tracey, 1994; Clark and KingM. Ac- cess to this bycatch pro- vided an opportunity to ex- amine a multispecies com- plex of sharks, which might be termed an assemblage — "a group of co-occurring pop- ulations — not necessarily interacting" as defined by Crowder (1990). The purpose of this study was to investigate the abundance and distribution of sharks on Chatham Rise to increase understanding of the effects of fishing on shark populations, the poten- tial of shark fisheries and utilization of bycatch, and interactions between sharks and commercially important teleosts. Materials and methods Data for this study were collected from deep-water bottom trawls during two cruises conducted by the Ministry of Agriculture and Fisheries on Chatham Rise, New Zealand (Fig. 1). The first survey (15 June to 5 August 1990) consisted of 281 trawls for orange roughy iH. atlanticus) and was conducted primarily on the north of Chatham Rise from the FV Cordelia. The second survey (24 October to 9 November 1993) consisted of 109 trawls for smooth oreo (P. macula- tus), primarily on the south of Chatham Rise from the RV Tangaroa. Fishing during both surveys was conducted at depths of 740-1503 m throughout the day and night (Fig. 2). Each survey consisted of a stratified random trawl design intended to provide relative biomass estimates and to illustrate patterns of distribution of deep-water species on Chatham Rise. Six-panel bottom otter- trawls with cut-away lower wings were used in each survey. The door-spread was 75 m for orange roughy trawls and 119 m for smooth oreo trawls, and distance between the net wdngs for both trawls was approxi- Figure 1 Map of Chatham Rise, New Zealand, showing depth contour lines and locations of trawls in a 1990 orange roughy survey and a 1993 smooth oreo survey. Trawls were grouped into 10 areas on the basis of major latitude and longitude meridians. mately 26 m. Codend mesh sizes were 110 mm for orange roughy trawls and 100 mm for smooth oreo trawls. Towing speed for both vessels was approxi- mately 3.0 kn. Orange roughy trawls were roughly 1 h in duration, and smooth oreo trawls ranged from several minutes to 45 min. For density estimates (kg shark/km''^ ) it was assumed that herding by, and escape from, nets were minimal, and that trawls sampled dif- ferent species of shark with equal effort. For each trawl, the catch was sorted into bins by species, and the total weight of each species caught at each station was recorded. Latitude, longitude, water temperature, minimum and maximum depth of fishing, towing speed, and start and end time were also recorded for each trawl. When the author was present on the research vessel (at all times other than from 15 June to 10 July 1990), all individuals of each species of shark were weighed and measured, except when large numbers of sharks were caught and a lack of time prevented examination of every shark. Because of size varied among species, an esti- mate of the total number of individuals captured in all fishing was derived by using the average weight for each species. Because there were differences in fishing methods (net characteristics, trawl duration) and time (season, year) between the two surveys, catch data from surveys were examined separately. When possible, comparisons were made between common areas fished during both surveys. For com- parison of the composition of the shark community at different locations on Chatham Rise, ten areas were designated based on major latitude and longi- tude meridians (Fig. 1). Wetherbee: Assemblage of deep-sea sharks on Chatham Rise, New Zealand 191 Consideration of sharks as an assemblage, which is separate from the rest of Chatham Rise commu- nity, is an artificial division. However, because the primary interest of this study was to describe the abundance and distribution of the sharks on Cha- tham Rise, several ecological indices were employed to compare different locations, depths, and species of shark. Abundance was expressed as density (kg shark/km- ) and was calculated for each species within each trawl based on the total weight of sharks caught, net width, towing speed, and trawl duration. Three features of distribution were examined for sharks: diversity, similarity, and randomness. Diversity was expressed as the number of species of sharks per trawl (Stephens et al. 1984; Garcia et al. 1998). The Bray-Curtis similarity index was used for compari- sons among the ten areas on Chatham Rise, depth intervals, and between the two surveys: S=l IK->^u|/X0.01), and data from common areas for the two surveys were therefore combined. In the orange roughy survey, there was a significant difference among areas for densities of all species except D. licha, whereas in the oreo survey, significant differ- ences were observed among areas for only two spe- cies, C. crepidater and D. calcea (ANOVA, P<0.05). The highest densities for all sharks combined were recorded at the eastern tip of Chatham Rise, in areas 5 and 6 (1003.4 and 2249.1 kg/km^), and the lowest were in the southwestern areas 8 and 10 (257.7 and 254.4 kg/km2). Composition of the shark catch varied consider- ably with location fished. In areas on the north of Chatham Rise, closest to New Zealand (areas 1 and 2), the catch was dominated by C. owstoni and C. crepidater, which accounted for 84% of the shark catch by weight (Fig. 3). On the mid-north of Cha- tham Rise (area 3), C. owstoni and C. crepidater still formed the majority of the catch; however D. calcea was also abundant and all eight major taxa were recorded. The northeast of Chatham Rise included those areas (4 and 5) where the most trawls were conducted. Here, C. crepidater still formed a large Welherbee: Assemblage of deep-sea sharks on Chatham Rise, New Zealand 193 Area Area 2 Area 3 1,798.9 Area 4 4,558.5 Area 7 2,054.2 2,936.5 Area 5 007.6 Area 10 2,282.9 Area 6 903.7 Q E.g. B Co n C.c ^ D.c A.s s S.p. D C.s m D.i. Figure 3 Species composition ( as percent of total kg of shark caught ) of the shark catch for both surveys combined in each of the 10 areas on Chatham Rise, New Zealand. Numbers below each area are the total weight (kg) of sharks caught in that area. E.g. = Etmopterus granulosus: Co. = Centroscymnus owstoni; C.c. = Centroscymnus crept- dater. D.c. = Deania calcea; A.s. = Apristurus spp.; S.p. = Scymnodon plunketi; C.s. = Centrophorus squamosus; and D.I. = Dalatias licha. 194 Fishery Bulletin 98(1) part of the catch, but the catch of C. owstoni declined substantially, and D. calcea began to dominate the shark catch. Along the eastern tip and southeast portion of Chatham Rise (areas 6 and 7), D. calcea accounted for the highest percentage of the catch, but E. granulosus was also prominent, and these two species formed over 85% of the shark catch by weight. Etmopterus granulosus dominated catches in areas along the south of Chatham Rise (areas 8-10); the proportion of the total catch increased with proximity to New Zealand (Fig. 3). The three large squalids (C. squamosus, S. plunketi, D. licha) were sporadically caught in areas 1-8, each with a peak density in area 6, but no squalids were recorded from the southwest of Chatham Rise (areas 9 and 10). Apristurus spp. were caught in small numbers throughout Chatham Rise, but their presence was more dependent upon depth than location (Fig. 3). Composition of the catch also varied with depth (Fig. 4) and several natural divisions were apparent. The three large squalids (C. squamosus, S. plunketi, D. licha ) appeared to have shallow distributions, and were not captured deeper than 1100 m (with the exception of one S. plunketi caught at 1170 m, and one C. squamosus at 1201 m). Densities (kg/km-) of the other species were fairly high at depths of 700-1200 m, although D. calcea was most abundant at depths of less than 1000 m, E. granulosus peaked 1,200, 1,000 ^ II ^ Depth (m) Figure 4 Density (kg/km^) of sharks collected at all locations on Chatham Rise, New Zea- land, at various depth intervals for both surveys combined. For abbreviations of species see Figure 3. For each species, P -values are given in parentheses for ANOVA comparisons of densities among depth intervals. at 900-1200 m, and Apristurus spp. densities were highest at depths greater than 1000 m (Fig. 4). At depths greater than 1200 m, Apristurus spp. were the only sharks regularly caught in the trawls. There was a significant difference among three depth inter- vals (700-1000 m, 1000-1300 m, and 1300-1- ml for mean densities (kg/km^) of each species in the orange roughy survey, but only for the three most abun- dant species iD. calcea, C. crepidater, and E. granu- losus) in the oreo survey (ANOVA, P<0.05). Density (kg/km^) for all sharks combined gradually declined with depth between 700 and 1200 m, but was low at depths greater than 1200 m (Fig. 5). Distribution Diversity (number of species per trawl) was signif- icantly higher for the orange roughy survey than for the oreo survey (Mest, P=0.0003), but diversity in areas common to both surveys (areas 6, 8, 9, 10) did not differ (Mest, P=0.14, 0.77, 0.81, and 0.63 respectively). There was a significant differ- ence among areas for mean diversity values in both surveys (ANOVA, P<0.01). Area 3 had the highest mean diversity value (4.2, SD=1.6) and area 10 had the lowest value (1.1, SD=0.9). Diversity dif- fered significantly among the three major depth intervals (700-1000 m, 1000-1300 m, 1300-(- m) for the oreo survey (ANOVA, P=0.007), but not for the orange roughy survey (ANOVA, P=0.50). The total number of species of shark caught in trawls was inversely proportional to depth, and declined by half from a max- imum often (800-900 m) to only five species at depths of 1200- 1300 m (Fig. 5). The index of similarity (S) between the two surveys was high (0.89), and for each survey there was a high degree of similarity between areas, except for those most distant from each other, par- ticularly between area 10 and other areas (Table 2). Indices of similar- ity between depth intervals were very similar for both surveys: high (0.80 for the orange roughy survey and 0.81 for the oreo survey) for the two shallowest depth intervals ( 700-1000 m versus 1000-1300 m), moderate (0.57 and 0.63) for the 1000-1300 m vs 1300-1- m intervals, and much lower (0.42 and 0.47) for S D.l. 10.270) C.s. (0.0021 n S.p. (0.711) A.s. (0-769) H D.c. (0.001) n C.c (0.001) Co. (0.006) □ E.g. (0.364) Wetherbee: Assemblage of deep-sea sharks on Chatham Rise, New Zealand 195 700-1000 m versus 1300+ m intei-vals. Morisita's index of dispersion (/^) indicated that all species of sharks tended to be aggre- gated on Chatham Rise to some degree. The three large squalid sharks were the most aggre- gated: S. plunketi (7^=104.5), C. squamosus (82.4), D. licha ( 55.4 ), followed hyE. granulosus {24.3), Apristurus spp. (21.2), C. owstoni (10.9), D. calcea (8.2), and the most randomly distrib- uted was C. crepidater {1^-5.4}. Discussion Kg sharli per km - Number of Species 12 10 3 Sharks are abundant and widely distributed on Chatham Rise. Even though sharks form a relatively small percentage of the overall catch in deep-water trawls they are frequently caught by the hundreds and occasionally dominate catches. The most abundant shark (by weight) on Chatham Rise, D. calcea, was often caught in large numbers, which suggests the presence of large aggregations. However, this species was caught in a high percent- age of trawls and was widely distributed on Chatham Rise, which resulted in a fairly low index of disper- sion (/^). D. calcea is abundant elsewhere in New Zea- land waters, accounting for as much as 70% of the shark catch off the North Island (Clark and King'). Kobayashi (1986) reported that D. calcea was one of the most common sharks in deep-water catches from Japan as well. The most ubiquitous shark in terms of presence at nearly all depths and locations on Cha- tham Rise was E. granulosus, although this species may have a fairly limited distribution outside of New Zealand and southeast Australia. Tachikawa et al. (1989) synonimized the New Zealand lantern shark, E. baxterl, with the widely-distributed southern lan- tern shark, E. granulosus; however, there may be sev- eral species within the E. granulosus group and E. baxteri may well be a valid species (Compagno et al., 1991; Wetherbee, 1996). Catsharks captured in trawls were keyed out to five undescribed species ( A-E ) belonging to the genus Apristurus (Paulin et al., 1989). That several hundred specimens of these undescribed sharks were collected, underscores the paucity of information on deep-sea sharks. This genus also contains many undescribed Depth (m) Figure S Average density (kg/km- ±SDl of all sharks combined and number of species captured within various depth intervals on Chatham Rise, New Zealand, for both surveys com- bined. For comparison of mean densities among intervals with ANOVA, P = 0.011, df=389. Table 2 Bray-Curtis index of simi arity between geographical areas on Chatham Rise, Vew Zealand for a 1990 orange roughy survey and a 1993 smooth oreo survey. For locat on of areas see Figure 1. Orange roughy survey Area 2 3 4 5 6 7 9 10 1 0.87 0.84 0.84 0.74 0.69 0.71 0.56 0.38 2 0.89 0.87 0.85 0.78 0.71 0.56 0..36 3 0.84 0.82 0.82 0.68 0.53 0.34 4 0.87 0.85 0.75 0.61 0.42 5 0.89 0.65 0.58 0.36 6 0.72 0.64 0.42 8 0.81 0.60 9 0.71 Oreo survey Area 7 8 9 10 6 0.76 0.71 0.35 0.32 7 0.76 0.47 0.37 8 0.57 0.53 9 0.85 species in other locations, such as Australia (Last and Stevens, 1994), and may eventually be recognized as one of the most speciose genera of sharks. 196 Fishery Bulletin 98(1) Very few individuals of three species of relatively large squalid sharks iCentrophorus squamosus, Scym- nodon plunketi, and Dalatias licha) were captured on Chatham Rise. These three species also happen to have liver oil that is high in squalene, or that has a high diacyl glycerol ether to triglycerol ratio, and the liver oil of these sharks is thus of high quality for industrial purposes (Bakes and Nichols, 1995; Weth- erbee, 1998). Because each of these species formed less than 1% of the total shark catch in all fishing, targeting of any of these species by commercial fisher- ies on Chatham Rise does not appear to be practical. However, these species have been captured in greater numbers in fishing that was conducted at shallower depths and that targeted different fishes at locations other than Chatham Rise in New Zealand (Clark and Kingi). The consistency in catch rate, regardless of time of day, indicates that there are few changes in the distribution of the deep-sea shark community on a diurnal basis. However, capture in a trawl may not provide information on activity patterns or feeding periodicity. For example, Kobayashi (1986) found that capture rate of sharks was higher at night than during the day in deep-sea fishing with baited lines. The present study was conducted over a three-month period and does not provide much information on seasonal differences in the distribution of deep-sea sharks. Some studies have suggested that there is continual movement of reproductive groups, or age classes, out of a particular area (Yano, 1991; Weth- erbee, 1996). Other studies have maintained that community structure, temperature, and salinity of the deep-sea environment vary little throughout the year (Kobayashi, 1986; Clark and KingM. Orange roughy appear to be common prey of two species of sharks (E. granulosus and C. owstoni) and are also consumed by two of the less common, large squalids (C. squamosus and D. licha ). Centroscymnus owstoni may exert the greatest predation pressure on orange roughy populations because both species are found in large numbers on the north of Chatham Rise. An expanded investigation of the feeding habits of sharks would provide more information on the relation- ship between sharks and commercially important tele- osts on Chatham Rise. Abundance Differences in abundance of sharks between the orange roughy and oreo surveys appear to be attrib- utable to the location at which fishing was concen- trated in each survey. For the oreo survey, fishing was restricted to the south of Chatham Rise, and during the orange roughy survey, fishing was con- centrated on the north of Chatham Rise (although trawls were made throughout Chatham Rise). The observation that there were no significant differences between the catches in areas common to both surveys indicates that overall differences between the sur- veys were probably not due to differences in fishing methods (duration of trawls and net specifications) or time (season or year). The contribution of sharks to total biomass of each survey was also remarkably similar. The composition of the shark community was dependent upon the area of Chatham Rise that was sampled. Moving from west to east on the north of Chatham Rise, the dominant species in trawls changed from C. owstoni to D. calcea. Fishing areas at the eastern tip of Chatham Rise were the most diverse. There, the highest densities (kg/km^) were recorded for six of the eight species of sharks and for all sharks combined. Etmopterus granulosus was the most abundant shark on the south of Chatham Rise, completely dominating the catch in the westernmost areas. Variation in the composition of the deep-sea shark community with location has been observed in other areas off New Zealand, and also in Japan and southern Africa (Kobayashi, 1986; King and Clark, 1987; Compagno et al., 1991). In several studies, the community of deep-sea sharks was thought to vary with latitude (Merrett and Marshall, 1981; Nakaya and Shirai, 1992; Yano and Kugai, 1993). The distri- bution of sharks on Chatham Rise may be influenced by a number of physical and biological factors. How- ever, the lack of information on the deep-sea envi- ronment in this area precluded examination of the relationship between shark abundance and either biotic or abiotic factors; in contrast to other studies (Graham and Hastings, 1984; Bianchi, 1991; 1992; Garcia et al., 1998). Large portions of Chatham Rise are as shallow as 500 m, and sharks with fairly deep distributions might not move freely between the north and south slopes. Thus, the patterns of distribution observed for several species of sharks near Chatham Rise may be related to this physical barrier. In this study, shark abundance also varied with depth. The depth range at which maximum shark density was recorded in the present study ( 700—800 m) was deeper than that reported by Kobayashi (1986) (300-500 m), and shallower than that found by Yano and Kugai ( 1993 ) ( 1100-1200 m). Depth dis- tributions of sharks caught on Chatham Rise are characterized by several patterns. Some of the larger species were rare at depths greater than 1100 m. which may approximate their maximum depth of occurrence. The density of other species declined abruptly beyond 1200 m, and the proportion of Apris- Wetherbee: Assemblage of deep-sea sharks on Chatham Rise, New Zealand 197 tiirus spp. in the catch increased at depths greater than 1200 m. The depths fished in the present study did not appear to reveal the minimum depth of occurrence for any species of shark. Compagno et al. ( 1991 ) noted the importance of determining mini- mum depth of occurrence describing the depth dis- tribution of a particular species. On Chatham Rise, density (kg/km-) of all sharks combined was fairly constant up to about 1200 m but dropped drastically beyond this depth. Nakaya and Shirai (1992) observed a similar dramatic decrease in shark density at 500 m, and Merrett and Marshall (1981) at 1000-1100 m. An inverse relation between shark abundance and depth has been described for a number of species in other regions of New Zealand, and in other parts of the world (Merrett and Mar- shall, 1981; Kobayashi, 1986; King and Clark, 1987; Yano and Kugai, 1993). Distribution Diversity (species of shark/trawl) was higher for the orange roughy surveys than for the oreo sui-veys, but there was no significant difference in mean diversity between areas common to both surveys. These obser- vations support the contention that differences in the shark catch between surveys are related to sam- pling location, rather than to temporal or method- ological differences between surveys. However, there were generally more trawls within each area during the orange roughy survey than for the oreo survey, which may have increased the total number of spe- cies caught on the north of Chatham Rise. Hill ( 1973) predicated that as the size of a sample is increased, so almost without limit will the diversity. Diversity also declined with increased depth on Chatham Rise, and Crowder (1990) suggested that such a decline in species diversity might be due to changes in the level of competition, predation, or environmental homogeneity. In this study, only 16 species of shark were caught, although many more species of deep-sea shark have been captured in New Zealand (Paulin et al., 1989). The total of only 16 species caught in the present study is also low in comparison to numbers (>30) of species caught in deep-water surveys in other parts of the world (Kobayashi, 1986; Compagno et al., 1991; Nakaya and Shirai, 1992; Yano and Kugai, 1993). Much of the fishing on Chatham Rise was conducted at depths of greater than 1000 m, which may be beyond the depth limit of a number of spe- cies of squalid sharks found in New Zealand waters (Yano, 1985; Compagno et al., 1991). The index of similarity between the orange roughy and oreo surveys was high, again suggesting that differences introduced as a result of variable fishing methods or time were probably not substantial. The nearly identical indices of similarity between depth intervals for each survey also support this conclu- sion. The high indices of similarity between areas for most species, along with the fairly low indices of dispersion, indicate that although their abundance is variable, most species have fairly wide distribu- tions on Chatham Rise. Sharks within the genera Etmopterus and Centro- phorus are thought to segregate by species in Jap- anese waters (Kobayashi, 1986; Baba et al., 1987; Yano and Tanaka, 1983). In the present study there was little evidence to suggest that any two species displayed such segregation. Compagno et al. (1991) found that Centroscymnus spp. were sympatric but had very different food habits. An examination of dietary overlap among sharks common on Chatham Rise may reveal whether these sympatric species compete for the same food resources. Although the sharks captured during this study were incidental to commercially important fishes, such as orange roughy and smooth oreo, the data collected from these trawls have provided informa- tion on the abundance and distribution of a number of species of deep-sea shark on Chatham Rise. Dis- tributional patterns of sharks vary among species, and the composition of the deep-sea shark commu- nity varies with depth and location. Therefore, the overall impact of deep-water trawl fisheries on shark populations would be expected to vary among spe- cies and to depend on the particular fishery, which in turn influences the location and depth where fishing is concentrated. Acknowledgments I thank A. Bush, T. Clarke, K. Holland, S. Kajiura, C. Lowe, C. Meyer, C. Mostello, and J. Parrish for their comments on the manuscript. P. Grimes, P. McMillan, and K. Mulligan were tremendously help- ful with collection of specimens and access to trawl data. K. Fields, J. Fenaughty, M. Clarke, A. Hart, and the captains and crews of the RV Tangaroa and FV Cordelia were instrumental in collection of data. J. Parrish, D. Yount, and B. Flannigan made funds available for travel to New Zealand. Literature cited Baba, O., T. Taniuchi, and Y. Nose. 1987. Depth distribution and food habits of three species of small squaloid sharks off Chosi. Nippon Suisan Gakkai- shi 53(31:417-424. 198 Fishery Bulletin 98(1) Bakes, M. J., and P. D. Nichols. 1995. Lipid, fatty acid and squalene composition of liver oil from six species of deep-sea sharks collected in south- ern Australian waters. Comp. Biochem. Physiol. 110(Bl :267-275. Bianchi, G. 1991. Demersal assemblages of the continental shelf and slope edge between the Gulf of Tehuantepec (Mexico) and the Gulf of Papagayo (Costa Rica ). Mar. Ecol. Prog. Ser. 73; 121-140. 1992. Demersal assemblages of the continental shelf and upper slope of Angola. Mar. Ecol. Prog. Ser. 81:101-120. Brower, J. E., and J. H. Zar. 1984. Field and laboratory methods for general ecology. W.C. Brown Publishers, Dubuque, lA, 226 p. Clark, M. R., K. J. King, and P. J. McMillan. 1989. The food and feeding relationships of black oreo, A//o- cyttus niger, smooth oreo, Pseudocyttus maculatus, and eight other fish species from the continental slope of the south-west Chatham Rise, New Zealand. J. Fish Biol. 35:465-484. Clark, M. R., and D. M. Tracey. 1994. Changes in a population of orange roughy, Hoploste- thus atlanticus, with commercial exploitation on the Chal- lenger Plateau, New Zealand. Fish. Bull. 92:236-253. Compagno, L. J. V., D. A. Ebert, and P. D. Cowley. 1991. Distribution of offshore demersal cartilaginous fish ( class Chondrichthyes ) off the west coast of southern Africa, with notes on their systematics. S. Afr. J. Mar. Sci. 11:43-139. Crowder, L.B. 1990. Community ecology, /n C. B. Schreck and P. B. Moyle (eds. ), Methods for fish biology, p. 609-632. Am. Fish. Soc, Bethesda, MD. Davenport, S., and P. Deprez. 1989. Market opportunities for shark liver oil. Aust. Fish. 11:8-10. Deprez, P. P., J. K. Volkman, and S. R. Davenport. 1990. Squalene content and neutral lipid composition of livers from deep-sea sharks caught in Tasmanian waters. Aust. J. Mar Freshwater Res. 41:375-387. Ebert, D. A., L. J. V. Compagno, and P. D. Cowley. 1992. A preliminary investigation of the feeding ecology of squaloid sharks off the west coast of southern Africa. S. Afr. J. Mar. Sci. 12:601-609. Field, J. G., K. R. Clarke, and R. M. Warwick. 1982. A practical strategy for analysing multispecies distri- bution patterns. Mar. Ecol. Prog. Ser. 8:37-52. Garcia, C. B., L. O. Duarte, and D. von Schiller. 1998. Demersal fish assemblages of the Gulf of Salamanca, Columbia (southern Caribbean Sea). Mar. Ecol. Prog. Ser. 174:13-25. Graham, J. H., and R. W. Hastings. 1984. Distributional patterns of sunfishes on the New Jersey coastal plain. Env Biol. Fish. 10:137-148. Hill, M. O. 1973. Diversity and evenesss: a unifying notation and its consequences. Ecology 54:427-432. King, K., and M. Clark. 1987. Sharks from the upper continental slope — are they of value? Catch ( May ):3-6. Kobayashi, H. 1986. Studies of deep-sea sharks in Kumano-nada Region. Bull. Fac. Fish. Mie Univ 13:25-133. Last, P. R., and J. D. Stevens. 1994. Sharks and rays of Australia. Commonwealth Sci- enfiic Industrial Research Organization (CSIRO), 513 p. Merrett. N. R., and N. B. Marshall. 1981. Observations on the ecology of deep-sea bottom-liv- ing fishes collected of northwest Africa (08°-27°N). Prog. Oceanogr. 9:185-244. Nakaya, K., and S. Shirai. 1992. Fauna and zoogeography of deep-benthic chondrich- thyan fishes around the Japanese Archipelago. Jpn. J. Ichthyol. 39(l):37-48. Paulin, C, C. Roberts, A. Stewart, and P. McMillan. 1989. New Zealand fish: a complete guide. National Mu- seum of New Zealand Miscellaneous Series 19, Wellington, 279 p. Sedberry, G. R., and R. F. Van Dolah. 1984. Demersal fish assemblages associated with hard bottom habitat in the South Atlantic Bight of the U.S.A. Env Biol. Fish. 11:241-258. Stephens, J. S. Jr., P. A- Morris, K. Zerba, and M. Love. 1984. Factors affecting fish diversity on a temperate reef: the fish assemblage of Palos Verdes Point, 1974-1981. Env Biol. Fish. 11:259-275. Summers, G. 1987. Squalene — a potential shark by-product. Catch (Oc- tober) 1987:29. Tachikawa, H., T. Taniuchi, and R. Aral. 1989. Etmoplerus baxten, a junior synonym of E. granu- losus (Elasmobranchii, Squalidae). Bull. Nat. Sci. Mus., Tokyo Ser. A 15:235-241. Wetherbee, B. M. 1996. Distribution and reproduction of the southern lantern shark from New Zealand. J. Fish Biol. 49:1186-1196. 1998. Biochemical and physiological buoyancy adaptations in deep-sea sharks. Ph.D. diss., Univ. Hawaii, Honolulu, HI, 144 p. Yano, K. 1985. Studies on morphology, phylogeny, taxonomy and biol- ogy of Japanese squaloid sharks, order Squaliformes. Ph.D. diss., Tokai University, Shimizu, 335 p. 1991. Catch distribution, stomach contents and size at maturity of two squaloid sharks, Deania calceus and D. crepidalhus from the southeast Atlantic of Namibia. Bull. Jpn. Soc. Fish. Oceanogr. 55:189-196. Yano, K., and K. Kugai. 1993, Deep-sea chondrichthyans collected from the waters around the Okinawa Islands: results of catch analysis of bottom longlines. Bull. Seikai National Fish. Res. Inst. 71:51-65. Yano, K., and S. Tanaka. 1983. Portuguese shark, Centrophorus coelolepis from Japan, with notes on C. owstoni. Jpn. J. Ichthyol. 30:208- 216. 199 Abstract.-Sagittae (n=2,263) and gonads (?i=870) from snowy grouper, Epmephelus nireatus, caught primar- ily with longlines. Kali poles, snapper reels, and chevron traps ofT North Caro- lina and South Carolina were examined 1) to compare growth rates, population age structure, and sex ratio between two periods 1979-85 and 1993-94, and 2) to determine reproductive seasonal- ity, size and age at maturity, and size and age at sex change. There were sev- eral indications that the population off North Carolina and South Carolina is overfished; 1 ) size at age of specimens caught with longlines and snapper reels has increased noticeably since the 1980s (possibly a a density-dependent popula- tion response to a high level of fishing mortality ); 2 ) 8 1'^f of the fish caught with commercial longlines during 1993-94 were ages 1-6. the majority (SB*?) of which were immature females; 3) the percentage of males in the population appears to have decreased significantly, from 7-23<7f in the 1970s and 1980s to 1% in the 1990s; and 4) mean length of fish landed in the longline fishery has steadily decreased from 65-80 cm in the early 1980s to 50-60 cm in the mid-1990s. There was a positive trend between water depth and total length in fishery-independent samples. Histo- logical examination of gonads revealed that mature gonads were present in 4% of the females at age 3, 52*^ at age 5, 95*7^ at age 7, and 1007c at ages >7 during 1993-94. The smallest mature female was 469 mm TL, and the larg- est immature female was 575 mm. Esti- mates of Lengthjy and Age^^ were 541 mm (95'» CI=529-553 mm) and 4.92 yr OS'/f CI=4.65-5.21 yr), respectively. Spawning females were caught during April through September on the upper continental slope off South Carolina at depths of 176-232 m. The size ( 767-1090 mm I and age i8-29 vr) of 97 male speci- mens and the capture of two specimens undergoing sex change provided conclu- sive evidence that snowy grouper are protogynous hermaphrodites. Growth, population age structure, and aspects of the reproductive biology of snowy grouper, Epinephelus niveatus, off North Carolina and South Carolina* David M. Wyanski D. Byron White Charles A. Barans South Carolina Department of Natural Resources Marine Resources Research Institute RO Box 12559 Charleston, South Carolina 29422 2559 E-mail address (for D M Wyanski) wyanskidiamrd.dnrstate.sc.us Manuscript accepted 26 August 1999. Fish. Bull. 98; 199-2 18 12000). The snowy grouper, Epinephelus niveatus, is a commercially impor- tant deepwater species that occurs in the western Atlantic from North Carolina (Cape Hatteras) to Brazil, including the Gulf of Mexico and the Bahamas (Smith, 1971). It also occurs in the eastern Pacific from California to Mexico (Miller and Lea, 1976; Fitch and Schultz, 1978 ). Along the coast of the southeast United States, adult snowy grou- per are predominantly found on the upper continental slope ( >75 m; Lee et al., 1985) at depths of 116-259 m (Low and Ulrich, 1983; Moore and Labisky, 1984; Parker and Ross, 19861, whereas juveniles are more common at shallower depths ( Moore and Labisky, 1984). Low and Ulrich (1983) noted a positive correlation between total length (TL) and water depth off South Carolina. Most fish- ing for this species occurs in habi- tats characterized by rocky ledges, cliffs, and swift currents (Matheson and Huntsman, 1984). Snowy grouper are captured pri- marily in commercial fisheries of the southeastern United States ( Parker and Mays, 1998). Most are caught with bottom longlines and snapper reels^ (handlines). Starting in 1991, the longline fishery was restricted to waters deeper than 91 m by the South Atlantic Fishery Management Council (SAFMC, 1991). Fishing for snowy grouper has occurred off North Carolina and South Caro- lina (the Carolinas) since the mid- 1950s (Huntsman, 1976); annual landings from the Carolinas aver- aged 119,657 kg during 1981-96 (Moran-). According to the spawn- ing stock ratio (SSR), the popula- tion off the Atlantic coast of the United States is considered to be overfished (SAFMC, 1993). Population age structure and individual growth rates in the snowy grouper population have not been assessed since the mid-1980s and should be assessed again given the sustained fishing pressure on the population. Studies of other fish populations have shown that size at age is often affected by the level and duration of exploitation ( Haug and Tjemsland, 1986; Harris and McGovern, 1997; Helser and Contribution 430 of the South Carolina Manne Resources Center, P.O. Box 12559. Charleston, SC 29422. ' Snapper reels are commonly known as '1)an- dits" or "one-armed bandits' by fishermen owing to the remote similarity between early snapper reels and gambling slot machines, and because luck is involved in what is caught. The early mechanical reels have typically been replaced by 12 volt DC auto- mobile starter motors or hydraulic systems. - Moran.J. 1996. S. Carolina Dept. of Nat- ural Resources, P.O. Box 12559, Charleston, SC 29422. Personal commun. 200 Fishery Bulletin 98(1) Almeida, 1997; Zhao etal., 1997; Goodyear and Schir- ripa^), thus trends in growth can be indicators of population stability. Additionally, nothing is known about trends in sex ratio. Recent studies along the southeast Atlantic coast and in the Gulf of Mexico have documented sharp decreases in the proportion of males in other grouper, gag (Mycteroperca microl- epis) and scamp (M. phenax), populations (Coleman et al, 1996; McGovern et al., 1998). This and other factors have led to reduced genetic diversity in the gag population along the southeast coast, a cause for serious concern, although the ramifications are not clearly understood (Chapman et al., 1999). Little is known about the reproductive biology of snowy grouper off the Carolinas because the only pre- vious study ( Moore and Labisky, 1984 ) was conducted in the Florida Keys. Using a histological technique, Moore and Labisky (1984) showed that the snowy grouper is a protogynous hermaphrodite; females reach sexual maturity at ages 3-5 and change to males as early as age 6. The objectives of the present study were 1) to compare individual growth rates, population age structure, and sex ratio between two periods, 1979-85 and 1993-94, and 2) to determine reproductive seasonality, size and age at maturity, and size and age at sex change for snowy grouper off the Carolinas. Materials and methods Specimen acquisition Snowy grouper were obtained from commercial boats, research vessels, and headboats, primarily off North Carolina and South Carolina (Table 1). All speci- mens were collected between 31°09'N and 34°44'N at depths of 42-302 m. Only seven specimens were caught south of 32°04'N. Fishery-independent sam- ples were collected during cruises of the MARMAP (Marine Resources Monitoring Assessment and Pre- diction) program with bottom longlines. Kali poles (an off-bottom longline; Russell et al., 1988), snap- per reels, rods and reels, and chevron traps (Collins, 1990). Specimens caught with longlines were col- lected primarily off South Carolina, whereas speci- mens caught with snapper reels were collected off South Carolina in the 1980s and primarily off North Carohna in the 1990s (Fig. 1). Total length (mm) was measured for all specimens and all length measure- ments in the text refer to total length (TL). During 1993-94, samples from the commercial fishery usually included the total catch of snowy grouper from a vessel or a random subsample; on three of 20 occasions, the catches from vessels that landed fish in South Carolina were subsampled to collect otoliths from small and large specimens. No documentation on sampling design was available for fishery-dependent samples that were collected with snapper reels during 1979-85. Age and growth The left sagitta, and the right sagitta when time permitted, was removed and stored dry prior to processing. Each otolith was embedded in paraffin ( 1979-85) or Araldite epoxy resin ( 1993-94) and sec- tioned along a dorsoventral plane through the focus with a single high-concentration diamond wheel on a Buehler Isomet low-speed saw. Otolith sections were mounted on glass slides with Crystalbond thermo- plastic or Accu-mount 60, covered with cedar wood oil, and examined under a dissecting microscope ( 10-63x) with reflected and transmitted light. We examined otoliths from 1937 specimens and used the age determinations (/!=326) of Waltz'* for speci- mens collected with snapper reels from 1979 to 1985. The width of the translucent zone along the margin of the section was measured to assess the periodicity of increment formation. Increments were counted inde- pendently by two readers for 1853 of 1937 specimens. In older fish, all increments could not be counted along one axis in many specimens. Counting commenced on one of three axes (ventral, ventromedial, or adjacent to the sulcus acousticus) and shifted to another axis by following an increment to the new axis. If counts dif- fered, both readers examined the otolith by projecting the image onto a TV monitor The otolith was rejected if agreement could not be reached. A small portion ( /; = 129 ) of the 1937 otoliths that we examined were used for an earlier MARMAP study (Waltz"*). Age assessments were compared to deter- mine if annual increment structure was being inter- preted in a similar manner. The specimens selected for the comparison were collected with longlines or Kali poles on research cruises during 1982-85. Age data from specimens collected with these two gear types were combined because they were fished simul- taneously in the same area and were deployed with the same hook type and bait. The sagittae of three young-of-the-year (YOY) specimens were hand-polished to thin (approx. 5pm) ■* Goodyear, C. P., and M. J. Schirripa. 1993. The red grouper fishery of the Gulf of Mexico, Report MIA92/93-75. Miami Lab- oratory, Southeast Fisheries Science Center, National Marine Fisheries Service, 75 Virginia Beach Dr, Miami, FL 33149. ■• Waltz, W. 1986. The size and age of snowy grouper (Ep/nep/i- elus niveatus) in the South Atlantic Bight. MARMAP Analytical Report, 16 p. S. Carolina Department of Natural Resources, PO. Box 12559, Charleston, SC 29422. Wyanski et a\ : Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 201 34''N SS^N - O Longline A Snapper reel X^ Trawls Figure 1 Approximate locations of commercial fishing effort with longlines and snapper reels for snowy grouper off North Carolina and South Carolina between June 1993 and September 1994. Snowy grouper were also caught in trawls in June 1978 during an exploratory squid cruise conducted by the government of Spain and the National Marine Fisheries Service on the RV Pescapuerta Segundo. Trawl site off Georgia is not shown. Table 1 Numbers of specimens of snowy grouper for which otoliths and gonad samples were examined by gear ings; H = headboats sampled by National Marine Fisheries Service, Beaufort, North Carolina; R = res Marine Resources Monitoring Assessment and Prediction Program (MARMAP). type, earch C = commercial land- cruises conducted by Gear Source Period Otoliths Gonads Snapper reel C,R 1979-85 326 309 Longline and Kali pole most R, C 1982-85 190 180 Rod and reel most H, R 1973-81 90 Other C,R 1980-84 10 Snapper reel C 1991-95 335 32 Longline C 1993-94 1332 146 Chevron trap R 1991-95 78 100 Other R 199.3-95 2 3 Total 2263 870 transverse sections according to the methods of Secor et al. (1992) and examined with a compound micro- scope. We counted the regular concentric rings, simi- lar to those reported as daily growth increments in other species, to estimate age. Assumed daily rings were visible, with the exception of a large opaque core area, on the smallest otolith but were only dis- tinguishable at discrete locations on the other two otoliths. For these two otoliths, age was estimated by extrapolating the increment count per unit dis- tance to the total radial measurement of the ven- tral axis, excluding the core. Radial measurement 202 Fishery Bulletin 98(1) from the core to the first annulus was made for a subsample of 23 specimens that were age 1 and com- pared with the measurements from the three YOY to estabhsh the position of the first annulus. Age-length keys were formed by obtaining a matrix of numbers at age by length interval for each gear type (longline and snapper reel ) in two periods ( 1980s and 1990s). Besides the differing selectivities of long- lines and snapper reels, another reason for develop- ing keys by gear type was the difference in sampling area in 1993-94 (Fig. 1) and potential differences in sampling depth due to the restriction (SAFMC, 1991) that limits the use of longlines to waters deeper than 91 m. Additional keys were generated to address two questions: 1) Do data from specimens with age estimates of lower precision affect the accuracy of the key for specimens caught with longlines in the 1990s? and 2) Are there differences in the keys for specimens caught with longlines or Kali poles during 1982-85 that were examined in the present study and in an earlier study by Waltz?^ To address the first question, keys based on all specimens and spec- imens for which there was a difference of 0-1 incre- ments between readers were compared. All analysis of age and growth data was conducted with SAS software (SAS Institute, Inc., 1990). Fish- er's exact test ( Siegel, 1956 ) was used to compare the age distributions of two age-length keys by 25-mm length intervals. The FREQ procedure was used to run this test. A comparison was made only if each key had >6 specimens in an interval because of the low power associated with small sample sizes (Bennett and Hsu, 1960). The large number of tests required to compare age-length keys necessitated compensating for experimentwise error by comput- ing an adjusted significance level (a*) using the for- mula presented by Hayes (1993). Nonlinear regression analysis with Marquardt's algorithm (NLIN procedure) and the NLIN weight statement were used to fit the von Bertalanffy growth model to observed length at age data (von Berta- lanffy, 1938). Lengths were weighted by the inverse of the number of fish at each age to moderate the effect of large and small sample sizes on the esti- mates of growth parameters. Age-length keys were applied to length data collected through the Trip Interview Program (TIP) in the Carolinas to gener- ate an age-frequency distribution. TIP is a commer- cial fisheries data collection program funded by the National Marine Fisheries Service (NMFS). landed whole by fishermen (Table 1). Ninety gonad samples from the headboat fishery during 1973-1981, collected in association with the study of Matheson and Huntsman ( 1984), were obtained from the Beau- fort Laboratory of the NMFS. The posterior portion of the gonad was fixed in 10*^ seawater-formalin for 1-2 weeks and transferred to 509^ isopropanol for 1-2 weeks. Gonad samples were processed, sectioned, and stained with double-strength Gill hematoxylin and eosin-y by using the methods of Schmidt et al. (1993). Sex and reproductive state were assessed primar- ily by one reader using histological criteria (Table 2), without reference to body length or date of capture. A second reader examined sections from 75 specimens to ensure accurate interpretations. If the assessments of the two readers differed, the section was viewed simultaneously by the readers and rejected if agree- ment could not be reached. Specimens with develop- ing, ripe, spent, or resting gonads were considered sexually mature. For females, this definition of sexual maturity included specimens wdth oocyte development at or beyond the cortical granule stage and speci- mens with beta, gamma, or delta stages of atresia (see Hunter and Macewicz, 1985). To ensure that females were correctly assigned to either the immature or resting categories, the length-frequency histogram of females with evidence of certain maturity (e.g. those that were developing, ripe, or spent) was compared with the histograms for immature and resting females. Females of uncertain maturity (Table 2) were excluded from data analyses. To estimate length at SO'X matu- rity (L50) and age at 50*7^ maturity (Ajg), the PROBIT procedure (SAS Institute, Inc., 1990) was used to fit gompit, logit, or probit models to maturity data in 25-mm length intervals or one year increments. The LOGISTIC procedure was used to determine which model to use in the PROBIT procedure. Females with hydrated oocytes or postovulatory follicles were considered to be in spawning condition. Macroscopic observations of snowy grouper caught in June 1978 during trawls (Fig. 1) of an explor- atory squid cruise conducted jointly by the govern- ment of Spain and the Northeast Fisheries Center of NMFS in Woods Hole, Massachusetts, on the RV Pescapuerta Segundo were also used to define the area and timing of spawning. Results Reproduction Gonads were obtained during 1979-95 from 870 spec- imens collected on research cruises and from fish Age and growth An age was assigned to 91.6'^ of 1937 otoliths that we examined (Table 1). Otoliths were rejected if the Wyanski et al : Growth, population age structure, and aspects of the reproductive biology of Epinephdus niveatus 203 Table 2 Histological criteria used to determine reproductive stage in snowy grouper (see Hunter and Goldberg, 1980; Wallace and Selman, 1981; Hunter and Macewicz, 1985; Wenner et al., 1986; West, 1990). Reproductive stage Criteria Immature Developing Ripe Developing, recent spawning Spent Resting Uncertain maturity Transitional Previtellogenic oocytes only, no evidence of atresia. In comparison with resting female, most previ- tellogenic oocytes are <70 pm, area of transverse section of ovary is smaller, lamellae lack muscle and connective tissue bundles and are not as elongate, oogonia are abundant along margin of lamellae, and ovarian wall is thinner. Oocytes undergoing cortical granule (alveoli) formation through nucleus migration and partial coalescence of yolk globules. Completion of yolk coalescence and hydration in the most advanced oocytes. Zona radiata becomes thinner. Developing stage as described above plus presence of postovulatory follicles. More than 50% of vitellogenic oocytes in alpha or beta stage of atresia. Previtellogenic oocytes only with traces of atresia possible. In comparison with immature female, most previtellogenic oocytes >70 pm, area of transverse section of ovary is larger, lamellae have muscle and connective tissue bundles, lamellae are more elongate and convoluted, oogonia are less abundant along margin of lamellae, and ovarian wall is thicker and exhibits varying degrees of expansion owing to previous spawning. Immature or resting. Inactive ovaries, previtellogenic oocytes only. Reproductive stage is uncertain. Proliferation of spermatogonia through limited spermatogenesis within lamellae of resting ovary, accompanied by development of peripheral sinuses in musculature of ovarian wall. readers couM not agree on the age or if the section was not adequate. Ages ranged from 1 to 29 yr and lengths from 226 to 1137 mm. From a subsample (n=21A) of 3-10 yr old specimens, we found that the mean width of the sagittal marginal translucent zone was smallest in April and May, indicating the period of increment formation ( Fig. 2 ). The unimodal nature of the data indicated that one increment was deposited per year. Data from longlines and snapper reels showed that size at age was greater during 1993-94 than during the previous decade (Figs. 3 and 4). Snowy grouper captured with longlines in 1993-94 exhibited a nearly constant growth rate until approximately age 10, after which there was a notable decrease. A similar growth pattern was noted for snowy gi-ou- per caught with snapper reels, although the trend was less definitive owing to smaller sample sizes. Estimates of theoretical maximum length (Lj were reasonable when compared with maximum observed lengths (Table 3). Data from both gear types indi- cated that L ., has increased 169-231 mm in the last decade. The application of the age-length key for samples caught with longlines during 1993-94 to TIP length data for the same period revealed recruit- ment to the fishery as early as age 1, and a modal age for recruitment of 5 (Table 4). The increase in size at age since the 1980s was also evident in comparisons of age-length keys between ■Q 80 075 70 65 0.60 55 ■a "^^ " S '•S 0.45 - ■§ I 40 - 35 - 30 - 25 A "1 — r ] J Month 1^ A -\ 1 \ 1 O N D Figure 2 Mean width (+SEl of marginal translucent zone in trans- verse sections of sagittae from 248 snowy grouper that were 3-10 yr old. periods for each gear type. For longline gear, the comparisons in 8 of 8 length intervals exceeded the adjusted significance level (P<0.00639; Table 5). The 204 Fishery Bulletin 98(1) Table 3 Von Bertalanffy parameters (± SEl desci-ibing the growth in mm total length of snowy grouper collected with snapper reels and longlines and Kali poles (LL, KP) during two decades. Mean observed length at age data were used to generate parameter esti- mates. Estimates from two earlier studies, generated using back-calculated lengths, are included for comparison. Maximum observed length L,, k ?„ 1090 970(24) 0.109(0.001) -2.123(0.336) 1110 1201(34) 0.103(0.008) -1.149(0.231) 1034 948(28) 0.122(0.017) -0.668(0.681) 1137 1117(13) 0.119(0.004) -1.409(0.121) 1130 1255 0.074 -1.92 1180 1320 0.087 -1.013 Gear Study and source Period n Waltz (1986) Snapper reels commercial, research 1979-85 326 Present Snapper reels commercial 1993-94 311 Present LL, KP, research 1982-85 163 Present LL, commercial 1993-94 1218 Matheson and Huntsman (1984) Hook and line, headboat 1972-79 536 Moore and Labisky(1984) Hook and line, research 1978-81 118 1.100 -1 • 1.000 - 900 - f%}. _ 800 - E E ~ 700 - P m n\ 1 % 600 - 500 - t F 400 - il / —^ 1982-85 300 - — •— 1993-94 1 1 1 1 1 1 1 1 5 10 15 20 25 30 35 Age increments Figure 3 Mean observed size at age ( ±SE i for snowy grouper collected with longlines and Kali poles during 1982-85 (n = 163) and longhnes during 1993-94 (;i = l,218l ofTNorth Carolina and South Carolina. trend was not as strong for snapper reel gear because comparisons in 5 of 13 intervals exceeded the 0.05 level; only one comparison exceeded the adjusted sig- nificance level (P<0.00394; Table 5). The age-length keys for the two gear types were very similar in 1993-94 because comparisons in only 3 of 17 inter- vals exceeded the 0.05 level, none of which exceeded the adjusted significance level (P<0.00284; Table 5). MOO -| • 1,000 - F^ k ' ^ 900 - d^h/ ^\ 800 - E E ~ 700 - c rjiT-i I Total le 8 8 1 1 / 400 - / / — »— 1979-85 300 - * — •— 1993-94 1 1 1 1 1 5 10 15 20 25 Age increments Figure 4 Mean observed size at age (±SEi for snowy grouper col- lected with snapper reels during 1979-85 («=326) and 1993-94 (f? =3 11) off North Carolina and South Carolina. Initial agreement between independent readers was 24.2% for the 1853 otoliths examined by two readers; however, there was a difference of 0-1 incre- ments between readers for 61. 37^ of the otoliths. The difference was >1 increment for 30.8'7f of the oto- liths and 7.9'7f were rejected as uninterpretable. The inclusion in the data sets of age data for otoliths that were difficult to interpret did not affect the assess- ment of population age distribution for snowy grou- Wyanski et al : Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 205 Table 4 Population age structure of snowy gi'ouper captured with longlines in North Carolina and South Carolina from July 1993 through May 1994. Age-length key was applied to length data from the same period collected during Trip Interview Programs. Age Number 1 5 2 32 3 77 4 172 5 209 6 163 7 82 8 23 9 13 10 8 11 6 12 7 13 4 14 2 15 3 16+ 3 Total 809 per caught with longHnes in 1993-94. Comparisons of age-length keys based on all specimens and speci- mens for which there was a difference of 0-1 incre- ments between readers revealed a strong similarity in age distribution for 20 intervals (P>0.00256; Table 6). Slightly less initial agreement was noted when our age estimates were compared with those of Waltz'* for specimens caught in 1982-85 with longlines and Kali poles. Ages were assigned to 85.3%^ of 129 speci- mens that were examined in both studies. The same age was assigned to 18. 2% of the specimens and there was a difference of 0-1 increments between studies for 40.0% of the specimens. Although the percent agreement was low, the estimates of mean size at age were similar (Fig. 5). Age distributions in the five length intervals that could be tested were similar (P>0.01021; Table 6). Given the similarity of age distributions, we decided to use the age data of Waltz^ for snowy grouper caught with snapper reels in 1979-85 without reexamining the otoliths. The low initial agreement between readers was due to a lack of a readily discernible growth pattern in many otoliths. Typical abnormalities included crys- talline areas that obscured increments (Fig. 6A) and rounded opaque deformities that distorted increment spacing ( Fig. 6B ). In addition, the axis of otolith growth frequently changed direction at least once after 6-7 Table 5 Comparison of age distributions by length interval with Fisher's exact test for snowy grouper collected off North Carolina and South Carolina with snapper reels during two periods (1979-85 and 1993-1994), with longlines during two periods ( 1982-85 and 1993-94 1, and with longlines and snapper reels during 1993-94. Dashed lines indicate that the sample size was less than seven in one or both groups. mm TL Snapper reel, 1979-85 vs. 1993-94 Longline, 1982-85 vs. 1993-94 Longline vs. snapper reel, 1993-94 226-250 — — — 251-275 — — — 276-300 — — 0.177 301-325 0.570 — 0.841 326-350 — — 0.074 351-375 0.020 — 0.488 376-400 0.933 — 0.730 401-425 0.057 — 0.407 426-450 1.000 — 0.502 451-475 0.021 — 0.036 476-500 0.630 — 0.595 501-525 0.430 — 0.043 526-550 0.019 — 0.753 551-575 0.001* — 0.323 576-600 0.209 — 0.972 601-625 0.045 <0.001 0.200 626-650 — <0.001" 0.668 651-675 0.064 J D "*«^ Q 18940007-6 ing categories (Fig. 8); only 29 females of uncertain maturity were excluded from analyses. Samples col- lected primarily with longlines and chevron traps in 1991-95 revealed that snowy grouper become sex- ually mature at lengths of 451 to 575 mm (ages 3-7; Tables 8 and 9). Mature gonads were present in 4% of females at age 3, 527f at age 5, 95% at age 7, and 100% at ages >7 (Table 9). The smallest mature female was 469 mm, and the largest imma- ture female was 575 mm. Estimates of L,;^ and Aj,^ were 541 mm (gompit model; 95% CI=529-553 mm) and 4.92 yr (probit model; 95% CI=4.65-5.21 yr). Samples collected primarily with bandit reels and longlines in 1980—85 showed that snowy grouper reached sexual maturity at similar lower limits of length and age, 476-500 mm and age 3, although sample sizes were small (<10) in the length and age intervals near the lower limit (Tables 8 and 9). Upper limits of size and age at maturity were higher, 626—650 mm and age 9, during this period and better defined owing to larger sample sizes in comparison to those for 1991-95. Mature gonads were present in 33% of females at age 3, 62% at age 5, 91% at age 7, 95% at age 9, and 100% at ages >9 (Table 9). The smallest WyanskI et al.; Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 209 Figure 7 Histological section of gonad tissue from a 787-mm-TL snowy grouper captured in July in which transition to male is nearly completed. Chromatin nucleolar (arrows) and perinucleolar oocytes are still present. Bar = lOOp. Table 7 Sex ratios of snowy grouper. Epinephelu male; F=female; T=transitional I collectec s nireatus, (M= off North Caro- lina and South Carolina during 1973-94. Ratios based on sexually mature individuals. LL=longline ;KP= Kali pole. Gear and source Period n '7cM 9cF '*r Rod and reel headboat 1973-81 83 7.2 91.6 1.2 Snapper reels commercial, research 1980-84 281 19.5 80.5 0.0 LL. KP, research 1982-85 153 22.9 76.4 0.7 LL. commercial 1993-94 82 1.2 98,8 0.0 mature female was 483 mm, and the largest imma- ture female was 634 mm. The estimate of Lj^ was 486 mm (logit model; 95Vf CI=449-509 mm) and A-^ was not estimated owing to the absence of specimens younger than age 3. A third data set, specimens col- lected by NMFS during 1973-81 primarily from head- boats, exhibited a pattern of size at maturity similar to that found for the 1980-85 samples, though sample sizes were <10 in every length intei'val and only three specimens were <55 1-575 mm (Table 8). Snowy grouper were in spawning condition from April through September based on the presence of hydra ted oocytes (Fig. 9A) and postovulatory follicles ( Fig. 9B ), with no obvious peak period ( Fig. 10 ). Given the small sample sizes for October through March, the spawning season could be longer. Ninety-nine female specimens were captured in spawning condi- tion. Seventy-two percent of the specimens were col- lected on research vessels off South Carolina (32°28' to 32"50'N ) at depths of 176-232 m, primarily during May and July through September. The remaining 27 fish in spawning condition were collected during April through August on headboats off South Caro- lina, on research vessels off North Carolina between Cape Fear and Cape Lookout, and by commercial fishermen on the upper continental slope off North and South Carolina; exact location data were not recorded. Commercial fishermen reported approxi- mate locations of 32°36' to 33°51'N and depths of 189-302 m for spawning fish. Trawl collections during exploratory squid cruises in June 1978 also provided evidence that snowy grou- per spawn on the upper continental slope (Fig. 1). Four large catches ( 1160 fish/8776 kg), which were made at depths of 180-316 ni off North Carolina, South Carolina, and Georgia (not shown), ranged from 90 to 520 snowy grouper per tow. Tow distance ranged from 7.4 to 18.5 km and estimates of snowy grouper density ranged from 2.2 fish/ha to 10.9 fish/ha (13.5 kg/ha to 79.5 kg/ha). Although the reproduc- 210 Fishery Bulletin 98(1) Table 8 Percentage of mature specimens by length interval for female snowy grouper collected off North Carolina and South Carolina with 1) primarily bandit reels and long- lines during 1980-85, 2) primarily longlines and chevron traps during 1991-95, and 3) primarily from headboats during 1973-81. Specimens in the developing, ripe, spent, or resting stages were considered mature. All specimens were examined histologically, n = number of specimens. 1980-85 1991-95 1973-81 mmTL n = 372 n = 235 n = 74 %ln) 9c (n) % (n) 251-275 (1) (3) — 276-300 — (6) — 301-325 (1) 0(10) — 326-350 (1) 0(12) — 351-375 — 0(101 0(1) 376-400 (5) 0(11) — 401-425 (3) 0(11) — 426-450 (1) 0(17) 0(1) 451-475 (2) 4(22) — 476-500 67 (6) 35(17) — 501-525 88 (8) 18(17) — 526-550 80(151 29 (7) 0(1) 551-575 100(20) 70(201 100(7) 576-600 90(21) 100(13) 86(7) 601-625 92 (24) 100 (9) 100(6) 626-650 97(311 100(12) 100(81 651-675 100(25) 100 (7) 100(9) 676-700 100(31) 100 (71 100(7) 701-725 100 (39) 100 (71 100(3) 726-750 100 (35) 100 (4) 100(6) 751-775 100 (28) 100 (3) 100(71 776-800 100(15) 100 (1) 100(31 801-1025 100(59) 100 (31 100(8) No length 100 (1) 100 (51 — tive stage of individual fish was not determined, it was noted that milt flowed freely from males in a collection of 420 fish from a depth of 278 m off South Carolina (32°57' to 33°02'N). Nearly all of the specimens were sexually mature as the mean lengths of subsamples collected off" North Carolina and off South Carolina and Georgia combined were 67.3 (48-96 cm; rt=89) and 79.2 (60-97 cm; n=98), respectively. Depth distribution There was a moderately positive trend (r^=0.53) be- tween water depth and total length in fishery-inde- pendent samples. Snowy grouper were caught at depths of 46-258 m. Larger adults were caught more frequently in upper continental slope waters >100 m, Table 9 Percentage of mature specimens by age class for female snowy grouper co Uected off North Carolina and South Car- | olina with 1 ) pri marily bandit reels and longl nes during 1980-85, and 2 primarily longlines and chevron traps | dunng 1991-95. Specimens in the developing. ripe, spent. or resting staget were considered mature. Al specimens were examined histologically, n = number of specimens. 1980-85 1991-95 Age n = 219 n = 197 (yr) % (n) 'Tc (n) 1 (5) 2 — 0(22) 3 33 (3) 4(26) 4 80 (5) 23(43) 5 62 (8) 52(40) 6 86(21) 83(29) 7 91(32) 95(21) 8 93 (30) 100 (7) 9 95 (20) 100 (11 10 10(26) 100 (1) 11 100(15) 100 (1) >11 100(59) 100 (1) X! e \ I = Immature y^ = Developing, npe or spent ^^B = Resting 60 Total length (cm) Figure 8 A comparison of length-frequency histograms for snowy grouper specimens collected during 1973-95 that were cat- egorized as immature (« = 166), definitely mature (n = 158l, or resting ( n = 130l. Definitely mature specimens were devel- oping, ripe, or spent. whereas small adults and juveniles (<600 mm) were caught more frequently at depths <100 m (Fig. 11). Wyanski et al.: Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 211 :.»,/ .^%' \S^* — ^ al Figure 9 Histological sections of ovarian tissue from female snowy grouper with evidence of imminent or recent spawning: (A) hydrated oocytes (arrow) in a 744-nim-TL specimen captured in May (bar=400p), and (B) age 24-48 h postovulatory follicles (arrows) in a 546-mm-TL specimen captured in August (bar=100 ]i). The fishery Historically, most snowy grouper landings along the Atlantic coast of the Unitecd States, as reported through TIP, have occurred in North Carolina and South Carolina (Fig. 12). Although landings sta- tistics are reported by state, fish are often caught throughout the region, especially by vessels fishing with longlines. Landings have varied widely, with peaks noted in 1983 and 1990 for South Carolina and in 1990 and 1992 for North Carolina. These fluc- tuations have often been the result of changes in effort. For example, the peak in 1983 reflected an approximate doubling in the number of vessels as 212 Fishery Bulletin 98(1) 100 16 73 69 24 42 130 127 14 60 3 Month -J Figure 10 Percentage composition (based on histological criteria) of reproductive stages by month for 510 female snowy grouper Number of specimens examined is above each bar. POP = postovulatory follicle. 1,200 1,000 800 % 600 - 400 - 200 - O 1981-85 * 1991-95 50 — I — 100 I 150 — 1 — 200 250 300 Depth (m) Figure 11 Relationship between length of snowy grouper and depth of capture in fishery-independent sampling off North Car- olina and South Carolina primarily with longhnes. Kali poles, and snapper reels during 1981-85 (^=359) and pri- marily with chevron traps during 1991-95 (n=114). the bottom longline fishery developed. The decreases in 1984 and 1985 reflect a shift in effort to the pelagic longline fishery for swordfish (Low et al., 1987). In South Carolina, the mean length of snowy grou- per caught with longlines decreased steadily from 66 to 72 cm during 1983-84 to a low of 49 cm in 1996 ( Fig. 13 ). No trend was evident in the length data for snowy grouper caught with snapper reels. The snowy grouper caught with snapper reels were consistently smaller than those caught with longlines because snapper reels were deployed in shallower water (Fig. 14). Length data from North Carolina for snowy grou- per caught with longlines showed a similar decreas- ing trend, though with greater interannual variation (Fig. 15). The mean length of snowy grouper caught with snapper reels has fluctuated, with peaks noted in 1985 and 1993. Discussion Status of the fishery There are several indications that the snowy grou- per population off the Carolinas is overfished: 1) size Wyanski et al : Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 213 at age of specimens caught with longlines and snapper reels has increased notice- ably since the 1980s (Table 3), which could be a density-dependent population response to a decrease in competition for resources, 2) 81'a^ of the specimens caught with longlines were ages 1-6, the major- ity (Se*^ ) of which were immature females ( Tables 4 and 9 ), 3 ) the percentage of males appears to have decreased significantly, from T^f to 237^ in the 1970s and 1980s to 19c in the 1990s (Table 7), 4) spawning stock ratio for the snowy grouper popula- tion in the South Atlantic Bight was 0.15 in the most recent assessment (SAFMC, 1993) — below the 0.30 level which means that the SAFMC considers the stock over- fished, and 5 ) mean length offish landed in the longline fishery has steadily decreased from 65 to 80 cm in the early 1980s to 50-60 cm in the mid-1990s (Figs. 13 and 15; see also Low, 1998). Snowy grouper are susceptible to rapid depletion in a localized area through fish- ing efforts. A study on a previously unexploited deep reef off North Carolina found that fishing can remove 3% of the reef population daily (Epperly and Dodrill, 1995). In less than three months, the catch per unit of effort and mean size of snowy grouper at that reef were reduced to levels comparable to other exploited sites. The mean size of snowy grouper landed in North and South Carolina during most of the 1990s (Figs. 13 and 15) is comparable to the size Epperly and Dodrill (1995) reported for exploited sites. The increase in size at age over a ten-year period for fish from both gear types is noteworthy because this trend has been documented in populations that had experienced moderate to high levels of fishing mortality. Increases in size at age have been noted for silver hake ( Helser and Almeida, 1997 ) and Atlan- tic halibut (Haug and Tjemsland, 1986) in the north Atlantic and several reef fish species (gag, red grou- per, and red porgy) off the southeast coast of the United States and in the Gulf of Mexico (Johnson et al., 1993; Johnson and Collins, 1994; Harris and McGovern, 1997; Goodyear and Schirripa'^). In our study, the increase in size at age may represent den- sity-dependent growth in response to an increase in fishing mortality (Rothschild, 1986). Decreases in the abundance (Low, 1998) of co-occurring species on a similar trophic level, such as gi'ay tilefish (Caulo- latilus microps) and tilefish (Lopholatilus chamaele- onticeps), also may reduce competition for food and shelter in deepwater habitats. Snowy grouper and gi-ay tilefish feed on macroinvertebrates, particularly 160 - _ 140 - o § ,20. X OO ■% 100- M 80 1 ^ 60 ^ U u 1 40 - o 20 . NC . . . SC FL A K. ^AA 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 Year Figure 12 Commercial landings of snowy grouper in North Carolina. South Caro- lina, and Florida for all gear types, primarily longlines and snapper reels. Data were collected in individual states through their Trip Inter- view Program (TIP). 1997 Year Figure 13 Mean length (±SE) of snowy grouper landed in South Carolina in the longline and snapper reel fisheries. Data were collected through the Trip Interview Program (TIPl at the South Carolina Department of Natural Resources. Sample sizes ranged from 22 to 851. crabs (Brachyura), and fishes closely associated with the substrate (Ross, 1982; Dodrill et al., 1993). Density-dependent increases in gi-owth rate gener- ally indicate that a population or community is heav- 214 Fishery Bulletin 98(1) Figure 14 Estimated mean water depth (±SE), based on minimum and maximum depths provided by captains, during fish- ing efforts in the longline and snapper reel fisheries. Data were collected through the Trip Interview Program (TIP) at the South Carolina Department of Natural Resources from vessels landing their catches in South Carolina. Sample sizes ranged from 141 to 851. ily exploited and possibly overexploited. A situation of greater concern would be one where size at age has decreased after a sustained high level of fishing- induced mortality, as has been reported for red porgy, Pagrus pagrus, and vermilion snapper, Rhombop- lites aurorubens, in our study region (Harris and McGovern, 1997; Zhao et al., 1997). There is evi- dence that faster-growing individuals in the popula- tions of red porgy and vermilion snapper have been effectively eliminated, thus causing a decrease in size at age. Red porgy exhibited a density-dependent response after an initially high level of fishing mor- tality, but the sustained high level of mortality even- tually removed the faster-growing individuals. Size at age should be monitored to ensure that this does not occur in the snowy grouper population. The age composition of the snowy grouper land- ings also needs to be monitored because the long- line fishery is presently supported by younger age classes ( 1-6). The present study showed that snowy grouper can attain an age of 29 yr, but only 19*7^ of the fish caught on longlines were >age 6 (Table 4). The low percentage of older age classes in the land- ings supports the preliminary sex ratio data from the 1990s, showing that the percentage of males had significantly decreased. • = Snapper reel o = Longline 1983 1985 1987 1989 1991 1993 1995 1997 Year Figure IS Mean length (±SE) of snowy grouper landed in North Carolina in the longline and snapper reel fisheries. Data were collected through the Trip Interview Program (TIP) at the North Carolina Department of Health, Environment, and Natural Resources. Sample sizes ranged from 5 to 1908 fish. Age and growth A comprehensive comparison of growth data in our study with previously published results was not pos- sible because 1) lack of large and old specimens and small sample sizes, and 2) differences in study area (Florida Keys in Moore and Labisky [1984|). One or more of these factors could explain the differences in size at age, k, and L^^ between the results of two published studies (Matheson and Huntsman, 1984; Moore and Labisky, 1984) and our results for long- line and snapper reel data from the 1980s. A primary reason for higher values of L^, in the published stud- ies (Table 3) is that the growth curves do not exhibit asymptotes, which is probably due to low numbers of specimens greater than approximately 900 mm and older than 15-17 yr. In our study, all the data sets (2 bandit reel and 2 longline) had individuals over 1000 mm and at least 21 yr old. Sample sizes in two data sets were very small (<200): our longline or Kali pole data set (;! = 163) and the data set of back calculations (n = 118) in Moore and Labisky (1984). Important factors that could not be evaluated on the basis of previous publications were 1) similarity of fishing gear, 2) method of increment interpretation, and 3) whether or not a weighting factor was used when fitting the von Bertalanffy growth model. Wyanski et al : Growth, population age structure, and aspects of the reproductive biology of Epinephelus niveatus 215 We feel confident that our assessment of the age structure in the snowy grouper population off the Carolinas was accurate, even though interpretation of growth increments was difficult and a minimal number of YOY specimens was available. The dif- ficulty of assigning an age to the sagittae of snowy grouper had not been reported by other investiga- tors, although it has been reported for other deepwa- ter species of continental slopes. The clarity of the hyaline and opaque zones in otoliths (presumably sagittae) from hoki, Macruroniis novaezelandidae, off New Zealand is highly variable and is divided into six categories based on internal features which are related to the ease of counting increments (Kuo and Tanaka, 1984). When otoliths are difficult to interpret, one option is to base population age struc- ture only on the specimens for which age is easily assessed. Alternatively, ages can be estimated for nearly all specimens despite the difficulties, as we did in our study, with the assumption that the larger sample will represent the population better. We found that limiting the data set to only those specimens for which the difference in counts between readers was 0-1 increments did not improve the accuracy of the age-length key for specimens caught with long- lines (Table 6). Thus, we advocate using the entire sample of specimens assigned an age. Crabtree and Bullock ( 1998) found that rejected otoliths tend to be from slower-growing older specimens, which could introduce bias into analyses. This bias was minimal in their study of growth in black grouper, M. bonaci, where each otolith was examined three times by two independent readers. Parameter estimates for the von Bertalanffy model based on all black grouper with ages were within one standard error of those based only on specimens for which the coefficient of varia- tion of the six readings was <12%. An important consideration in age determination is positive identification of the first annulus and any settling mark that may be deposited prior it. We believe that we have identified the first annulus because the largest YOY specimen (172 mm) had an estimated age of 191-291 days and the measure- ments of otolith radius for all YOY in -3) were less than radial measurements to the first annulus in a subsample of 23 specimens that were age 1. Evi- dence to support our conclusion that these three specimens were YOY was found in another sam- pling effort, where six specimens 4-5 cm in length were caught with a trawl during August and Sep- tember (Machowski'^), the last two months of the ^ Machowski, D. 1998. S. Carolina Department of Natural Resources, P.O. Box 125.59. Charleston, SC, 29422. Personal commun. spawning season. Moore and Labisky (1984) consid- ered 150-175 mm specimens to be YOY, although they did not examine daily increments. Validation of a technique for aging snowy grouper with otoliths has been weakly supported by pre- vious studies. We found that marginal increments form annually and there is a peak in April and May that corresponds to the beginning of the spawning season. This finding concurs with the limited results of Matheson and Huntsman (1984) and Moore and Labisky (1984) who found that increment formation appeared to begin in April and peaked in June. Mathe- son and Huntsman (1984) measured marginal incre- ments in 18 specimens collected during April through October and Moore and Labisky (1984) examined specimens collected during March through July in not reported). Waltz"* found a wider period of increment formation, April through September, although he was not able to conclude that increments form annually because samples were lacking for four months. Reproduction The reproductive pattern of snowy grouper needs to be investigated more comprehensively and the sex ratio should be assessed again, given the small sample size in 1993-94 («=82), because there is evi- dence that reef fish species, particularly grouper, which change sex and aggregate to spawn, are more susceptible to size-selective mortality and overex- ploitation (Bannerot, 1987; Huntsman and Schaaf, 1994; Coleman et al., 1996). The capture of only one male in the 1993-94 samples, which appeared to be representative of the population based on commer- cial landings, is reason for concern because the per- centage of males has apparently decreased from the 7-23*^ for samples collected with three gear types in the 1970s and 1980s (Table 7). Large decreases in the number of males have been documented for two other grouper species in the southeast region. Percentages of males in popula- tions of gag and scamp in the Gulf of Mexico, grou- pers that are known to form small spawning (lO's to lOO's of individuals) aggregations, decreased from 17% to 19c and 38% to 18%, respectively, between the 1970s and early 1990s (Coleman et al., 1996). A similar decrease, from 20% to 6%, was noted for gag along the Atlantic coast of the southeastern United States during the same period (McGovern et al., 1998). The resultant decrease in genetic diversity has been documented for gag (Chapman etal., 1999), and its ramifications are currently of great concern to many fishery scientists in the southeast region. The size (767-1090 mm) and age (8-29 yr) of 97 male specimens in the present study and the capture 216 Fishery Bulletin 98(1) of two specimens in the process of changing from female to male is conclusive evidence that snowy grouper are protogynous hermaphrodites. Moore and Labisky ( 1984 ) reported males as young as age 6 and some males with evidence of recent sex change. It is likely that we collected only two transitional speci- mens because sex change occurs after a female fin- ishes spawning, during months when sample sizes in our study were small. Sex change in other grouper species, gag for example, occurs primarily during the first two to three months after the spawning season (McGovern et al., 1998), before males and females become spatially separated (Coleman et al., 1996). Our findings on age at maturity and spawning season are in general agreement with the results of Moore and Labisky ( 1984) for snowy grouper in the Florida Keys. They found that the smallest mature female and largest immature female were age 3 and age 5, respectively, whereas the smallest mature female from 1980 to 1985 in our study was also age 3, but small percentages (<10'7f ) of the age 7-9 females were immature (Table 9). Females and males in the Florida Keys were in spawning condition from April through July, although no mature fish were sam- pled in other months (Moore and Labisky, 1984). We found that females off the Carolinas spawn from April through September — possibly longer owing to small sample sizes in October through March. The capture of 1160 specimens, some of which were assessed macroscopically as spawning, in four trawl collections on the exploratory squid cruise in June 1978 suggests that snowy grouper may form spawning aggregations. Estimates of density would have been much higher than 13.5-79.5 kg/ha calculated from the trawl data if the fish were caught in only a small part of the area sampled during tows of 7.4-18.5 km. Dodrill and Epperly (1995) reported that the initial density of exploitable snowy grouper on a 2700-m-^ virgin reef off North Carolina was 11 kg/m^. Depth distribution Fishery-independent data collected over several years and with various gear types show that fish length is positively correlated with water depth (Fig. 11). Longlines, Kali poles, and snapper reels were the pri- mary gear types deployed in waters >150 m and the waters <100 m were sampled primarily with chev- ron traps. Chevron traps are not known to be selec- tive for snowy grouper <600 mm. Dodrill et al. ( 1993 ) speculated that the low abundance of adults in shal- low waters may in part reflect years of intensive fish- ing pressure in 40-120 m depths and only relative recent fishing activity at depths >183 m off North Carolina. Alternatively, we propose that snowy grou- per may migrate to deeper water toward the end of the juvenile stage. Fish length and water depth data from the com- mercial fisheries ( Figs. 13 and 14 ) concurred with the positive correlation noted in the Florida Keys (Moore and Labisky, 1984) and off Georgia and South Car- olina (Low and Ulrich, 1983), although depth data from fishermen may be less accurate than fishery- independent data. We found small juveniles in shal- low water, a finding that agrees with the results of Moore and Labisky ( 1984) and with observations from submersible dives that documented juvenile snowy grouper (<300 mm) between 46 and 91m, but not in deeper waters (Parker^). Accurate assessment of population parameters re- quires knowledge of juvenile and adult distributions of snowy gi'ouper as well as characteristics of the fish- ery. The snapper reel fishery catches a greater pro- portion of younger age classes than does the longline fishery because fishing efforts are generally restricted to areas <100 m in depth (Figs. 13 and 14). The long- line fishery presently catches a greater proportion of older age classes than does the snapper reel fish- ery because regulations established by the SAFMC restrict longlines to waters deeper than 91 m. Conclusions As other investigators have suggested, rebuilding grouper populations may require a novel approach such as long-teiTH area closures or individual trans- ferable quotas (Epperly and Dodrill, 1995; Coleman et al., 1996). At present, the regulations enacted to rebuild the snowy grouper population include an annual quota of 245,082 kg, with a trip limit of 1134 kg (SAFMC, 1993). Traditional management mea- sures such as minimum size limits will not be effective because snowy grouper experience fatal embolisms while being brought to the surface from deep waters (Matheson and Huntsman, 1984). Future research should focus on improving our understanding of repro- ductive pattern (e.g. spawning behavior, spatial and temporal aspects of distribution) and include a thor- ough assessment of sex ratio and an updated assess- ment of population age structure. Acknowledgments We thank F. Rhode, J. Francesconi, and other assisting members of the North Carolina Department of Health, '^Parker, R., Jr. 1997. National Marine Fisheries Service, 101 Pivers Island Rd., Beaufort, NC, 28516. Personal commun. Wyanski et al : Growth, population age structure, and aspects of the reproductive biology of Epmephelus niveatus 217 Environment and Natural Resources for obtaining otoliths and gonads from snowy gi-ouper in North Carolina. In South Carolina, the personnel of the MARMAP program of the S.C. Department of Natu- ral Resources assisted with port sampling. Captains D. Juel, S. Juel, J. Mun'ay, S. Shelley, and the late J. D. Skipper III let us remove otoliths from their catches of snowy grouper and brought in whole speci- mens for the study of reproductive biology. We very much regret the loss of Captain Skipper, his vessel, and a crew member. C. Jackson and T. Prince of the Southport Fish Market, and F. McGinn and R. McGinn of the Little River Fish House, allowed us to process specimens at their businesses. R. Dixon and G. Huntsman provided gonad samples collected by the National Marine Fisheries Service Beaufort Labo- oratory. K. Grimball and O. Pashuk prepared the his- tological sections. B. Zhao generated the parameter estimates for the von Bertalanffy model. T. Azarov- itz, T. Reisinger, and D. Machowski contributed to the retrieval of data from the 1978 squid cnaise origi- nally collected by E. Gutherz. Earlier drafts of the manuscript were examined by P. Harris, J. McGov- em, and three anonymous reviewei's. This project was funded through the National Marine Fisheries Service MARFIN grant NA37FF0046-01 and the MARMAP contract 50WCNF006002. Literature cited Bannerol, S., W. W. Fox Jr., and J. E. Powers. 1987. Reproductive strategies and the management of snap- pers and groupers in the Gulf of Mexico and Caribbean. In J. J. Polovina and S. Ralston (eds.). Tropical snappers and groupers: biolog>' and fisheries management, p. 561-603. Westview Press, Boulder, CO. Bennett, B. M., and P. Hsu. 1960. On the power function of the exact test for the 2x2 contingency table. Biometrika 47:393-397, Chapman, R. W., G. R. Sedberry, C. C. Koenig, and B. M. Eleby. 1999. Stock identification of gag, Myctcroperca microlepis, along the southeast coast of the United States. Mar. Bio- technol. 1:137-146. Coleman, F. C, C. C. Koenig, and L. A. Collins. 1996. Reproductive styles of shallow-water groupers ( Pisces: Serranidae) in the eastern Gulf of Mexico and the conse- quences of fishing spawning aggregations. Environ. Biol. Fishes 47:129-141. Collins, M. R. 1990. A comparison of three fish trap designs. Fish. Res. 9:32,5-332, Crabtree, R. E., and L. H. Bullock. 1998. Age, growth, and reproduction of black grouper, Mycte- roperca bonaci. in Florida waters. Fish. Bull. 96:735-753. Dodrill, J., C. S. Manooch III, and A. B. Manooch. 1993. Food and feeding behavior of adult snowy grouper, Epinephelus niveatus (Valenciennes! (Pisces: Serranidae I, collected off the central North Carolina coast with ecologi- cal notes on major food groups. Brimleyana 19:101-135. Epperly, S. P, and J. W. Dodrill. 1995. Catch rates of snowy grouper, Epinephelus niveatus, on the deep reefs of Onslow Bay, southeastern U.S.A. Bull. Mar. Sci. 56:450-461. Fitch, J. E., and S. A. Schultz. 1978. Some rare and unusual occurrences of fishes off Cali- fornia and Baja California. Calif Fish Game 64:74-92. Harris, P. J., and J. C. McGovern. 1997. Changes in the life history of the red porgy, Pagrus pagrus, from the southeastern United States, 1972-1994. Fish. Bull. 95:732-747. Haug, T., and J. Tjemsland. 1986. Changes in size- and age-distributions and age at sexual maturity in Atlantic halibut. Hippoglossus hippo- glossus, caught in north Norwegian waters. Fish. Res. 4:145-155. Hayes, D. B. 1993. A statistical method for evaluating differences between age-length keys with application to Georges Bank haddock, Melanngrammus aeglefinus. Fish. Bull. 91:550-557. Helser, T. E., and F. P. Almeida. 1997. Density-dependent growth and sexual maturity of silver hake in the north-west Atlantic. J. Fish Biol. 51:607-623. Hunter, J. R., and S. R. Goldberg. 1980. Spawning incidence and batch fecundity in northern anchovy, Engraulis mordax. Fish. Bull. 77:641-652. Hunter, J. R., and B. J. Macewicz. 1985. Rates of atresia in the ovary of captive and wild north- ern anchovy, Engraulis mordax. Fish. Bull. 83:119-16. Huntsman, G. R. 1976. Offshore headboat fishing in North Carolina and South Carolina. Mar Fish. Rev 38(3): 13-23. Huntsman, G. R., and W. E. Schaaf. 1994. Simulation of the impact of fishing on reproduction of a protogynous grouper, the graysby. North Am. J. Fish. Manage. 14:41-52. Johnson, A. G., and L. A. Collins. 1994. Age-size structure of red grouper, Epinephelus morio, from the eastern Gulf of Mexico, Northeast Gulf Sci. 13:101-106. Johnson, A. G., L. A. Collins, and J. J. Isley. 1993. Age-size structure of gag, Mycteroperca microlepis, from the northeastern Gulf of Mexico. Northeast Gulf Sci. 13:.59-63. Kuo, C. L., and S. Tanaka. 1984. Otolith features and reliability for age-determina- tion of hoki, Macruronus novaezelandidae (Hector), in waters around New Zealand. Bull. Jpn, See. Sci. Fish. 50:1349-1355. Lee, T. N., V. Kourafalou, J. D. Wang. W. J. Ho, J. O. Blanton, L. P. Atkinson, and L. J. Pietrafesa. 1985. Shelf circulation from Cape Canaveral to Cape Fear during winter. In L. P. Atkinson, D. W. Menzel, and K. A. Bush (eds.). Oceanography of the southeastern U.S. con- tinental shelf p. 33-62. American Geophysical Union, Washington, D.C. Low, R. A. 1998. South Carolina marine fisheries, 1996. South Caro- lina Mar Resour. Center Data Rep. 27, 78 p. Low, R. A., and G. F. Ulrich. 1983. Deep-water demersal finfish resources and fisheries off South Carolina. South Carolina Mar. Resour. Center Tech. Rep. 57, 24 p. 218 Fishery Bulletin 98(1 ) Low, R. A., D. Theiling, and E. B. Joseph. 1987. South Carolina marine fisheries. South Carolina Mar. Resour. Center Tech. Rep. 67, 78 p. Matheson, R. H., Ill, and G. R. Huntsman. 1984. Growth, mortahty, and yield-per-recruit models for speckled hind and snowy grouper from the United States South Atlantic Bight. Trans. Am. Fish. Soc. 113:607-616. McGovern, J. C., D. M. Wyanski, O. Pashuk, C. S. Manooch, III, and G. R. Sedberry. 1998. Changes in the sex ratio and size at maturity of gag, Mycteroperca microlepis, from the Atlantic coast of the southeastern United States during 1976-1995. Fish. Bull. 96:797-807. Miller, D. J., and R. N. Lea. 1976. Guide to the coastal marine fishes of California. Calif Dep. Fish Game Fish Bull. 157, 235 p. Moore, C. M., and R. F. Labisky. 1984. Population parameters of a relatively unexploited stock of snowy grouper in the lower Florida Keys. Trans. Am. Fish. Soc. 113:322-329. Parker, R. O., Jr., and R. W. Mays. 1998. Southeastern United States deepwater reef fish assemblages, habitat characteristics, catches, and life his- tory summaries. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 138, 41 p. Parker, R. O., Jr., and S. W. Ross. 1986. Observing reef fishes from submersibles off North Carolina. Northeast Gulf Sci. 8:3 1-49. Ross, J. L. 1982. Feeding habits of the gray tilefish, Caulolatilus microps (Goode and Bean, 1878) from North Carolina and South Carolina waters. Bull. Mar. Sci. 32:448-454. Rothschild, B. J. 1986. Dynamics of marine fish populations. Harvard Univ. Press, Cambridge, MA, 277 p. Russell, G. M., E. J. Gutherz, and C. A. Barans. 1988. Evaluation of demersal longline gear off South Caro- lina and Puerto Rico with emphasis on deepwater reef fish stocks. Mar Fish. Rev 50(1):26-31. SAFMC (South Atlantic Fishery Management Council). 1991. Amendment 4, regulatory impact review, initial regu- latory flexibility analysis and environmental assessment for the snapper grouper fishery of the South Atlantic region, 177 p. SAFMC, Charleston, "SC. 1993. Amendment 6, regulatory impact review, initial regu- latory flexibility analysis and environmental assessment for the snapper grouper fishery of the south Atlantic region, 202 p. SAFMC, Charleston, SC. SAS Institute, Inc. 1990. SAS/STAT® user's guide, version 6, part 2, 4th ed., 1686 p. SAS Institute, Cary. NC. Schmidt, D. J., M. R. Collins, and D. M. Wyanski. 1993. Age, growth, maturity, and spawning of Spanish mackerel, Scomheromorus maculatus (Mitchill), from the Atlantic coast of the southeastern United States. Fish. Bull. 91:526-533. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstructural examination. In D. K. Stevenson and S. E. Campana (eds.). Otolith microstructure examination and analysis, p. 19-57. Can. J. Fish, and Aquat. Sci. Special Publ. 117. Siegel, S. 1956. Nonparametric statistics for the behavioral sciences. McGraw-Hill, New York City. NY, 312 p. Smith, C. L. 1971. ArevisionoftheAmerican groupers: Bp/nep/ie/w.s' and allied genera. Bull. Am. Mus. Nat. Hist. 146:228-241. von Bertalanffy, L. 1938. A quantitative theory of organic growth. II. Inquiries on growth laws. Human Biol. 10:181-213. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in tele- osts. Am. Zool. 21:32.5-343. Wenner, C. A., W. A. Roumillat, and C. W. Waltz. 1986. Contributions to the life history of black sea bass, Centroprtstis striata, off the southeastern United States. Fish. Bull. 84:723-741. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshwater Res. 41:199-222. Zhao, B., J. C. McGovern, and P. J. Harris. 1997. Age, growth, and temporal change in size at age of the vermilion snapper from the South Atlantic Bight. Trans. Am. Fish. Soc. 126:181-193. 219 Notes on the biology of Cepha/urus cephalus and Parmaturus xaniurus (Chondrichthyes: Scyliorhinidae) from the west coast of Baja California Sur, Mexico Eduardo F. Balart Division de Biologia Manna, Centra de Investigaciones Biologicas del Noroeste, S,C. Apdo Postal 128, La Paz, B C S , Mexico 23000 E-mail address ebalarKa'cibnor mx Jeanette Gonzalez-Garcfa Carlos Villavicencio-Garayzar Laboratono de Elasmobranquios, Universidad Autonoma de Baia California Sur Apdo Postal 19-B, La Paz, B C S, Mexico 23000 were dissected, the diameter of all visible oocytes were recorded, and embryos were removed. For the sub- sequent analysis of C cephalus, pre- viously published data were also used. The length-weight relation- ship was determined with STATIS- TICA software (StatSoft, Inc., 1995) by using the exponential function of the form W^aTL'', as used by Cross ( 1988) for the file- tail catshark. Length-weight rela- tionships were compared between males and females by using an analysis of covariance. The head, or lollipop, shark (Cephal- urus cephalus) and filetail catshark (Parmaturus xaniurus) are found in the eastern Pacific in waters rang- ing from 245 to 828 m and from 91 to 1251 m depth, respectively (Castro, 1983). Little is known of their biology because they are rarely captured. Cephalurus cephalus is a benthic catshark and has been recorded from southern Califor- nia, Gulf of California, the Revil- lagigedo Archipelago, and off the coasts of Peru and Chile (Kato et al., 1967; Melendez and Meneses, 1989; Pequeho, 1989). Mathews and Ruiz (1974) reported it from the upper Gulf of California, Mexico, and Castro- Aguirre (1981) gave basic information about its morphology, taxonomy, and ecology based on 11 specimens from this locale. Parmaturus xaniurus is distrib- uted from central California to the Gulf of Cahfornia (Castro, 1983). Cross ( 1988) has described its gen- eral biology and Lee ( 1969 ) reported that, in the Santa Barbara basin, it consumed myctophids. Cailliet' concluded that juveniles were meso- pelagic, adults demersal, and noted that efforts to determine their age have failed because of the absence of annual rings on the vertebrae. The objective of this study was to provide additional data on the biol- ogy of C cephalus and P. xaniurus, with information on length-weight relationships, and notes on repro- ductive biology, including the sizes of oocytes and embryos, the sex ratio of embryos, and clasper length, for specimens from the west coast of Baja California Sur, Mexico. Materials and methods Fish were sampled by bottom trawl- ing with commercial shrimp nets (20-m mouth, 30-mm mesh size) during the oceanographic expedi- tion EP9505 along the Pacific coast of Baja California Sur, Mexico. Trawls were made at depths of 50-280 m from the RV El Puma during May 1995. On 5 May two scyliorhinid species (C. cephalus and P. xaniurus) were collected and whole specimens were frozen. All specimens were captured at the following three stations: 26°01.9' N, 113 26.9W (280 m); 25'59.6'N, 113 20.5 W (260 m); 26 00. 7N, 113° 18.3'W(230m). Bottom temperature at the collection sites was 10.0°C. Total weight (W, g), and total length (TL, mm) were recorded. Females Results and discussion Seventy-five catsharks were ob- tained: 51 P. xaniurus and 24 C. cephalus. The latter species is recorded for the first time along the west coast of the Baja Cali- fornia peninsula. At one station, C. cephalus was captured together with P. xaniurus. as reported by Mathews and Ruiz (1974). Other organisms collected at the three sta- tions were the fishes Merluccius angustimanus , Physiculus rastrel- liger, Kathetostoma averruncus, Sy- nodus lucioceps, Hippoglossina sto- mata, and Lepophidium spp., and the crustaceans Pleuroncodes pla- nipes and Cancer johngarthii. Cephalurus cephalus Of the 24 specimens, 23 were fe- male, with a range of 224 and 295 mm TL (.v=249 mm TL). The single male specimen was 298 mm TL. Previously reported maximum total ' Cailliet, G. 1981. Ontogenetic changes in the depth distribution and feeding habits of two deep-dwelhng demersal fishes of Cahfornia: sablefish and filetail cat sharks. Paper presented at the sixty- first annual meeting of the Am. Soc. Ich- tyol. Herpetol. Manuscript accepted 17 August 1999. Fish. Bull. 98:219-221 (2000). 220 Fishery Bulletin 98(1) lengths for females and males were 295 and 245 mm TL respectively (Castro-Aguirre, 1981). The length- weight relationship suggested allometric growth and was described by the following equation W = 0.000012 rL2 8-* (Fig. lA) The b coefficient was not significantly different (i^j 33= 2.15) between the sexes. Females of this species have two functional ovaries and oviducts. Sixteen (70Vf ) of the females had mature oocytes in their ovaries and a mean oocyte diameter of 10.5 mm (range 2.0-21.8 mm). The relationship of mean oocyte diameter with TL indicated that all females were mature (Fig. IB). The size at first matu- •a 140 110 80 50 20 -10 W = 0000012Tl r^ = 0.99 n =40 50 100 150 200 Total length (mm) 250 300 40 80 J. I, 60 I s ■3 ~ u ?. 20 ^ -e 220 90 •^ 30 B a ^Ok Oocyte • " Embryo f OD = -O047 2 + 044 Tl a a 8 « r^= 62 n=21 / ^-i-8- a _a- / ___Q OoO -0-^ . ^ 9 230 240 250 260 Female total length (mm) 270 J '0 U -10 50 100 150 200 250 Total length (mm) 300 Figure 1 Cephalurus cephalus. (A) Length-weight relationship. (Bl Relationships between oocyte diameter and embr>o total length with female total lengths. (Ci Relationship between total length and clasper length. Data less than 75 mm TL are from embryonic material. Data from Castro-Aguirre ( 1981 ) are denoted by solid circles. rity has previously been estimated to be 190 mm TL (Compagno, 1984). The large diameter of the oocytes suggested that spawning occurs in early summer. Embryonic development of C. cephalus occurs in the uterine tract and is described as aplacental vivi- parity or ovoviviparity (Wourms, 1977, 1981; Com- pagno, 1984). The thin-walled egg cases of C. cephalus, which are retained in the oviducts, were observed in 11 individuals. The three largest females had neither eggs in the oviducts nor large oocytes in their ovaries and may have finished their reproductive cycle. The mean size of the 19 embryos was 43 mm TL ( size range 21 to 65 mm TL), which is less than the esti- mated size at birth (100 mm; Compagno, 1984), and the sex ratio of embryos (9 males, 10 females) was aproximately equal. The relation between clasper length and total length of adults and embryos is illustrated in Figure IC. Parmaturus xaniurus Of 51 specimens captured, 22 were female and 29 were male. Females measured 108 to 350 mm TL (.v=184 mm TL), males 117 to 380 mm TL (.r=233 mm TL). The largest recorded lengths are 574 mm TL for females and 600 mm TL for males (Springer, 1979). The length-weight relationship of P. xaniurus indi- cated allometric growth (Fig. 2A) and was described by the following equation: W= 5.5 X 10-^ TL3 34. The analysis by sex (males: W= 6.76 x 10"^ TL^-^i; females: W=9.53 x 10'" TL^-) did not reveal sig- nificant differences in the b coefficient (Fj 45=0.14) as reported by Cross (1988) for specimens from the upper continental slope off southern California. All females examined were immature and did not contain mature oocytes. Cross (1988) noted that females first reached maturity at 425-475 mm TL in southern California waters. Parmaturus xaniurus is oviparous, and the egg cases of this species have been described by Cox (1963). Embryonic develop- ment lasts approximately one year (Compagno, 1984; Cross, 1988). The relationship between total length and clasper length suggested that males matured at 340 mm TL (Fig. 2B), and this size at first maturity was smaller than that previously recorded (375 to 425 mm TL; Cross, 1988). This may reflect latitudinal differences in the reproductive biology of P xaniurus. Acknowledgments We thank Edgar Amador, Carolina Downton, Dinorah Herrero, M. Angeles Cobarrubias, and Carmifia A. NOTE Balart et a\ Notes on the biology of Cephalurus cephalus and Parmaturus xaniurus 221 250 200 — 150 ■fi 100 ^ 50 f A W = 00000055 Ti 3_M Q / r ' = 98 n = 49 c o/ ^^^ ^^^"^ • Male O Female . . 50 100 150 200 250 300 350 400 Total length (mm) g 60 45 30 B Y=0 00175EXP(X"') r^=0 90 o / n = 29 °9 1 ^_^^^^^^^ ^ 50 100 150 200 250 300 Total length (mm) 350 400 Figure 2 Parmaturus xaniurus. (Al Length-weight relationship. (B) Relationship between total length and clasper length. Fernandez, and the crew of the RV El Puma who were always helpflil in our work. We gratefully acknowledge the valuable comments of Gregor Cailliet, Robert N. Lea, Jim Ellis, and two anonymous referees. We thank Ellis Glazier for editing the text in English. We also thank the Universidad Nacional Autonoma de Mexico, especially Ingvar Emilsson, for supporting the projects of CIBNOR, UABCS, and CICIMAR and providing the RV El Puma for the cruise EP9505 at no cost. Literature cited Castro, J. I. 1983. The sharks of North american waters. University Press, Austin, TX, 180 p. Texas, A&M. Castro-A^irre, J. L. 1981. Especies de la Familia Scyliorhinidae (Elasmo- branchii, Galeoidea), de la Costa Occidental de Mexico, con especial referenda a Cephalurus cephalus (Gilbert). An. Esc. Nac. Cienc. Biol. Mex. 24:71-93. Compagno, L. J. V. 1984. FAO species catalogue. Sharks of the world: an anno- tated and illustrated catalogue of shark species known to date. Part 2: Carcharhiniformes. FAO Fish. Biol. Synop. (125), vol. 4, pt2: 2.51-6.55. Cox, K. W. 1963. Egg-cases of some elasmobranchs and a cyclostome from California waters. Calif Fish and Game 49(4):271- 289. Cross, J. N. 1988. Aspects of the biology of two scyliorhinid sharks, Apristurus hrunneus and Parmaturus xaniurus. from the upper continental slope off southern California. Fish. Bull. 86(41:691-702. Kato, S., S. Springer, and M. H. Wagner. 1967. Field guide to Eastern Pacific and Hawaiian sharks. U.S. Fish and Wildl. Serv., Circ. 271:1-47. Lee, R. S. 1969. The filetail catshark, Parmaturus xaniurus, in mid- water in the Santa Barbara basin off California. Calif. Fish and Game 55( 1 ):88-90. Mathews, C. P., and M. F. Ruiz. 1974. Cephalurus cephalus. a small shark, taken in the northern Gulf of California, with a description. Copeia 1974(2): 556-560. Melendez, C. R., and D. R. Meneses. 1989. Tiburones del talud continental recolectados entre Anca ( 18 T9'S) e Isla Mocha (38''30'S ), Chile. Invest. Mar.. Valparaiso, 17:3-73. Pequeiio, R. G. 1989. Feces de Chile. Lista sistematica revisada y comen- tada. Rev. Biol. Mar. 24(2): 1-132. Springer, S. 1979. A revision of the catsharks, family Scyliorhinidae. U.S. Dep. Commer., NOAA Tech. Rep. NMFS Circ. 422, 152 p. StatSoft, Inc. 1995. STATISTICAfor Windows. StatSoft, Inc., Tulsa, OK. 3782 p. Wourms, J. P. 1977. Reproduction and development m chondrichthyan fishes. Am. Zool. 17:379-410. 1981. Viviparity: the maternal-fetal relation in fishes. Am. Zool. 2:473-515. 222 Penaeid shrimp landings in the upper Gulf of California in relation to Colorado River freshwater discharge Manuel S. Galindo-Bect Institute de Investigaciones Oceanologicas Universidad Autonoma de Baia California Km 103 carretera Ti|uana-Ensenada Ensenada, Baia California, Mexico Edward P. Glenn Environmental Research! Laboratory 2601 East Airport Drive Tucson, Anzona 85706 E-mail address (for E- P Glenn, contact author) eglennidag anzona edu Henry M. Page' Kevin Fitzsimmons^ Luis A. Galindo-Bect^ Jose M. Hernandez-Ayon^ Robert L. Petty' Jaqueline Garcia-Hernandez^ David Moore^ A commercial trawl fishery in the upper Gulf of California provides the principle source of income for the coastal communities of the region, but catches of estuarine- dependent crustaceans and fish have declined in recent years (Her- nan, 1997; Cudney-Bueno and Turk- Boyer, 1998). Declines in shrimp landings, mainly Litopenaeus styl- irostris (formerly classified as Pen- aeus stylirostris ) i Perez-Farfante and Kinsley, 1997 ) have been attrib- uted primarily to over-exploitation of the resource and to viral diseases (Rosas-Cota et al., 1996; Hernan, 1997). The Biosphere Reserve of the upper Gulf of California and Col- orado River Delta was created in 1993 to address some fisheries man- agement problems. A more funda- mental problem, however, may be the lack of river flow after construction of upstream dams. Historic reduct tions in river discharge have caused dramatic increases in salinity in the estuary and changes in the distri- bution of nutrients ( Alvarez-Borrego et al., 1975; Hemandez-Ayon et al, 1993). Since 1979, occasional flood releases have entered the upper Gulf of California by means of the Colo- rado River when upstream impound- ments are filled (Glenn et al., 1996). Effects of freshwater on penaeid shrimp population development are controversial (Garcia and Le Reste, 1981; Day et al., 1989), but recruit- ment of spawning stocks of white shrimp (Penaeus setiferus ) has been positively correlated with river dis- charge in the southwestern Gulf of Mexico and has been attributed to an expansion in estuarine nursery habitat for white shrimp (Garcia, 1991). River discharge also can stimulate the migration of sub- adults from estuaries (Deben et al., 1990; Vance et al., 1998). Fish- ermen have a strong perception that shrimp and fish catches in the northern Gulf of California are related to freshwater discharge from the Colorado River (Cudney- Bueno and Turk-Boyer, 1998). To evaluate their perception we con- ducted a correlation analysis of shrimp landings at San Felipe Baja California (nearest shrimping sta- tion to the delta) with freshwater discharges from the Colorado River to the northern Gulf of California. Materials and methods Data on annual shrimp landings and number of trawlers legally fishing from San Felipe were obtained ft-om the Secretary of Environment, Nat- ural Resources and Fish (SEMAR- NAP), San Felipe, Mexico. Landings were available from 1977 and number of trawlers from 1982. The artisanal catches by small boats (pangas) or the significant illegal shrimp fishery are not accounted for in reported shrimp landings. Annual shrimp landings serve as indicators of the variability in the total landings and are reported for all species of shrimp, even though landings are >90% L. styl- irostris in San Felipe (Rosas-Cota et al., 1996). Data on freshwater discharge of the Colorado River were from the Southerly Inter- national Border (S.I.B.) gauging station which is below the last diversion on the river and were obtained from the United States ' Mainne Science Institute University of California Santa Barbara. California 93106 - Environmental Research Laboratory 2601 Ea.st Airport Drive Tucson, Arizona 85706 ' Institute de Investigaciones Oceanologicas Universidad Autonoma de Baja California Km 103 carretera Tijuana-Ensenada Ensenada, Baja California, Mexico Manuscript accepted 2.5 August 1999. Fi.sh. Bull. 98:222-22.5 (20001. NOTE Galindo Beet et al ; Penaeid shrimp landings in the upper Gulf of California 223 Bureau of Reclamation, Yuma, Arizona (Wil- liamsM. We assumed that this flow entered the delta and the upper Gulf of California. Annual shrimp landings and landings divided by numbers of trawlers { normal catch per unit of effort, CPUE) were correlated with river flow and number of trawlers. Our normal CPUE was a crude approximation of stock abundance or catchability. We lacked actual fishing time (days, weeks, hours of net deployment), size frequency of the legal ves- sels, and number of small boats ( pangas ) fish- ing. We made landings lag river discharge by one year because the life cycle of shrimp from hatching to capture is approximately one year (Gracia-Pamanes^). Transformed river flow (logjij) was tested for nonlinearity; then we conducted a multiple regression analysis to predict shrimp landings from variables that were individually correlated (P<0.05) with landings. Results Annual shrimp landings ranged from 701 metric tons (t) ( 1983-84) to 217 t ( 1992-93), decreasing significantly from 1977 to 1996 (r=0.78, P<0.001, Fig. lA). The reported number of trawlers legally fishing from San Felipe ranged from a high of 59 in 1988 to a low of 20 in 1995 (Fig. IB). Catch per unit of effort (CPUE) increased from 1982 to 1984, then markedly decreased back to the 1982-83 level in 1985, remaining low until 1993, after which a positive trend was achieved and the highest CPUE ever was recorded in 1995 (Fig. IC). There were substantial flows (>700 million cubic meters, Mm-^) in 8 of the 21 years from 1976 tol996 and varied over 10'*-fold, ranging from 1 Mm^ in 1990 and 1996 to 15,657 Mm'^ in 1984 (Fig. ID). High- est volume occurred between 1980 and 1987 as a result of overflow from Lake Powell in the United States (Glenn et al., 1996). The flow spike in 1993 was due to releases from Painted Rock Dam on the Gila River in Arizona. Periods of significant river flow at the S.I.B. were closely z a! 2 b l'^ 1975 1980 1985 1990 1995 2000 80 60 40 20 B -^ 1975 1980 1985 1990 1995 2000 20,000 15,000 - 10,000 - 5,000 1975 1980 1985 1990 1995 2000 Figure 1 (A) Annual San Felipe shrimp landings (metric tons); (B) Number of shrimp trawlers fishing annually in San Felipe; (C) CPUE (metric tons/boat); (D) Colorado River freshwater discharge flow below the Southerly International Border (million m-Vyrl. 1 Williams, B. 1998. United States Bureau of Reclamation. Yuma, Arizona. Personal commun. ' Garcia-Pamanes, F. C. 1992. Biologia reproductiva y din- amica poblacional del camaron azul Penaeus stylirostris en el Alto Golfo de California. Instituto de Investigaciones Oceano- logicas, Universidad Autonoma de Baja California, En.senada. Unpubl. final report. matched to El Nino Southern Oscillation (ENSO) events that occurred in 1983 and 1993. Shrimp landings were significantly (P<0.05) cor- related with same year river discharge, but logj^- transformed river discharge in the year prior to shrimp harvest produced the highest correlation coefficient (r=0.67, P<0.001) (Table 1). The number of trawlers also significantly correlated with shrimp landings (r=0.77, P<0.001), as expected. The best correlation (r) of shrimp landings was the product 224 Fishery Bulletin 98(1) Table 1 Correlation coefficients (r) and significance levels (P) from regression analysis relating San Felipe annual shrinr ( 1977-96) and CPUE ( 1982-96) to rainfall and discharge of the Colorado River at the Southern International Border A indicates that shrimp landings were paired with the previous year's river discharge in the correlation analysis. p landings '1-year lag" Correlation with shrimp landings Correlation with CPUE Variable r P r P River discharge 0.47 0.0362 0.25 0.3360 Logn, river discharge 0.54 0.0112 0.25 0.3368 River discharge ( 1-yr lag) 0.52 0.0127 0.34 0.1826 Log[Q of river discharge ( 1-yr lag) 0.67 0.0006 0.38 0.1304 Number of shrimp trawlers 0.77 0.0003 0.18 0.4804 Log,,, of river discharge (1-yr lag) number of shrimp trawlers 0.80 0.0004 0.29 0.8771 of logjQ-lagged river discharge and number of trawl- ers (/•=0.80, P<0.001). CPUE was not significantly (P>0.05 ) correlated with river flow or number of trawl- ers ( Table 1 ), nor with total landings ( r=-0.26, P=0.3 1 ). The equation of best fit (0.64) for predicting shrimp landings took the form where X,= Y = M = Y=a +m{X,X2), log,Q-lagged river discharge (Mm'^/yr); number of trawlers; shrimp landings (t/yr); the slope of the equation (1.67); and the y-intercept (232 t/yr). Discussion Our analyses represent a first attempt to identify re- lationships between variability in shrimp landings in the upper Gulf of California and factors influencing these landings. Total shrimp landings and the size of the shrimping fleet at San Felipe have declined over the past 15 years. Social and economic changes have affected shrimping. In the late 70s and early 80s shrimping was reserved for social units (cooper- atives), with the result that privately owned shrimp trawlers were banned from the fishery. In addition, the government subsidized building of additional vessels and many new unskilled fishermen entered the industry. Then policies were reversed in the late 1980s, private boats returned, interest rates increased, and many of the shrimp trawlers were removed from the fleet. We found a significant relationship (P<0.001) be- tween total catch and the rate of freshwater discharge of Colorado River water into the marine ecosystem, although the mechanisms through which river dis- charge might affect the shrimp fishery are unknown. Lower salinity may improve the survival of early life stages by providing "enlarged nursery" protected habitat (Garcia, 1991), even though L. stylirostris and P. californiensis are generally considered eury- haline species ( Hernan, 1997 ), having large numbers of postlarvae and juveniles in hypersaline habitats (Brusca, 1980; Page'^). Salinity and nutrient gradi- ents in the estuary and upper Gulf during river flows have not been reported to our knowledge. Future plans for the Colorado River will likely decrease freshwater discharge into the estuary as more water is diverted upstream for farms and domestic use (Morrison et al., 1996). Our analyses suggest that decreases in river discharge to the delta and estuary may adversely affect shrimp landings. The United States and Mexican governments should initiate a research program on the effects of river flow on ecologically and commercially important spe- cies in the upper Gulf of California and incorporate these findings into a comprehensive management plan for the Biosphere Resei-ve as well as the Colo- rado River Basin at large. Literature cited Alvarez-Borrego, S., B. P. Flores-Baez, and L. A. Galindo-Bect. 1975. Hidrologia del Alto Golfo de California II. Condiciones durante invierno, primavera y verano. Ciencias Marinas '2(1 ):2 1-36. Brusca, R. C. 1980. Common intertidal invertebrates of the Gulf of Cali- fornia. Univ. Arizona Press, Tucson, AZ, 513 p. ' Page, M. 1999. Marine Science Institute, University of Cali- fornia, Santa Barbara, CA 93106. Unpubl. data. NOTE Gallndo-Bect et a\ Penaeid shrimp landings in the upper Gulf of California 225 Cudney-Bueno, R., and P. J. Turk-Boyer. 1998. Pescando entre mareas del Alto Golfo de California. Centro Intercultural de Estudios de Desiertos 7 Oceanos, Puerto Penasco, Sonora. Mexico. 166 p. Day, J. W., Jr., C. A. S. Hall, W. M. Kemp, and A. Yanez-Aranchibia. 1989. Estuanne ecology. Wiley. New York, I^TV'. .'i.S8 p. Deben, W., W. Clotheir, G. Ditsworth and D. Baumgartner. 1990. Spatiotemporal fiucuation.-i in the distribution and abundance of demersal fish and epibenthic crustaceans in Yaquina Bay. Oregon. Estuaries 13(4):469-478. Garcia, A. 1991. Spawning stock recruitment relationships of white shrimp in the southwestern Gulf of Mexico. Trans. Am. Fisheries Soc. 120i4):519-527. Garcia, S., and L. Le Reste. 1981. Life cycles, dynamics, exploitation and management of coastal penaeid shrimp stocks. FAO Fish. Tech. Paper 203, 215 p. Glenn, E. P., Ch. Lee, R. Felger, and S. Zengel. 1996. Effects of water management on the wetlands of the Colorado River Delta, Mexico. Conserv. Biol. 10(41:1175- 1186. Heman, A. 1997. Morphologic and genetic characterization of wild popu- ulations of shrimp of the genus Penaeus within the Gulf of California, Mexico: new social, political and management dilemmas for the Mexican shrimp fishery. Univ. of Ari- zona, Tucson, AZ, 323 p. Hernandcz-Ayno, J. M., M.S. Galindo-Bect, B.P. Flores-Baez, and S. Alvarez-Borrego. 1993. Nutrient concentrations are high in the turbid waters of the Colorado River Delta. Estuarine Coastal Shelf Sci., 37:593-602. Morrison, J., S. Postel, and P. Gleick. 1996. The sustainable use of water in the Lower Colorado River Basin. Pacific Institute for Studies in Development, Environment, and Security, Oakland, California. Perez-Farfante, L, and B. Kinsley. 1997. Panaeoid and sergestoid shrimps and prawns of the world: keys and diagnoses for the families and genera. Memoires du Museum National D'Histoire Naturelle 175:1- 233. Rosas-Cota, J. A., V. M. Garcia-Tirado, and J. R. Gonzalez-Camacho. 1996. Analisis de la pesqueria del camaron de altamar en San Felipe, B.C., durante la temporada de pesca 1995-1996. Secretaria de Medio Ambiente Recursos Naturales y Pesca, Institute Nacional de la Pesca, Centro Regional de Inves- tigacion Pesquera de Ensenada, Boletin 2, 23-30 p. Vance, D. J., M. Haywood, D. Heales, R. Kenyon, and N. Loneragan. 1998. Seasonal and annual variation in abundance of post- larval and juvenile banana prawns Penaeus merguiensis and environmental variation in two estuaries in tropical northeastern Australia: a six year study. Mar. Ecol. Prog- ress Ser. 163:21-36. 226 Fishery Bulletin 98(1), 2000 Superintendent of Documents Publications Order Form *5178 Dyes, please send me the following publications: Subscriptions to Fishery Bulletin for $50.00 per year ($62.50 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) (Please type or print) Charge your order. IT'S EASY! VISA Please Choose Method of Payment: I I Check Payable to the Superintendent of Documents GPO Deposit Account 1 - VISA or MasterCard Account 1 (Credit card expiration date) To fax your orders (202) 512-2250 (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! Fishery Bulletin Guide for Contributors Content Articles published in Fishery Bulletin de- scribe original research in fishery marine science, engineering and economics, and the environmental and ecological sciences, including modeling. Articles may range from relatively short to extensive. Notes are reports of 5 to 10 pages without an abstract and describing methods or re- sults not supported by a large body of data. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not upon the editor or the publisher. It is there- fore important that the contents of the manuscript are carefully considered by the authors. Submission of an article is understood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts should be written in English. Authors whose native language is not En- glish are strongly advised to have their manuscripts checked by English-speaking colleagues prior to submission. Preparation Title page should include authors' full names and mailing addresses, the corre- sponding author's telephone, FAX number, and e-mail address, and a list of key words to describe the contents of the manuscript. Abstract should not exceed one double- spaced typed page. It should state the main scope of the research but emphasize its conclusions and relevant findings. Because abstracts are circulated by abstracting agencies, it is important that they repre- sent the research clearly and concisely. Text must be typed double-spaced through- out. Abrief introduction should portray the broad significance of the paper; the re- mainder of the paper should be divided into the following sections: Materials and methods, Results, Discussion (or Con- clusions), and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e. there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be spelled out the first time they are mentioned. The scientific names of species must be written out the first time they are mentioned; subsequent mention of scientific names may be abbreviated. Follow Scientific Style and Format: the CBE Manual for Authors, Editors, and Publish- ers (6th ed. ), for editorial style, and the most current issue of the American Fisheries Society's Common and Scientific Names of Fishes from the United States and Canada, for fish nomenclature. Dates should be writ- ten as follows: 11 November 1991. Measure- ments should be expressed in metric units, e.g. metric tons as t; if other units of mea- surement are used, please make this fact ex- plicit to the reader. The numeral one ( 1 ) should be typed as a one, not as a lower-case el (1). Use of appendices is discouraged. Text footnotes should be numbered with Arabic numerals and inserted at the bot- tom of text. Footnote all personal commu- nications, unpublished data, and unpub- lished manuscripts with full address of the communicator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid refer- ences to nonstandard (gray) literature, such as internal, project, processed, or ad- ministrative reports. Where these refer- ences are used, please include whether they are available from NTIS (National Technical Information Service) or from some other public depository. Literature cited comprises published works and those accepted for publication in peer reviewed literature (in press). Follow the name and year system for citation format. In the text, cite as follows: Smith and Jones (1977) or (Smith and Jones, 1977). If there is a sequence of citations, list alphabeti- cally by authori s ): Smith, 1987; Smith and Jones, 1985; Smith and Peterson, 1986. Ab- breviations of serials should conform to ab- breviations given in Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and complete- ness of all citations. Tables should not be excessive in size and must be cited in numerical order in the text. Headings within tables should be short but ample enough to allow the table to be intelligible on its own. All unusual sym- bols must be explained in the table legend. Other incidental comments may be foot- noted with italic numerals. Use the aster- isk only to indicate probability in statistical data. Zeros should precede all decimal points for values less than one. Figures must be cited in numerical order in the text. The senior author's last name and the figure number should be written on the back of each one. Hand-drawn il- lustrations should be submitted as origi- nals and not as photocopies. Submit photographs as glossy prints or slides with good contrast, otherwise we cannot guar- antee a good final printed copy. Graphs should be submitted as laser-printed cop- ies, not as photocopies. Label all figures with Helvetica typeface and capitalize the first letter of the first word in axis labels. Italicize species name and variables in equations. Use zeros before all decimal points. Use uppercase Times Roman bold typeface to label the parts of a figure, e.g. A, B, C, etc. Each figure legend should ex- plain all symbols and abbreviations in the figure and should be double-spaced and placed at the end of the manuscript. Copyright law Government publications are not copyrighted; they fall within the public domain. If an author reproduces any part of a government publication in his work, reference to source is appreciated. Submission Send printed copies (original and three copies without staples) to the Scientific Editor: Dr. John V. Merriner Southeast Fisheries Science Center National Marine Fisheries Service 101 Pivers Island Road Beaufort, NC 28516 Once the manuscript has been accepted for publication, you will be asked to submit an electronic copy of your manuscript to the Managing Editor. The electronic copy should be submitted in WordPerfect or Microsoft Word text format and placed on a 3.5-inch disk that is double-sided, double or high density, and that is compatible with either DOS or Macintosh systems. A copy of page proofs will be sent to the au- thor for final approval prior to publication. Reprints (copies of published articles and notes) are available free of charge to the senior author (50 copies) and to his or her laboratory (50 copies). Additional copies may be purchased in lots of 100 when the author receives page proofs. U.S. Department of Commerce Volume 98 Number 2 April 2000 U.S. Department of Commerce William M. Daley Secretary National Oceanic and Atmospheric Administration D. James Baker Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Penelope D. Dalton Assistant Administrator for Fisheries Scientific Editor Dr John V. Merriner Editorial Assistant Sarah Shoffler Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 101 Pivers Island Road Beaufort, NC 28516 ^ATES OV The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service. NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA98115-0076. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office. Washington, DC 20402. Subscrip- tion price per year: $.50.00 domestic and $62.50 foreign. Cost per single issue: $19.00 domestic and $23.76 foreign. See back for order form. Managing Editor Sharyn Matriotti National Manne Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Dr. Andrew E. Dizon National Marine Fisheries Service Dr. Harlyn O. Halvorson University of Massachusetts, Boston Dr. Ronald W. Hardy University of Idaho, Hagerman Dr. Richard D. Methot National Marine Fisheries Service Dr. Theodore W. Pietsch University of Washington, Seattle Dr. Joseph E. Powers National Manne FIshenes Service Dr. Harald Rosenthal Universitat Kiel, Germany Dr. Fredric M. Serchuk National Marine Fisheries Service Fishery Bulletin web site: fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents. U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 98 Number 2 April 2000 Fishery Bulletin Contents Tln' LuiuluMuii^ and iipimons expressed in Ff,s/it'r\' Bullet III are solely those of the authors and do not represent the official position of the National Mai'ine Fisher- ies Service (NOAAi or an> other agency or institution. The NalionaJ Marine Fisheries Service iNMFS' does not approve, recommend, or endorse any proprietary product or pro- prietary material mentioned in this pub- lication, No reference shall be made to NMFS, or to this publication furnished by NMFS. in any advertising or sales pro- motion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or pro- prietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. Articles 227-235 Acha, Eduardo M., and Gustavo J. Macchi Spawning of Brazilian menhaden, Brevoortia aurea, in the Rio de la Plata estuary off Argentina and Uruguay 236-249 Brewster-Geisz, Karyl K., and Thomas J. Miller Management of the sandbar shark, Carcharhinus plumbeus: implications of a stage-based model 250-263 Keen Alonso, Mariano, Enrique A. Crespo, Susana N. Pedraza, Nestor A. Garcia, and Mariano A. Coscarella Food habits of the South American sea lion, Otaria flavescens, off Patagonia, Argentina 264-282 Lo, Nancy C. H, John R. Hunter, and James H. Churnside Modeling statistical performance of an airborne lidar survey system for anchovy 283-289 Macchi, Gustavo J., and Eduardo M. Acha Spawning frequency and batch fecundity of Brazilian menhaden, Brevoortia aurea, in the Rio de la Plata estuary off Argentina and Uruguay 290-298 Manickchand-Heileman, Sherry C, and Dawn A. T. Phillip Age and growth of the yellowedge grouper, Epinephelus flavolimbatus, and the yellowmouth grouper, Mycteroperca interstitialis, off Trinidad and Tobago 299-318 Mollet, Henry F., Geremy Cliff, Harold L. Pratt Jr., and John D. Stevens Reproductive biology of the female shortfin mako, Isurus oxyrindtus Rafinesque, 1810, with comments on the embryonic development of lamnoids 319-335 Mortensen, Donald, Alex Wertheimer, Sidney Taylor, and Joyce Landingham The relation between early marine growth of pink salmon, Oncorhynchus gorbuscha, and marine temperature, secondary production, and sui"vival to adulthood Fishery Bulletin 98(2). 2000 336-344 Orbacz, Elizabeth A., and Patrick M. Gaffney Genetic structure of tautog (Tautoga onltis) populations assayed by RFLP and DGGE analysis of mitochondrial and nuclear genes 345-352 Rivera, Glauco A., and Richard S. Appeldoorn Age and growth of dolphinfish, Coryphaena hippurus, off Puerto Rico ^- 353-363 Rocha-Olivares, AxayacatI, H. Geoffrey Moser, and Jason Stannard Molecular identification and description of pelagic young of the rockfishes Sebastes constellatus and Sebastes ensifer 364-374 Schaefer, Kurt M., and Charles W. Oliver Shape, volume, and resonance frequency of the swimbladder of yellowfin tuna, Thunnus albacares 375-388 Sevigny, Jean-Marie, Patrice Gagne, Yves de Lafontaine, and Julian Dodson Identification and distribution of larvae of redfish (Sebastes fasciatus and S. mentella: Scorpaenidae) In the Gulf of St. Lawrence 389-399 Stevens, Bardley G., Ivan Vining, Susie Byersdorfer, and William Donaldson Ghost fishing by Tanner crab (Chionoecetes bairdi) pots off Kodiak, Alaska: pot density and catch per trap as determined from sidescan sonar and pot recovery data 400-409 Stokesbury, Kevin D. E., Jay Kirsch, Evelyn D. Brown, Gary L. Thomas, and Brenda L. Norcross Spatial distributions of Pacific herring, Clupea pallasi, and walleye pollock, Theragra chalcogramma, in Prince William Sound, Alaska 410-420 Veistad, Jon H., Alexei F. Sharov, Glenn Davis, and Brenda Davis A method for estimating dredge catching efficiency for blue crabs, Callinectes sapldus, in Chesapeake Bay Notes 421-426 Ahrenholz, Dean W., Deborah D. Squires, James A. Rice, Stephen W. Nixon, and Gary R. Fitzhugh Periodicity of increment formation in otoliths of overwintering postlarval and prejuvenile Atlantic menhaden, Brevoortia tyrannus 427-438 Powell, Allyn B., David G. Lindquist, and Jonathan A. Hare Larval and pelagic juvenile fishes collected with three types of gear in Gulf Stream and shelf waters in Onslow Bay, North Carolina, and comments on ichthyoplankton distribution and hydrography 439 Subscription form 227 Abstract.— Brazilian menhaden, Bre- voortia aurea, is the only species of the genus Brevoortia in South American Atlantic waters and is abundant in the Rio de la Plata estuar\'. We found that B. aurea in this area spawns almost ex- clusively in this estuary. We studied the temporal and spatial reproductive pat- tern of this menhaden and related the pattern to the major hydrographic fea- tures of the region. We based evidence of spawning activity on the presence of females with hydrated oocytes and on the occurrence of menhaden eggs in plankton samples. Our results show that B. aurea spawn during virtually every month of the year, but that they spawn mainly from September (late winter) to January (early summer). In the Rio de la Plata estuary, spawned eggs occur in a thermohaline range of 13-23"C and 10-25 psu, mainly in strat- ified waters. Breroortia aurea .spawn very near the bottom salinity front, probably in a convergent flow between the riverine and estuarine waters that helps to retain eggs. In contrast to men- haden of the northern hemisphere (B. tyrannus and B. patronus ), which spawn offshore and which drift during early life history stages, Brevoortia aurea in the Rio de la Plata estuary are spawned and held in estuarine waters near spawning sites. The latter reproductive pattern is also shared by Micropogonias furnieri (whitemouth croaker), the most abun- dant fish species in the area. Spawning of Brazilian menhaden, Brevoortia aurea, in the Rio de la Plata estuary off Argentina and Uruguay* Eduardo M. Acha Dto Ciencias Mannas Universidad Nacional de Mar del Plata Funes 3350 (7600) Mar del Plata, Argentina Present address; Institute Nacional de Investigacion y Desarrollo Pesquero (INIDEP) Paseo V Ocampo N° 1 (7600) Mar del Plata, Argentina E-mail address macha a inidep-eduar Gustavo J. Macchi Conseio Nacional de Investigaciones Cientificas y Tecnicas (CONICET) Rivadavia 1917 (1033) Buenos Aires, Argentina Manuscript accepted 29 November 1999. Fish. Bull. 98:227-235 (2000). The Rio de la Plata drains the second largest basin of South America. It flows into the Atlantic Ocean with an average discharge of 22,000 m^/s, generating a large estuary of about 35,000 km- and 5-15 m in depth, located at 36°S, 56°W (Framinan and Brown, 1996). Brazilian menha- den, Brevoortia aurea , is abundant in this estuary ( Cousseau, 1985; Boschi, 1988); it also inhabits coastal and estuarine environments from 13^S (Brazil) to 40°S (Argentina). Histori- cally, two species of menhaden were thought to inhabit Brazilian-Argen- tine waters (de Ciechomski. 1968; Weiss et al., 1976; Weiss and Krugg, 1977; Whitehead, 1985; Lasta and de Ciechomski, 1988); however, Cous- seau and Diaz de Astarloa (1993) concluded that B. aurea is the only species that inhabits South Ameri- can Atlantic waters. Little is known about the reproduc- tive biology of S. aurea. De Ciechom- ski (1968) described eggs and early lai"val stages of fi. aurea and reported a period of 86—88 hours from the time of spawning to hatching at 13-14'C. Spawning has been detected in the Rio de la Plata estuary (Lasta and de Ciechomski, 1988). Although their planktonic eggs also have been found in inshore waters along the Uruguayan and Argentine coasts (de Ciechomski, 1968; Hubold and Ehr- lich, 1981; Cassia and Booman, 1985; Sanchez and de Ciechomski, 1995), Samborombon Bay (a shallow area inside the estuary) seems to be the locus of intensive spawning, where B. aurea eggs are exposed to low salinities (5—15 psu) (Lasta and de Ciechomski, 1988). Estuarine spavin- ing by species producing pelagic eggs is uncommon ( Dando, 1984; Haedrich, 1992; Potter et al., 1993 ); however no attempts have been made to describe the basic spawning habitat require- ments of S. aurea. Furthermore, in the Rio de la Plata, Micropogonias furnieri, the species with the largest biomass, releases pelagic eggs in the inner zone of the estuary (Macchi et al., 1996; Acha etal., 1999). The life cycles of Northern Hemi- sphere menhaden, mainly Brevo- ortia tyrannus and Brevoortia pat- Contribution 1130 of the Instituto Nacional de Investigacion y Desarrollo Pesquero (INIDEP). (7600) Mar del Plata, Argentina. 228 Fishery Bulletin 98(2) ronus (the Atlantic and gulf menhaden, respectively) are well understood. Both species are typical repre- sentatives of estuarine dependent species that spawn in the marine environment (Lawler et al., 1988; Day et al., 1989) in contrast to the Brazilian menhaden, which spawns in an estuarine environment. Since 1983 the coastal resources of the Argen- tine—Uruguayan Common Fishing Zone have been monitored by a number of cruises, and results per- taining to menhaden form the basis for this paper. Our objectives were to describe the timing and spa- tial occurrence of spawning in relation to the major hydrographic features of the region in order to gain insight into the spawning habitat requirements of the Brazilian menhaden. Whenever possible, comparative analyses with other species of the estuary and men- haden of the Northern Hemisphere were performed. Materials and methods Samples from 47 cruises from 1983 through 1998 were analyzed in our study. Twenty-two of these cruises were on research vessels of the National Institute for Fisheries Research and Development (INIDEP), covering the Rio de la Plata estuary and adjacent coastal waters, throughout which stations were randomly distributed. Twenty-five cruises took place in Samborombon Bay, on small fishing boats using a systematic sampling design. Monthly distri- bution of the sampling effort is shown in the insert of Figure 1. During 1983 and 1987, nineteen cruises in Samborombon Bay were performed every 30-45 days during the entire year. The remaining six cruises in this bay correspond only to spring months (October, November, and December). Cruises were made with several objectives, hence covering different periods, but sampling effort was higher in spring (October- November) when cruises for stock assessment were performed. The high number of plankton samples in May was due mainly to one cruise, designed to study physical variables and plankton in the estuary. Cruises on the small boats and on the research ves- sels were not simultaneous. All data were employed as a composite representing mean conditions. Plankton was collected at 980 sampling stations by oblique tows of 60-cm bongo nets, 20-cm bongo nets, or a Nackthai sampler ( a modified Gulf V high-speed plankton net, see Nellen and Hempel, 1969). All nets were equipped with flowmeters. The volume filtered in each tow ranged from 10 to 400 m''. Sampling depth was estimated from measurements taken with an angle indicator (inclinometer) and a wheelout meter. All samples were preserved in 5*^ buffered formalin. Plankton samples were sorted and ana- lyzed in the laboratory. Brevoortia aurea eggs were identified following the description of de Ciechomski (1968). Estimates of egg density from the different sam- plers were not intercalibrated. They were arranged into four broad density classes (<10, 10 to 99, 100 to 999, >1000 eggs/m'^), and marked on a map to delin- eate the geographic location of the spawning area. Monthly distribution of the average number of eggs per tow (catch per unit of effort, CPUE) and the per- centage of positive stations for menhaden eggs were plotted to identify the spawning season. Stations with no catch were included in the CPUE estimates. Data from the November 1995 cruise were useful for gaining insight into the vertical distribution of menhaden eggs. During that cruise, plankton was sampled at five stations in a small area near Mon- tevideo, where schools of spawning B. aurea were previously detected. In each of those stations, two tows were conducted with the Nackthai net, one to sample the surface and the other to sample the bottom layers as defined by a strong halocline. The depth of the halocline was previously measured with a conductivity-temperature-depth (CTD) profiler. On all cruises, menhaden were collected with bot- tom trawls, and the presence of females with hydrated oocytes (macroscopic examination) was considered evidence of imminent spawning. The percentage of females with hydrated oocytes for each tow was also indicated on the map to identify the location of the spawning area. In addition, the mean monthly per- centage of females with hydrated oocytes was used to determine the spawning season. Temperature and salinity information was taken from the oceanographic database created by Guer- rero et al. (1997a), which included all the CTD sta- tions sampled during the cruises that we performed. Bottom salinity and temperature data for spring- summer were used because they could be matched to the vertical distribution of the eggs and the main reproductive period. Salinity is expressed as psu (practical salinity units; Anonymous, 1981). Hori- zontal and vertical salinity contour lines were made to compare the spawning area with the salinity field. Egg density was plotted into a temperature-salinity diagram to show the environmental ranges of the spawning habitat of the Brazilian menhaden. Results Eggs in the plankton Brazilian menhaden eggs were detected virtually all year round (Fig. 1). The highest CPUE occurred from Acha and Macchi: Spawning of Brevoortia aurea in the Rio del la Plata estuary off Argentina and Uruguay 229 1000 ■^ 40- Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Figure 1 The spawning season of the Brazihan menhaden in the Ri'o de la Plata estuary. Line and dots show monthly distribution of the mean number of menhaden eggs per tow (CPUE). Black bars show the percentage of positive plankton stations for menhaden eggs. Gray bars show the females with hydrated oocytes as a percentage of the total adult females caught each month. Insert; the gray bars show the monthly distribution of plankton (above) and adults (below) samples. Numbers below the bottom axis correspond to cruises on small boats (SB) and research vessels (RV) each month. September through January and diminished during late summer, followed by a secondary peak during the fall (April-June). The percentage of positive stations for menhaden eggs had a similar pattern to that of the CPUE. There was no plankton sampling in July. Menhaden eggs were identified in 220 samples that represented 22. 5C^ of all samples analyzed. The eggs were found mainly in estuarine waters (Fig. 2A), especially at depths <10 m. Highest densities of eggs were found in Samborombon Bay and west of Montevideo in the innermost part of the estuary, where salinity values ranged between 10 and 20 psu. Medium egg densities were found along the Uruguayan coast between Montevideo and Atlan- tida (55'45'W) and in the middle area of the river between Montevideo and Punta Piedras. Lowest den- sities (<10 eggs/m'^) were present in the rest of the estuary and adjacent coastal waters. The study area (5 stations) of the vertical distri- bution of B. aurea eggs is shown by an arrow in Figure 2A. Total depth in this area was 6.5-7.5 m and halocline depth was 4.8-6.5 m. Menhaden eggs were present in the water layer below the halocline but were extremely scarce in the upper layer. In the bottom layer, egg density ranged from 50-2100 eggs/m'. Eggs were found within a thermohaline range of 9.7-27.3 psu and 18.5-20.2°C. The upper layer was less saline (0.7-10.8 psu) and warmer (19.7-21.7-C). Strong haloclines up to 21.5 psu/m were observed. Figure 3 shows the vertical distri- bution of salinity and B. aurea eggs along a tran- sect in the estuary. This transect includes four of the five plankton stations sampled above and below the halocline. Menhaden eggs were detected in the region where the salt wedge intersects the bottom (the bottom salinity front). In this area of stratified 230 Fishery Bulletin 98(2) waters, depth of bottom layer (measured from the halochne to the bottom) was less than 3 m, and men- haden eggs were located below the halocline. Menhaden eggs occurred in a wide environmental range (Fig. 4). Densities of > 100 eggs/m^ were found 34 S 35 - 36 3 37 34 S 33 - 36 37 53 W 75-100% Figure 2 Location of the spawning area of the Brazilian menhaden, (A) The size of the symbols is proportional to the egg density from the plankton samples; the arrow shows the study site for examining the vertical distribution of eggs (see the text). (B) The size of the .symbols is proportional to the per- centage of females with hydrated oocytes. In both maps, the isohalines (expressed as psui represent the bottom salinity field for spring-summer. PE = Punta del Este; A = Atlantida; M = Montevideo; PP = Punta Piedras; SB = Samborombon Bay; SAC = San Antonio Cape. in a salinity range of 10 to 25 psu; minor densities were located in waters with salinities reaching 33 psu (continental shelf waters). The low salinity boundary for egg distribution seems more abrupt than the high salinity boundary; very few eggs were detected below 10 psu. Most eggs produced over the prolonged spawning season were found at temperatures be- tween 13 and 23°C. Gravid females A total of 375 sampling stations were analyzed, in which 1084 gravid females were identified. All the gravid females were caught from September through December. In these months, the total percentage of females with hydrated oocytes ranged from 72.8% to 89.4'7r (Fig. 1). Gravid females were detected across the river between Mon- tevideo and Punta Piedras, and in Samborombon Bay (Fig. 2B). In that region, percentages of gravid females in each tow were >50%. The highest percentages of females with hydrated oocytes appeared to be associated with the highest horizontal salinity gradient (the salinity front). No females with hydrated oocytes were detected in continental shelf waters. Discussion The distribution of eggs and gravid females reveals that B. aurea spawn in the estuary (Fig. 2). Gravid females were found in a more restricted area than that where eggs were found, thus providing a more accurate picture of the spawning spatial pattern, because currents may move eggs to more distant areas. An echo sounder showed Brazilian menhaden below the halocline on all cruises, includ- ing the occasions when spawn- ing individuals were collected. The halocline was detected by South America* J \/ ( M Study H" Area 53W Acha and Macchi: Spawning of Brevoortia aurea in the Rio del la Plata estuary off Argentina and Uruguay 231 ID Salinity (psu) J) 10 20 10 20 10 20 10 20 ,, 766n/rrf I 29 n/ni J Q 40 60 Distance (km) Figure 3 Vertical distribution of Brazilian menhaden eggs. (A) Vertical distribution of eggs at the head of the salt wedge, where the upper and bottom layers were sampled separately. Numbers in the white areas show eggs/m' above the halocline, numbers in the gray areas show egg concentrations below the halocline. Black lines show salin- ity profiles. (B) Location of the transect in the estuary. (C) Salinity section in the estuary. Isohalines are each 2.5 psu; vertical dots indicate CTD observations. Numbers above the top axis show eggs/m' in oblique plankton tows from bottom to surface. the echo sounder as a scattering layer, owing to the concentration of zooplankton ( Madirolas et al. , 1997 ). Brazilian menhaden appears to spawn close to the bottom, at the salinity front. The mean position of this front is strongly related to bottom topography (Guerrero et al., 1997a), having a relatively fixed location. This spawning area extends into the strati- fied waters of the estuary between Montevideo and Punta Piedras, and along Samborombon Bay. How- ever, because waters of the inner zone of the bay are vertically homogeneous, or weakly stratified (Guer- rero et al., 1997a), some high egg concentrations oc- cur in nonstratified waters. Menhaden eggs occurred in a wide range of salini- ties and temperatures. Because the bottom front is a zone of major salinity changes, the eggs are exposed to a wide range of salinity values, and because the reproductive season extends over a long period, eggs are exposed further to a wide thermal range (Fig. 4). Low egg densities at low temperatures and high salinities represent samples taken in continental shelf waters, which were cooler than the estuarine waters during the warm season (Guerrero et al., 1997a, 1997b). Presence of Brazilian menhaden eggs in the plank- ton year-round (Fig. 1) is strong evidence that they have a protracted reproductive season. Major spawn- ing activity, based on incidence of eggs and hydrated ovaries (gravid females), occurs from September to January. Other reports of menhaden eggs, based on 232 Fishery Bulletin 98(2) partial temporal or spatial coverage, or both, fall within the temporal limits that we detected (de Ciechomski, 1968; Hubold and Ehrlich, 1981; Cassia and Booman, 1985; Lasta and de Ciechomski, 1988). Brevoortia aurea eggs remain in the bottom layer of the water column, near where the halocline intersects the bottom (Fig. 3). At this frontal interface, a conver- gent flow near the bottom (Largier, 1993) would serve to retain eggs near the con- fluence of river and marine waters, mini- mizing their drift. Thus, specific gravity of the eggs seems to be an important feature of the reproductive strategy of B. aurea, allowing the eggs to stay in the saltier ( and denser) bottom waters. Typically, bottom waters move landward in salt-wedge estu- aries (Kjerfve, 1989; Mann and Lazier, 1991 ). Lower egg concentrations in the rest of the estuary probably indicate some dis- persion of eggs or reduced spawning in these areas. Disruptive events such as storms, which destroy the halocline (Guerrero et al., 1997a), intermittently occur in this region. These meteorological events are character- ized by strong winds over 13 m/s from the southeast (Balay, 1961). The events have a characteristic duration of 1-3 (occasion- ally 5) days, occur throughout the year, and are usu- ally the strongest from May to October (Anonymous, 1995). During these events, the estuarine circulation is modified, thus altering the egg retention properties of the system as well. The protracted reproductive season and multiple spawning of B. aurea (Macchi and Acha, 2000) may help to ensure that enough eggs survive in this unpredictable environment. The four menhaden species in the Northern Hemi- sphere are the small-scaled menhaden B. smithi and B. gunteri and the large-scaled menhaden B. tyran- nus and B. patronus (Ahrenholz, 1991). Thei-e is little information on the biology of 6. smithi and B. gunteri. Conversely, B. tyraruius and B. patronus have been intensively studied, and the information on their life histories has been broadly reviewed (Ahrenholz, 1991; Powell, 1994). Both species have protracted reproduc- tive seasons and a main spawning period in winter (Nelson et al, 1977; Shaw et al., 1985; Powell, 1994). Spawning of these menhaden takes place in marine waters. Larvae swim and drift with tides and currents toward estuaries where metamorphosis from larvae to juveniles takes place. In the case of B. tyrannus. larvae must be transported from the intensive spawn- ing area south of Cape Hatteras, up to 100 km to L ^ D D 22 5, 20 °to°° : - Z 18 3 E ^ 16 ^ -^^ p? ° a 14 eggsW ' a °" ° 1000 1°° - . - 1 5 10 15 20 25 30 Salinity (psu) Figure 4 Thermohaline range of Brazilian menhaden eggs determined from plank- ton samples taken from the Rio de la Plata estuary and adjacent coastal waters. The size of the symbols is proportional to egg density. estuarine nursery areas (Warlen, 1992; Powell, 1994). Brevoortia patronus seems to spawn relatively close to estuarine nursery areas (Shaw et al., 1985). Brevoortia aurea, B. tyrannus. and B. patronus all have protracted spawning periods. The southern spe- cies is a spring-summer spawner; the northern ones spawn during cooler months. In the case of 6. tyran- nus. however, adults search for temperatures >17°C to spawn, and as larvae are transported nearer the coast, they enter cooler water (Warlen, 1992). Notwithstand- ing, the reproductive thermal range of the northern species (12.9-21.2 C for B. patronus, and >17€ for B. tyrannus (Warien, 1988; 1992) lies close to that of B. aurea. The proposed advantage of winter spawn- ing refers to maximum onshore transport during that season, providing a mechanism for transporting larvae of both species into the vicinity of estuaries (Nelson et al, 1977; Warlen, 1988). The main differences between B. aurea and those northern menhaden seem to be the population bio- mass and the spawning habitat. Brevoortia tyrannus and B. patronus exhibit large population sizes and are a significant component of United States fishery landings (Vaughan, 1991; Powell, 1994). Although there is no biomass assessment for B. aurea, it has Acha and Macchi: Spawning of Brevoortia aurea in the Rio del la Plata estuary off Argentina and Uruguay 233 little importance as a fishery species (Argentine landings in 1997 were 893 t (Anonymous, 1998). Life history characteristics of Brevoortia aurea are probably more similar to those of B. smithi and B. gunteri, which tend to be more nearshore and estu- arine-oriented species, form loose aggregations in coastal waters, and have protracted spawning peri- ods during fall-winter months (Ahrenholz, 1991). Not enough information, however, on their reproduc- tive biology exists to make further comparisons. Drift of early life history stages is a major feature in the life cycle of 5. patronus, and especially of B. tyrannus. On the contrary, egg retention seems to be a main property of the life cycle of the Brazilian menhaden in the Rio de la Plata estuary. However, in southern Brazil, B. aurea does not seem to be an estuarine spawner ( reported as B. pectinata , Weiss et al., 1976; Weiss and Ki-ug, 1977; Weiss, 1981; Sinque and Muelbert, 1997 ). The greatest number of eggs occurred in nearshore waters, in a saline range of 33.04-35.50 psu. Several eggs were also found in the access channel to Lagoa dos Patos (a large coastal lagoon in southern Brazil, 32''S) during high salinity events, although its larvae and juveniles were dis- tributed throughout the estuary in low salinity con- ditions (Weiss et al., 1976; Weiss, 1981; Sinque and Muelbert, 1997). The spawning area of M. furnieri, the most abundant fish species in the Rio de la Plata, partially overlaps that of S. aurea, and the eggs of M. furnieri are retained in the estuary below the halo- cline (Acha et al., 1999). Like B. aurea, M. furnieri is an estuarine spawner in the Rio de la Plata (Macchi et al., 1996) but seems to be a saltwater spawner in southern Brazil (Sinque and Muelbert, 1997). Both species have the potential for estuarine reproduc- tion, but the eventual spawning site depends on the dynamic properties of the environment. Major reviews of the role of estuaries for fishes state that spawning of pelagic eggs within estuar- ies is an unusual episode (e.g. Day et al., 1981; Day et al.,1989; Haedrich, 1992). Owing to the net sea- ward movement of estuarine waters, export of early life-history stages from the estuaries seems to be a major problem for estuarine spawners ( Boehlert and Mundy, 1988). In the Rio de la Plata estuary, at least two fish species (S. aurea and M. furnieri) spawn pelagic eggs, taking advantage of the retention properties at the head of the salt wedge. These retention proper- ties are not a feature unique to the Rio de la Plata because estuaries with salt wedges are well known features (e.g. Mann and Lazier, 1991; Officer, 1992); however spawning of pelagic eggs at those locations is an uncommon event. Retention of eggs and larvae in the convergence zone may not be complete and the net nontidal flows (residual currents) may drive them seaward. Given the large size of the Rio de la Plata estuary, distance from the spawning area to the offshore limit of the salt wedge may reach 250 km (Guerrero et al., 1997a, 1997b). Therefore, larvae probably have enough time to develop to the stage where they become able to control their vertical posi- tion in the water column. The migratory behavior of larvae is manifested mostly in the vertical rather than horizontal plane (Norcross and Shaw, 1984). Thus, larval menhaden could take advantage of the two-lay- ered circulation pattern and through vertical migra- tions compensate for the horizontal transport. This movement would be an important adaptation to main- tain population (as the reproductive unit) coherency (Boucher, 1988; Sinclair and lies, 1989). Literature cited Acha, E. M.. H. W. Mianzan, C. A. Lasta, and R. A. Guerrero. 1999. Estuarine spawning of the whitemouth croaker Micro- pogonias furnieri in the Rio de la Plata, Argentina. Mar. Freshwater Res. 50( 1 1:57-65. Anonymous. 1981. Background papers and supporting data on the Prac- tical Salinity Scale 1978. Tec. Pap. Mar. Sci. 37, UNESCO, Paris, 144 p. 1995. Derrotero Argentine. Parte 1 Rio de la Plata, 10th ed. Servicio de Hidrografia Naval, Armada Argentina, Buenos Aires. 273 p. 1998. Capturas maritimas totales 1997. Ministerio de Eco- nomia y Obras y Servicios Publicos. Secretaria de Agri- cultura, Ganaderia, Pesca y Alimentacion. Subsecretaria de Pesca. Direccion Nacional de Pesca Acuicultura, Buenos Aires, 45 p. Ahrenholz, D. W. 1991. Population biology and life histor>' of the North Amer- ican menhadens, Brevoortia spp. Mar. Fish. Rev. 53(4): 3-19. Balay, M. A. 1961. El Rio de la Plata entre la atmosfera y el mar. Servi- cio de Hidrografia Naval, Argentina, Publico H-621, 153 p. Boehlert, G. W., and B. C. Mundy. 1988. Roles of behavioral and physical factors in larval and juveniles fish recruitment to estuarine nursery areas. Am. Fish. Soc. Sym. 3:51-67. Boschi, E. E. 1988. El ecosistema estuanal del Rio de la Plata. An. Inst. Cienc. del Mar y Limnol., Univ. Nal. Auton. Mexico 15i2): 159-182. Boucher, J. 1988. Space-time aspects in the dynamics of planktonic stages. In Toward a theory of biological-physical interac- tions in the world ccean: proceedings of the NATO advanced workshop, Casteria Verduzan, France, 15 June 1987 (B. J. Rothschild, ed.l, p. 203-214. Kluwer Academic Press, Dordrecht. Cassia, M. C, and C. I. Booman. 1985. Distribucion del ictioplancton en el Mar Argentine en los afios 1981-1982. Physis 43(105):91-111. 234 Fishery Bulletin 98(2) Cousseau, M. B. 1985. Los peces del Rio de la Plata y su Frente Maritimo. In Fish community ecology in estuaries and coastal lagoons: towards an ecosystem integration ( A.Yafiez-Arancibia, ed. ), p. 515-534. DR(R) UNAM Press, Mexico. Cousseau, M. B., and J. M. Diaz de Astarloa. 1993. El genero Brevoortia en la costa Atlantica Sudameri- cana. Frente Maritimo 14:49-57. Dando, P. R. 1984. Reproduction in estuarine fish. In Fish reproduc- tion: strategies and tactics (G. W. Potts and R. J. Wootton, eds.l p. 155-170. Academic Press, London. Day, J. H., S. J. M. Blaber, and J. H. Wallace. 1981. Estuarine fishes. In Estuarine ecology with par- ticular reference to southern Africa (J. H. Day, ed.), p.197-221. A.A. Balkema, Rotterdam. Day Jr., J. W., C. A. S. Hall, W. M. Kemp, A. Yanez-Arancibia, and L. A. Deegan. 1989. Nekton, the free-swimming consumers. In Estua- rin ecology (Day et al., eds.), p. 377-437. John Wiley and Sons, New York, NY. de Ciechomski, J. D. 1968. Huevos y larvas de tres especies de peces marines, Anchoa marinii, Brevoortia aiirea y Pnonotus nudigula de la zona de Mar del Plata. Bol. Inst. Biol. Mar. 17, 28 p. Framiiian, M. B., and O. B. Brown. 1996. Study of the Rio de la Plata turbidity front. Part I: spatial and temporal distribution. Cont. Shelf Res. 16(101:1259-1282. Guerrero, R. A., E. M. Acha, M. B. Framiiian, and C. A. Lasta. 1997a. Physical oceanography of the Rio de la Plata Estu- ary, Argentina. Cont. Shelf Res. 17(7):727-742. Guerrero, R. A., C. A. Lasta, E. M. Acha, H. W. Mianzan, and M. B. Framiiian. 1997b. Atlas hidrografico del Rio de la Plata. Comision Administradora del Rio de la Plata-Instituto Nacional de Investigacion y Desarrollo Pesquero, Buenos Aires-Monte- video, 109 p. Haedrich, R. L. 1992. Estuarine fishes. In Ecosystems of the world, vol. 26, Estuaries and enclosed seas (B. H. Ketchum, ed.), p. 185-207. Elsevier, Amsterdam. Hubold, G., and M. D. Ehrlich. 1981. Distribution of eggs and larvae of five clupeoid fish species in the Southwest Atlantic between 25'S and 40S. Meeresforschung 29:17-29. Kjerfve, B. 1989. Estuarine geomorphology and physical oceanography. In Estuarine ecology (J. W. Day, C. A. S. Hall, W. M. Kemp, and A. Yafiez-Arancibia, eds. ), p. 47-78. John Wiley & Sons, New York, NY. Largier, J. L. 1993. Estuarine fronts: How important are they? Estuar- ies 16(1):1-11. Lasta, C. A., and J. D. de Ciechomski. 1988. Primeros resultados de los estudios sobre la distribu- cion de huevos y larvas de peces en Bahfa Samborombon en relacion a temperatura y salinidad. Publ. Com. Tec. Mix. Fr Mar 4:1.33-141. Lawler, J. P., M. P. Weinstein, W. Y. Chen, and T. L. Englert. 1988. Modelling of physical and behavioral mechanisms influencing recruitment of spot and Atlantic croaker to the Cape Fear estuary. Am. Fish. Soc. Sym. 3:115-131. Macchi, G. J., and E. M. Acha. 2000 .Spawning frequency and batch fecundity of Brazil- ian menhaden, Brevoortia aurea, in the Rio de la Plata estuary off Argentina and Uruguay. Fish. Bull. 98:283- 289. Macchi, G. J., E. M. Acha, and C. A. Lasta. 1996. Desove y fecundidad de la corvina rubia (Micropogo- nias furnieri, Desmarest, 1826) en el estuario del Rio de la Plata, Argentina. Bol. Inst. Espafiol Oceanogr. 12:99-113. Madirolas, A., E. M. Acha, R. A. Guerrero, and C. A- Lasta. 1997. Sources of scattering near an estuarine frontal system. Sci. Mar. 61(4):431-438. Mann, K. H., and J. R. N. Lazier. 1991. Dynamics of marine ecosystems: biological-physical interactions in the oceans. Blackwell Science Publica- tions, Boston, MA, 466 p. Nellen, W., and G. Hempel. 1969. Versuche zur Fangigkeit des "Hai" und des modifi- zierten Gulf-V-Plankton-Samplers "Nackthai." Ber. Deuts. Wiss. Komm. Meeres. 20:141-154. Nelson, W. R., M. C. Ingham, and W. E. Schaaf. 1977. Larval transport and year-class strength of Atlantic menhaden, Brevoortia tyrannus. Fish. Bull. 75(1):23-41. Norcross, B. L., and R. F. Shaw. 1984. Oceanic and estuarine transport of fish eggs and larvae: a review. Trans. Am. Fish. Soc. 113:153-165. Officer, C. B. 1992. Physics of estuarine circulation. In Ecosystems of the world, vol. 26, Estuaries and enclosed seas (B. H. Ket- chum, ed.), p. 1.5-41. Elsevier, Amsterdam. Potter, I. C, G. A. Hyndes, and F. M. Baronie. 1993. The fish fauna of a seasonally closed Australian estuary. Is the prevalence of estuarine-spawning species high? Mar. Biol. 116:19-30. Powell, A. B. 1994. Life history traits of two allopatric clupeids, Atlantic menhaden and gulf menhaden, and the effects of harvest- ing on these traits. N. Am. J. Fish. Manage. 14:53-64. Sanchez, R. P., and J. D. de Ciechomski. 1995. Spawning and nursery grounds of pelagic fish spe- cies in the sea-shelf off Argentina and adjacent areas. Sci. Mar .59(3-4):4.5.5-478. Shaw, R. F., J. H. Cowan Jr, and T. L. Tillman. 1985. Distribution and density of Brevoortia patronus (gulf menhaden) eggs and larvae in the continental shelf waters of western Louisiana. Bull. Mar Sci. 36( 1 ):96-103. Sinclair, M., and T. D. lies. 1989. Population regulation and speciation in the oceans. J. Cons. Int. Explor Mer 45:165-175. Sinque, C, and J. H. Muelbert. 1997. Environment and biota of the Patos Lagoon Estuary: ichthyoplankton. In Subtropical convergence environments: the coast and sea in the Southwestern Atlantic (Seeliger et al., eds.), p. 51-56. Springer-Verlag, Heidelberg. Vaughan, D. S. 1991. Menhaden: the resource, the industry, and a manage- ment history. Preface. Mar. Fish. Rev. 53(4): 1-2. Warlen, S. M. 1988. Age and growth of larval gulf menhaden, Brevoortia patronus, in the northern Gulf of Mexico. Fish. Bull. 86(l):77-90. 1992. Age, growth and size distribution of larval Atlantic menhaden ofT North Carolina. Trans. Am. Fish. Soc. 121: 588-598. Weiss, G. 1981. Ictioplancton del estuario de Lagoa dos Patos, Brasil. Ph.D. diss., Univ. Nacional de La Plata, Facultad de Cien- cias Naturales y Museo, La Plata, Argentina, 164 p. Acha and Macchi: Spawning of Brevoortia aurea in the Rio del la Plata estuary off Argentina and Uruguay 235 Weiss, G., and L. C. Krug. Whitehead, P. J. P. 1977. Caracteristicas do desenvohnmento e metamorfose de 1985. Clupeoid fishes of the world: an annotated and illus- Lycengraulis olidus (EngrauUdae) e Brevoortia pectinata trated catalogue ofthe herrings, sardines, pilchards, sprats, (Clupeidae)noestuarioda LagoadosPatos, Brasil. Allan- anchovies and wolf-herrings. Part I: Chirocentridae, Clu- tica 21 1):83-117. peidae and Pristigasteridae. FAO Fisheries Synopsis 125, Weiss, G., J. A. Souza. and A. Santos. vol. 7, 303 p. 1976. Contribuifao ao conhecimento do ictioplancton marinho da plataforma sul do Brasil. Atlantica li l/2);7-78. 236 Abstract.— Sandbar sharks iCarcha- rhinus plumbeus) support an impor- tant commercial fishery. They are man- aged as a component of a multispecies group, termed large coastal sharks, by the National Marine Fisheries Service (NMFS) under the Fishery Manage- ment Plan (FMP) for Atlantic sharks. Currently, large coastal sharks, gener- ally, and sandbar sharks, specifically, are considered overfished. Several man- agement options, including nursery ground closures and size limits, are being considered to conserve the fish- ery. We explored the implications of management options for large coastal sharks within the framework of a stage- based model. Based on biological crite- ria, the life cycle of the sandbar shark was represented as five stages: neonate, juvenile, subadult, pregnant adults, and resting adults. The model followed only females. From a fishing mortality rate (F) of 0.20, estimated in the 1996 stock evaluation workshop (SEW), the model projects a population decline to 13% of its current abundance within 20 years. The population is not stabilized until F is reduced to 0.07. In one run of the model, we assumed that F on neonates and pregnant adults was zero in order to assess the impact of a "perfect" nursery ground closure. Under this scenario, the population continued to decline unless F on the remaining stages was reduced to 0.097. Even with the closure of nurs- ery grounds or the introduction of size limits to protect neonates and juveniles, F has to be reduced substantially. The model is highly sensitive to the dynam- ics of juveniles and subadults, which implies that management should pro- tect these immature sharks to rebuild the stock. Management of the sandbar shark, Carcharhinus plumbeus: implications of a stage-based model* Karyl K. Brewster-Geisz Highly Migratory Species Management Division National Manne Fisheries Service 1315 East West Highway Silver Spring. Maryland 20910 E-mail address Karyl Brewster Geisz 3 noaa gov Thomas J. Miller Chesapeake Biological Laboratory University of Maryland Center for Environmental Sciences P O Box 38 Solomons, Maryland 20688-0038 Manuscript accepted 26 November 1999. Fish. Bull. 98:236-249 (2000). The sandbar shark (Carcharhinus plumbeus) is a species of primary importance in the Atlantic and Gulf of Mexico shark fishery (NMFS, 1993; NMFSi; Branstetter and Bur- gess- ). It is managed as a part of the large coastal shark group defined under the Atlantic shark Fishery Management Plan (FMP; NMFS, 1993). Since the mid 1980s, the demand for shark has increased (NMFS, 1993). The fishery peaked in 1989 with landings of approxi- mately 4600 metric tons (t) dressed weight ( dw; NMFS-^ ). Catch per unit of effort of large coastal sharks declined rapidly during the 1970s and 1980s. To prevent overfishing, the FMP imposed an annual quota of 2570 t dw from 1994 to 1996 for the large coastal fishery, required mandatory reporting of landings, and prohibited the removal of fins (NMFS, 1993). At the 1996 stock evaluation workshop (SEW), scien- tists found no evidence of improve- ments in the large coastal stocks and recommended reducing fishing mortality by SC}^ (NMFS, 1996). In response, the National Marine Fisheries Service (NMFS) reduced the annual quota in 1997 by SC^ to 1285 t dw and reported to Con- gress that large coastal sharks were overfished (NMFS, 1997). The most recent data indicate that fishing mortality rates have not declined as much as expected and may still be too high to stabilize the sandbar shark stock (NMFS^). A size limit equivalent to approximately 12-13 years of age (140 cm fork length) was recommended. The NMFS is mandated, through the Magnuson-Stevens Fishery Con- servation and Management Act, to * Contribution 3264 of the University of Maryland Center for Environmental Sci- ences Series, Chesapeake Biological Labo- ratory, University of Maryland, Solomons, MD 20688-0038." ' 1996. Report ofthe shark evaluation work- shop. Southeast Fisheries Science Center, Natl. Mar Fish. Serv., NOAA. Miami, FL, 80 p. - Branstetter, S., and G. Burgess. 1997. Continuation of an observer program to characterize and compare regional efforts in the directed commercial shark fishery in the eastern Gulf of Mexico and South Atlantic. Gulf and South Atlantic Fisher- ies Development Foundation, Inc.. Tampa, FL. University of Florida, Gainesville, FL., 83 p. ' 1997. Shark evaluation annual report. Southeast Fisheries Science Center, Natl. Mar Fish. Serv.. NOAA, , Miami, FL, 12 p. ^ 1998. Report of the shark evaluation work- shop. Southeast Fisheries Science Center, Natl. Mar. Fish. Ser\'., NOAA, Panama City, FL. Brewster-Geisz and Miller: Management of Carcharhinus plumbeus 237 rebuild the large coastal stocks to the optimum yield level. In October 1998, the NMFS released a draft FMP for Atlantic tunas, swordfish, and sharks. The measures in this FMP were designed to halt over- fishing and to rebuild these stocks. Management options under consideration for large coastal sharks included restrictions on effort, size limits, and area closures that were focused on nursery gi-ounds. Many traditional approaches that could be used to compare management options, such as surplus produc- tion models or age- and size-structured approaches, rely on catch or effort data, or both. However, because logbook reporting in the shark fishery was not man- datory until 1993, fishery-dependent time series have been insufficiently long to permit reliable application of these approaches. Yet a comparison of the efficacy of the potential management options is still required. The paucity of fishery-dependent data suggests that demographic approaches, such as life-table or stage- based analyses, may be the appropriate tools to explore the potential response of shark populations to management actions. Life-table analysis is a common age-structured demographic approach with a long history in pop- ulation dynamics (Kingsland, 1985). It is a matrix projection approach that estimates the contribution of each age class to future generations. Sminkey and Musick (1996) applied a life-table approach to sandbar sharks. From the intrinsic rate of nat- ural increase, r, predicted by the model, they con- cluded that the population could not sustain the observed rates of fishing mortality. Heppell et al. (1999) developed matrix-based life tables for leopard {Triakis semifasciata) and angel (Squatina califor- nica) sharks. Heppell et al. calculated the elasticity or proportional sensitivity of the population growth rate to changes in survival and fecundity and con- cluded that the two species differ in the degree to which each can compensate for changes in exploi- tation. Simpfendorfer (1999) developed a life table for the dusky shark {Carcharhinus obscurus). He concluded that in the absence of exploitation, dusky shark populations in southwestern Australia would increase at 4.3% annually. Analysis also indicated that current patterns of exploitation were sustain- able. However, there are problems in application of life-table analysis to long-lived marine species. The intrinsic rate of natural increase predicted is depen- dent on the products of survival and fecundity for all ages and the estimated generation time. Thus, life tables require estimates of the schedules of mortal- ity (survival) and fecundity over the entire age range (Gotelli, 1995). Consequently, in a long-lived species such as the sandbar shark, small errors in para- meter estimates can become magnified. Stage-based modeling is a matrix-based demo- graphic approach that considers aggregate stages (defined in terms of size or life history stages) that represent functional biological units (Gotelli, 1995). It too has a long history in the field of ecological pop- ulation dynamics (Kingsland, 1985). A stage-based model can be formed by collapsing a life table into discrete stages. Thus, unlike the life-table analysis that requires estimates for every year the organism lives, the stage-based model requires only estimates for each stage. Therefore, the realism of a many- staged model can be balanced with the precision of a simpler model when parameter estimation error is of concern. As with life tables, stage-based projection models can easily be solved analytically to permit formal sensitivity analysis (Caswell, 1989). Ander- son ( 1990), and Hoenig and Gruber (1990) have sug- gested that stage-based models may provide a more realistic view of the dynamics of some populations. The population dynamics of several marine spe- cies, including sandbar sharks (Cortes, 1999), sea turtles (Grouse et al., 1987; Crowder et al., 1994), blue crabs (Miller and Houde''), sardines (Lo et al., 1995), and anchovies (Pertierra et al., 1997) have been explored by using stage-based models. Cortes (1999) developed a stochastic stage-based model for sandbar sharks in the western North Atlantic. He used the model to explore the implications of three different harvest strategies on population viability when fecundity varies. He concluded that in the absence of exploitation, the sandbar shark popula- tion should increase slowly by about 1.3% annually. Additionally, Cortes concluded that all three pat- terns of exploitation would cause declines in popula- tion abundance. Cortes' model and results indicate the utility of stage-based models in exploring poten- tial management alternatives for sandbar sharks. Here, we develop a deterministic stage-based model for sandbar shark populations. The model includes the two-year reproductive cycle of fertile and rest- ing periods known to occur in sandbar sharks, but which were not included in Cortes's (1999) original description. We chose to use a deterministic frame- work to permit a formal elasticity (proportional sen- sitivity) analysis of the basic model. Stages to which the population dynamics are most sensitive can be interpreted either as being stages at which manage- ment action is likely to have the most impact or as stages at which parameter estimates have to be most precise because of impacts of potential environmen- tal stochasticity. We use the model to examine the expected change in population growth resulting from 5 Miller, T. J., and E. D. Houde. 1998. Blue crab targeting. U.S. EPA Chesapeake Bay Program Report, Annapolis, MD, 167 p. 238 Fishery Bulletin 98(2) two particular management options, nursery ground closures and size limits. We exercised the general model framework to address four fundamental ques- tions. What is the intrinsic rate of increase of sandbar shark populations under current patterns of exploi- tation? What is the sustainable level of fishing mor- tahty '•P critical'''^ What is the effect of ehminating fishing mortality on early stages, either through nurs- ery ground closures or through the introduction of size limits? For each question, we provide the results, predictions, and interpretation of sensitivity analyses to indicate the reliability of our conclusions. Materials and methods Life history of sandbar sharks The first step in developing a stage-based model is to review the life history of a species to identify appropriate stages. Results of tagging and age and growth studies (Springer, 1960; Casey et al., 1985; Casey and Natanson, 1992; Sminkey and Musick, 1995, 1996) indicate that sandbar sharks are a long- lived species with low fecundity. These studies also indicate that females, males, and juveniles segregate by water depth and distance from shore. However, estimates of key vital rates are inconsistent. The generally accepted estimate of mortality and fecun- dity schedules indicates that sandbar sharks mature between 12 and 15 years of age and live to around 30 years of age (Casey et al., 1985; Sminkey and Musick, 1996 ). Another estimate suggests that sandbar sharks may not mature until age 29 and may live past 50 (Casey and Natanson, 1992). In our model we used an age at maturity of 15 years. From the biological func- tion and the migration pathways determined by these studies, we identified five stages in the life history of the sandbar shark: neonates, juveniles, subadults, pregnant adults, and resting adults (Fig. 1). Neonates are young-of-the-year sharks. Sandbar shark neonates are bom fully developed at a fork length (FL) between 43 and 52 cm (Castro, 1983; Branstetter and Burgess^). They remain in this stage for one year before becoming juveniles. Juveniles are the first stage to show a seasonal pattern of movement. In the winter, juveniles migrate to warmer waters, often to the edge of the Gulf Stream off North Carolina. In the summer, juveniles return to their nursery grounds. They con- tinue this seasonal migration until they are between 6 and 10 years old (Casey et al., 1985; Branstetter and BurgessM. In contrast, subadults, while still not yet mature, follow the adult migration pattern. This migra- tion pattern consists of swimming along the Atlantic coast of the United States as far nor:th as New England / Neonates 0-1 yr / - ; l^ Juveniles 1-7 yr ; \^ Subadults 7-15 yr \ ; /^ Pregnant adults lyr f V V Resting adults lyr Figure 1 Diagram of the five-stage model. Arrows indicate individuals sur- viving and growing to the next stage or surviving and remain- ing in the same stage. in the summer and traveling south to warmer waters in the winter (Castro, 1983). In this model, individu- als remain in the subadults stage for 8 years, at which point they may be 15 years of age, and then join the reproductive population. Fifty percent of female sandbar sharks are mature at about 150 cm FL (Springer, 1960; Casey et al., 1985; Sminkey and Musick, 1996) or 12 to 15 years of age depending on the von Bertalanffy model. Female sandbar sharks give birth at an average of 8 or 9 pups once every other year (Springer, 1960; Smin- key and Musick, 1996). Larger sharks do not appear to give birth to a greater number of pups (Sminkey and Musick, 1996). Gestation lasts between 9 and 12 months (Castro, 1983). Pregnant females pup in shallow bays and estuaries along the east coast of North America, including Chesapeake Bay (Smin- key and Musick, 1996), Delaware Bay (Pratt and Merson^) and the waters off the coast of South Caro- •* Pratt Jr., H. L, and R. R. Merson. 1996. Report of the 1996 apex predators investigation: sandbar .shark nursery grounds project. Apex predators investigation. Northeast Fisheries Sci- ence Center, Natl. Mar., Fish. Serv., NOAA, Narragansett, HI. Brewster-Geisz and Miller Management of Carcharhlnus plumbeus 239 lina (Castro, 1993). In the model, females alternate between pregnant and resting adult stages, spend- ing one year in each. Thus, the stage durations used in the model were the following: 1 year for neonates; 6 years for juveniles; 8 years for subadults; 1 year for pregnant females; and 1 year for resting females. Model development The approach we present below is based on the gen- eral framework presented by Caswell ( 1989). Details on the general background of the approach can be found in Caswell (1989). In all equations, matrices and vectors are shown in boldface type, parameters in italic type. The model is a postbreeding census, follows only females, and uses a yearly time step. The total number of sharks in the population at time t can be expressed in vector form as N^. Each element of N, represents the number of sharks in the appropriate stage at time t. There are three possible transitions for each individual in each stage: the probability of surviving and growing to the next stage, G,; the prob- ability of surviving yet remaining in the same stage, P,; or the probability of dying, 1-G— P,. The individual transition probabilities G, and P, may range between and 1. The sum of G, and P, is further constrained such that when a stage is not subject to mortality, G, + Pj = 1. One other parameter, stage-specific fecun- dity, is required to estimate the number of young females produced per breeding female per year. The vital rates governing the dynamics of the shark population can be expressed mathematically in a 5 X 5 transition matrix, A. The fundamental equation to estimate the stage-structure in the pop- ulation at any time t is given by N, = A' - N„ (1) where N^ = a vector of the number of individuals in each stage at time /; and the transition matrix A for sandbar sharks given by A = G^xf^ Gi P„ G2 ^3 G, G. (2) For large values of t, AN, = AN,=N,^j, where the scalar A is the finite rate of population increase. Fur- ther, InA = r, the intrinsic rate of increase. The columns of the matrix represent the fates of individuals in each stage. For example, surviving neo- nates can grow only to the juvenile stage (G,) where- as surviving juveniles can either remain a juvenile (P2) or survive and grow into a subadult iG^)- Sur- viving pregnant adults can give birth and become resting adults (G^). Surviving resting adults (Gr,) can grow only into a pregnant female. The rows repre- sent the origins of individuals in each stage. Neo- nates arise from pregnant adults who survive (G^x/^j) whereas juveniles arise from neonates surviving and growing into juveniles (Gj) or from juveniles surviv- ing and remaining juveniles (Pg). Pregnant adults can arise from subadults surviving and growing into a pregnant female (G3) or from resting adults sur- viving and becoming pregnant adults (Gr,). Resting adults can arise only from pregnant adults that sur- vive (G4). The transition probabilities, P, and G,, can be cal- culated from estimates of the probability that during a single time step an individual of stage /' survives, a,, and an individual of stage / grows, y^. In this way G,, the probability of surviving and gi'owing to the next stage is given by G, = oj. Consequently P,, the probability of surviving, not growing to the next stage is given by P, = a,(l-y,). (3) but (4) The probability of survival, a^, over a single time step can be expressed as a, = e-^i. (5) Following traditional fisheries models, total mortal- ity (Z,) is calculated by using the equation Z, = F^ + M,, where F, is the rate of fishing mortality and M, is the I'ate of natural mortality at stage /. Estimates of 7, can be obtained in several different ways. Caswell (1989) recommended assuming con- stant stage duration for all individuals in the stage when only a relatively crude estimate of survival over broad age ranges is available. For this approach, individuals entering the stage have an equal prob- ability of survival as individuals nearing the end of the stage. Employing this assumption yields an esti- mate of Y, as Y, (CT,/A„,„) -((T, A J.-i <^- ^,n„ .1-1 (6) where T the expected stage duration of a single stage; and 240 Fishery Bulletin 98(2) A = the finite rate ofpopulation increase given by the dominant eigenvector of A (lnA=r, the intrinsic rate of increase). We began with an initial value of A,„„ = 1. We then iterated A until the value of A given by an eigenanal- ysis of the matrix A (see below), equaled the value of A„j„ used in Equation 6. Together, Equations 1-6 described above allow estimates of G, and P, within A to be defined. Fecundity must also be defined. The fecundity term, /j, is simply the expected number of female offspring produced by a female in stage i. For sand- bar sharks only one stage is reproductively active and thus the only fecundity term in the matrix A is f^ = 4.5. This estimate is based on an equal sex ratio, 9 pups per brood, and one brood per year. However, as the model is a postbreeding census, the fecundity has to be discounted by the probability that a preg- nant female will survive the gestation year to pup. Thus the realized fecundity term used in the model is G4 X f^. All parameters within the matrix A are now defined. One feature of stage-based projection models that motivated their use was that they allowed us to solve A analytically in order to calculate important demo- graphic features and find the sensitivity of the model to parameter estimates. The two demographic fea- tures that can be calculated from A are the stable stage distribution and the reproductive value of each stage. Once the stable stage distribution has been reached, the relative proportion of individuals in each stage remains constant over time. The I'epro- ductive value is the relative number of offspring that are yet to be born by individuals in a given age (Gotelli, 1995). This value depends on individu- als surviving to maturity and reproducing. Thus, the youngest stages should have the lowest reproductive values because individuals in those stages must sur- vive and reach maturity before they can reproduce. Both features can be calculated from an eigenanaly- sis of A. For any |/!xn| matrix one may define up to n scalar values (Aj ,, ) and /! -associated right and left vectors such that Aw = Aw vAT - Av (7) (8) where A^ = the transpose of A; A = the eigenvalue; and w and V = the right and left eigenvectors of A. interpretation is simplified for the sandbar shark transition matrix. A, because it is non-negative, irre- ducible, and primitive. Thus, we are guaranteed that there is a single, dominant eigenvalue, Aj, that is real, positive, and strictly greater than all other pos- sible As. This dominant eigenvalue, Aj will eventu- ally describe the population rate of increase and In Aj = r, the intrinsic rate of increase of the popula- tion. Moreover, the right and left eigenvectors associ- ated with Aj will be strictly positive. The population structure will eventually become proportional to a single stable stage distribution, given by Wj. Finally, there will be a single vector, Vj, associated with Aj, that expresses the relative contributions of each stage to the future population — a vector of reproduc- tive values. Reproductive values are standardized so that the reproductive value of an individual in the first stage is one. We were interested in calculating the change in A following a change in vital rates expressing a tran- sition from stage / to any other stage (including remaining in i ) that may have been caused by man- agement activities. This change reflects the sensitiv- ity of A to the transition probability. If entries in the transition matrix A are represented as a, „ it can be 'J shown that 3A da,, (9) i>,v) where = the scalar product of the two vectors. Simply stated, the sensitivity of the population growth rate to changes in any vital rate is the prod- uct of the reproductive value of stage / and the proportional level of stagey in the stable stage dis- tribution. Because transition probabilities are censored parameters, varying only between and 1, and fecundity is noncensored, it is more helpful to report the elasticity of A. This is defined as the proportional change in A for proportional changes in a . Elasitici- ties are calculated as 3A A da. (10) Importantly, elasticities are additive, such that the sum of elasticities for each stage defines the pro- portional contribution of a to overall population growth, A. as The sandbar shark transition matrix has five pos- sible eigenvalues and eigenvect9rs. However, our II'„ Brewster-Geisz and Miller: Management of Carcharhinus plumbeus 241 Elasticities depend on a stable stage distribution and should be compared qualitatively. Transition probability estimation for management options Current conditions Fishing mortality (F) and espe- cially natural mortality (M ) are difficult to estimate. Owing to the uncertainty in estimates and in order to simplify the model, we used an estimate of M = 0.1 for all stages and all projections (NMFS^ Sminkey and Musick, 1996). We projected the population for- ward using F = 0.20 for juveniles and older stages, as estimated in the 1996 shark evaluation workshop (SEW) for sandbar sharks only (NMFS'^). For neo- nates a lower value of F - 0.10 was used because small sharks may be, but are not as likely to be, caught on the same gear as older sharks (Branstet- ter and Burgess' ). Using these values of F and M and f^=4.5, we iterated Equation 6 to estimate all P, and G, values. We initialized the population with 1000 neonates. Then we estimated the initial number in subsequent stages using the dO'Jr survival schedule for sandbar sharks given in Sminkey and Musick ( 1996). These calculations yielded an initial popula- tion of 9640 sharks (Nq). Estimate of Fcritical ^^ defined Ff^j^jj^f^j^i^ as the limiting level of fishing mortality that is sustainable, i.e. the F for which r = or A = 1, where r = InA. We systematically reduced F on all stages to define the relationship between F and r. For our estimations, Fj remained 0.10 as long as ^234 5 ^^^ >0.10. For any ^2 3 4 5 <0- 1^- ^i =^2345- Thus, the fishing mor- tality of neonates was never greater than the fishing mortality on other stages. Protecting neonates and pregnant adults: an extreme example We used the model to determine how effi- cient protecting different stages would be in pro- moting recovery of sandbar shark stocks. We asked the question: If neonates and pregnant adults are removed from the commercial fishery, how much will F on other stages need to be reduced to arrive at a sustainable population level? To address this question, we modified the model to remove all mor- tality on neonates (F^=0, Mj=0) and to protect all pregnant adults from fishing pressure (^^=0). In reality, we could not completely protect neonates from mortality (i.e. Mj>0) and we could not fully pro- tect pregnant adults from commercial fishing. Thus the scenario represents an idealized nursery closure scheme. Fishing mortality on juveniles, subadults, and resting adults remained at 0.20. The fecundity for pregnant adults was left at 4.5. Nursery closures and size limits We also ran the model using more realistic scenarios. In this case F on neo- nates and juveniles was 0, and F on the older stages was 0.20. Natural mortality for all stages remained at 0.10. This scenario is a fairly realistic size limit or nursery ground closure because sandbar sharks seg- regate by size. This scenario is similar to, but not as strict as, the 1998 SEWs recommended size limit. Size limits protecting only one stage are another management option available. This method can be used to reduce the fishing mortality on any range of sizes. In this paper, two scenarios of this type are pre- sented: 1 ) a size limit which reduces the F on juve- niles to 0; and 2) a size limit which reduces the F on subadults to 0. Fishing mortality was equal to 0.20 in all other stages except neonates, where F=0.10. Implementing such management actions would be difficult because the gear (longlines) cannot realisti- cally avoid catching only the restricted stage, but the results would be indicative of the potential of these mechanisms to improve stocks. Using the relationships (Eqs. 1-10) and vital rate estimates defined above, we now proceed with an analysis of the population dynamics of sandbar sharks. For each scenario, we calculate the stable stage distribution, the proportional reproductive value for each stage, and the elasticity of A to changes in each matrix parameter, and compare the popula- tion growth rate and potential population reduction after 20 years for each scenario. Results Current conditions When F = 0.20 for all stages except neonates, the population decreases (Table 1). The model predicts the intrinsic rate of natural increase, r, of the popu- lation as r = -0.124. The population is 13% of the initial abundance when projected 20 years forward. Population growth rate, stable stage distribution, and reproductive values are not affected by choice of the actual numbers used for initial abundance. The stable stage distribution is reached after 21 years in this scenario. The largest proportion of the population (>0.56) are juveniles (Table 2). The smallest proportions (0.04, 0.03) are pregnant and resting adults, respec- tively. Adults have much larger reproductive values than prereproductive stages (Table 3). The pattern of model proportional sensitivity is shown in Figure 2. The elasticity of A to a small change in fecundity was expressed only in the preg- 242 Fishery Bulletin 98(2) Table 1 The reduction in popu ation abundance after 20 years of each scenario. Scenario Fishing and natural mortality rates used Intrinsic rate of increase, r Reduction i^c) in population abundance after 20 years' Current conditions Fi=0.10;F2_5=0.20 M=0.10 -0.124 87 Protecting neonates and pregnant adults Mi=0;M2_5=0.10 -0.079 62 Protecting neonates and juveniles f'l. 2=0; /^3 -5=0-20 M=0.10 -0.058 51 Protecting juveniles only Fi=0.10;F2=0 F3_5=0.20;M=0.10 -0.069 62 Protecting subadults only F,=0.10;F3=0 Fj ,,g=0.20;M=0.10 -0.048 50 ' This percent reduction in population abundance is by using the equation 1 - e", where r = the intrinsic rate of increase; and t = the number of years. based on numbers generated in the model . Population abundance can also be calculated Use of this equation results in a greater reduction in the population abundance- Table 2 The stable stage d stributions for each stage of the model and each scenario. The values of all stages under a scenario should sum to one. Current Current Protecting Protecting conditions, conditions. neonates and neonates and Protecting Protecting Model stage unstablized stablized pregnant adults juveniles juveniles subadults Neonate 0.15 0.16 0.15 0.10 0.12 0.19 Juvenile 0.56 0.54 0.60 0.54 0.51 0.51 Subadults 0.22 0.22 0.18 0.31 0.31 0.21 Pregnant adults 0.04 0.04 0.03 0.03 0.03 0.05 Resting adults 0.03 0.04 0.03 0.02 0.03 0.04 nant adult stage (because this is the only stage that is reproductively active). The highest elastic- ities were for the transition from neonate to juve- nile and juvenile to subadult stages (Fig. 2). The elasticities for sharks remaining in the stage were equal for neonate, juvenile, subadult, and pregnant female stages. As discussed, the individual elastici- ties can be summed to estimate the overall contri- bution of each stage to A. It is clear from Figure 2 that the peak elasticity occurs in the subadult stage. Estimates of elasticity suggest that the model is 2.3 times more sensitive to changes in this stage than in pregnant adults. Estimate of Fc«/r/c>ii The relation between F and r is linear ( Fig. 3 ). Fp^^j-^ f.^^ = 0.071 when M - 0.10. Therefore, the population is sustainable if F = 0.071. The value of Ff^^/y-^f.^^ will vary with the value of M that is chosen. Our analy- ses showed that total mortality (Z) must be less than 0.17 for all stages if the population is to increase (Fig. 4). The population will increase at a rate r = 0.05 if Z = 0.122, and the population will decrease at a rate r = -0.05 if Z = 0.222. At FcRiTiCAL' ^^^ population abundance stabilizes and the population reaches a stable stage distribu- Brewster-Geisz and Miller; Management of Carcharhinus plumbeus 243 Table 3 The proportional standardized reproductive values of each stage under eac h scenario. Neonates will always equal one. Model stage Current conditions, unstablized Current conditions, stablized Protecting neonates and pregnant adults Protecting neonates and juveniles Protecting juveniles Protecting subadults Neonate 1.00 1.00 1.00 1.00 1.00 1.00 Juvenile 1.08 1.19 0.92 1.04 1.26 1.29 Subadults 3.10 3.32 3.48 1.34 1.52 5.83 Pregnant adults 12.72 13.06 20.44 9.23 9.65 8.84 Resting adults 10.67 11.00 16.38 7.25 7.66 6.87 tion (Table 2). In this scenario, the proportion of neonates, pregnant aduhs, and resting adults in the population has increased compared with the proportion of stages in previous scenarios. The pattern in repro- ductive values remains largely unchanged (Table 3). As with the baseline scenario, the elas- ticities show the model to be the most sensitive to changes at the juvenile and subadult stages. In summary, these model runs suggest that to stabilize and increase the sandbar shark pop- ulation, F needs to be reduced below 0.07 (Z<0. 17 ). An F of this magnitude requires more than the full 50'7r quota reduction to be implemented. Protecting neonates and pregnant adults: an extreme example Results show that, even after protection of neonates and pregnant females, at current levels of F the pop- ulation still decreases rapidly (r=-0.079). The popu- lation, projected 20 years forward, is only 38% of the initial abundance (Table 1). This percentage com- pares with reductions to 139f of initial abundance and ;-=-0. 124 in the base run, where F^ 345 = 0.2. In order to have a stable population ir-O) under this scenario, we needed to decrease F to 0.097. Without protecting neonates and pregnant females, F must be reduced to 0.07. Thus, protecting neonates and pregnant females provides a 37% increase in the F required to maintain a sustainable population (Table 4). However, it must be emphasized that although Neonate Juvenile Subadult Pregnant adult Resting adult I Fecundity i/'i H Growth from stage iG, i D Stage residence iP, i Figure 2 The proportional sensitivities (elasticity I of each stage to fecundity, growth, and stage residence under current fishing conditions (F=0.20). this option does provide some protection, implemen- tation would still require a 52% reduction in F over those levels currently estimated to be operating in the fishery. This reduction is in contrast to the 64% reduction in F required to reach sustainable rates of exploitation in the absence of this protection. Juveniles had the highest proportion of individuals in the stable stage distribution (Table 2). Pregnant females and resting adults have the highest repro- ductive values (Table 3). Similar to the previous sce- nario, projections show the highest overall sensitivity to transitions involving the abundance of subadults (Fig. 5). However, in contrast with earlier simulations, projections show additional substantial sensitivity to transitions into the resting adult stage (Fig. 5). 244 Fishery Bulletin 98(2) Nursery ground closures and size limits Results of this scenario show that the protection of neonates and juveniles from all fishing mortality slowed the decline (r=-0.058) but could not stabilize the population. When projected forward 20 years, the population abundance is 49% of the initial abun- dance (Table 1). In order to stabilize this model, F had to be reduced on subadults and adults to 0.109. Thus, this closure provides a 54% increase in the F -0.10 -0.20 required to maintain a sustainable population over that required in the absence of nursery closures and a 12% increase in F over the extreme option mod- eled above (Table 4). Protecting neonates and juve- niles through nursery ground closures or size limits would require a 46% reduction in F over those levels currently estimated to be operating in the fishery. Juveniles have the highest proportion of individu- als in the stable stage distribution ( Table 2 ). Pregnant adults and resting adults have the highest reproduc- tive values (Table 3). Again, the model shows the highest sensitivi- ties to the juvenile and subadults stages (Fig. 6). Protecting either juveniles or sub- adults alone still leads to a declin- ing population. When F2=0, after 20 years the population is 38% that of the initial population (Table 1). When F3=0, the population at 20 years is 50% that of the initial population (Table 1). Further runs indicated that the population is sta- Instantaneous rate of fishing mortality Figure 3 The relation between the intrinsic rate of increase (r) and fishing mortality IF). F,.r,i,cai '^ reached at 0.071. If F is less than F^r,r,ra/' ^^^ population will increase. If F is greater than f, .„(„.„/. the population will decrease. 0.03 Fishing mortality Figure 4 Isoclines showing the intrinsic rate of increase (r) at different rates of fishing iF) and natural mortality (A/l. The population will increase if r is greater than zero. The population will decrease if r is less than zero. bilizedifF. 3,4,5 0.101 (whenF2=0) or if F245 = 0.120 (when ^3=0). Quota reductions of 50% and 40%, respectively, are required to achieve these critical levels of F. In both cases the stable stage distribution is achieved within 24 years. The stable stage distribution proportions and reproductive values of each stage are listed in Tables 2 and 3. Figure 7 shows the sensitivity of the model to fecundity, growth, and stage res- idence when F.2 - 0. These sensitivi- ties were approximately the same when F3 = 0. As in the other models, juveniles and subadults have the highest sensitivity. Discussion The model projects that the sand- bar shark population is unlikely to increase unless F is reduced below F CRITIC M.- '^^^ value calculated here is less than the F critical value of 0.10 that Sminkey and Musick (1996) predicted by using a life table. Both Sminkey and Musick's (1996) and Cortes's (1999) results and those presented here indicate that current estimates of F are Brewster-Geisz and Miller: Management of Carcharhinus plumbeus 245 Neonate Juvenile Subadult Pregnant adult Resting adult Fecundity if) B Growth from stage ( G, ) D Stage residence i P, ) Figure 5 The proportional sensitivities (elasticity) of each stage to fecundity, growth, and stage residence if neonates and pregnant adults are protected (Fj ^=0; F 23 5=0.20). Table 4 The percent reduction in fishing mortality needed to stabilize the population 'f^r.rr,.,,, = 0.20 ^rr,„™; is thefi shing mortality level at which the population abundance is stable Currfrt is the fishing mortality 1 evel estimated in the 1996 stock assessment. Fishing and natural '7f reduction Scenario mortality rates used '^ Cnlical needed Current conditions F,=0.10;F2_5=0.20 M=0.10 0.071 64 Protecting neonates and pregnant adults f-i. 4=0; ^2. 3. 5=0.20 Mi=0;M2_5=0.10 0.097 52 Protecting neonates and juveniles Fi .,=0; F, _5=0.20 M=0.10 0.109 46 Protecting juveniles only Fi=0.10;F2=0 F3_5=0.20;M=0.10 0.101 50 Protecting subadults only F,=0.10;F3=0 F., , -=0.20;Af=0.10 0.120 40 too high to maintain the population and must be reduced. At the 1996 SEW, it was determined that reducing F levels by 50% was likely to increase the chances of recovering the large coastal stock (NMFSi). In response to this, NMFS reduced the quota in 1997 for Atlantic large coastal sharks by 50% in order to reduce F by BOVc. Assuming that the estimate of F from the SEW is accurate, that a 50% quota reduction is equivalent to a 50% decrease in F, and that the reduction in F is equally distrib- uted across age classes, we believe our results indi- cate that a 50% quota reduction may not stabilize the stock. Our model predicts that without alter- native management strategies, the population will not begin to recover unless F on sandbar sharks is reduced to below 0.07, requiring a reduction in cur- rent estimates of F of greater than 50%. Nursery ground closures and size limits are possi- ble management strategies. These strategies would protect neonates, pregnant adults who are in the 246 Fishery Bulletin 98(2) nursery grounds to pup, and any juveniles who may have returned for the summer. One of the scenarios we ran is extreme in that every neo- nate survives (Z=0) and preg- nant adults are not fished for the entire time they are preg- nant. But in this scenario, juveniles are not protected. If F is not reduced on juveniles, subadults, and resting adults, the model shows that the pop- ulation will decrease until F is reduced to 0.097. This is higher than the F = 0.07 which would be needed to stabilize the population without any protection for neonates, and would almost be met by the 50'^ reduction in quota as sug- gested by the 1996 SEW. How- ever, these scenarios assume complete protection of protected stages from either fishing or natural mortality. Thus, they probably over-estimate the effectiveness of the potential man- agement action. Overall, these models indicate that nursery ground closures or size limits that protect only neonates and juveniles, or neonates and preg- nant adults, are not likely to be the ultimate solu- tion. Additional measures will need to be taken to protect the sandbar shark. Subsequent runs of the model showed that size limits that protect juvenile and subadult stages will not act to rebuild the population alone, despite the fact that the model indicates these stages are the most sensitive to survival. In these cases, F needs to be reduced to 0.10 or 0.12, respectively. If the current F estimate of 0.20 is correct and if a 50% reduction in quota reduces F by 50%, size limits to protect either stage and a reduction in quota of between 40% and 50% may be sufficient to stabilize the population. All scenarios indicate that the sandbar shark stock will most likely be rebuilt through a combination of management strategies. With nursery closures or size limits that protect only one stage, the stock will decline if fishing mortality remains the same as that currently estimated. Because the model's esti- mates of population growth are sensitive to survival at the juvenile and subadult stages, because these stages have the highest proportion of the population in the stable stage distribution, and because sub- adults have relatively high reproductive values, ide- ally any management strategies selected should be those that conserve these stages. , Neonate Juvenile Subadult Pregnant adult Resting adult Fecundity (/) Growth from stage ( G, ) D Stage residence (P,l Figure 6 The proportional sensitivities (elasticity) of each stage to fecundity, growth, and stage residence if neonates and juveniles are protected iF, .,=0; F., ^ -=0.20). Most of the model projections indicate that the total sensitivities of juveniles and subadults are the greatest. The sensitivity of population growth to events during these stages suggests two things: man- agement needs to focus on protecting juveniles and subadults, and scientists need to collect accurate esti- mates of F and M for juveniles and subadults. Possi- ble conservation efforts could include minimum sizes to protect immature sandbar sharks or time-area clo- sures to protect both juveniles and subadults during their migrations. If our model is correct, it is impor- tant to take measures to protect these stages soon, not only because the model shows that the population abundance decreases quickly at current estimates of F, but also because there is evidence of strong year classes of immature sandbar sharks entering the fishery ( Branstetter and Burgess^). In 1994, Sminkey suggested that the 1987, 1989, and 1992 year classes in Chesapeake Bay were exceptionally strong. It will be easier for the fishery to recover if we have strong year classes on which to build. This is not the first time the use of a stage-based model has concluded that conservation efforts should target juveniles and subadults of a long-lived species more than newborns. Rates of population growth in many marine species are effected more by changes in survival of juvenile and subadult stages than by changes in survival of other stages, or by changes in fecundity (Heppell et al., 1999). For example, Grouse et al. (1987) and Crowder et al. (1994) concluded that population growth rate was most sensitive to the survival of large juvenile loggerhead sea turtles. Brewster-Geisz and Miller: Management of Carcharhinus ptumbeus 247 0.35 0.30 0.25 g 0.20 Hi « S 0.15 0.10 0.05 t- 0.00 Neonate Juvenile Subadult Pregnant adult Resting adult I Fecundity ^f) M Growth from stage (G, ) D Stage residence (P, ) Figure 7 The proportional sensitivities (elasticity) of each stage to fecundity, growth, and stage residence if only juveniles are protected lF,,=0; F,=0.10; F, ^ -=0.20 >. They suggested that the use of turtle excluder devices would protect juvenile sea turtles and aid in conserva- tion and recovei7 of this species. Additionally, Heppell et al. (1996) indicated that population enhancement by means of hatchery production of sea turtles would likely not be successful. In contrast, marine mam- mals show a different pattern of sensitivity. For these taxa, population gi-owth appears most sensitive to events occurring during the adult stages (Heppell et al., 1999). Studies indicate that maiine fish may also show a different pattern of sensitivity, where there is increased sensitivity to events in the early life history (Heppell et al. 1999; Quinlan and Crowder, 1999). Three features of our approach require one to use caution in interpreting our conclusions. First, there are problems in using stage-based models, or demo- graphic models of all types, with highly migratory animals such as sharks. For instance, most demo- graphic models assume that the population is closed. In the case of the sandbar shark this assumption may not hold true. Tagging studies show that a small percentage of sandbar sharks tagged in U.S. waters are caught in Mexican waters. Because it is currently unknown if there are nursery grounds in Mexican waters, this migration to Mexico may rep- resent an additional source of loss to the population that may not be replaced. The model does not con- sider this loss. If a significant number of sandbar sharks are found to migrate to Mexican waters, cur- rent estimates oiF may be underestimates. If this is the case, even greater reductions in F may be neces- sary to help the stock recover. Second, we have presented a deterministic model of sandbar shark population dynamics. Thus, we have ignored uncertainty of and plasticity in vital rates such as growth and fecundity. Tuljapurkar ( 1997 ) and Nations and Boyce ( 1997 ) have discussed the potential biases that may result from basing har- vest strategies on results from a deterministic model, particularly when juvenile survival is closely tied to environmental conditions. In addition to potential biases in the results, a deterministic model yields only a point estimate of population growth rate. Cortes (1999) included a stochastic term for fecun- dity in his model for sandbar sharks. Subsequently, he used Monte Carlo simulations to generate distri- butions of predicted population gi'owth rates when fecundity varies stochastically. His results indicate that predicted population growth rates may vary by 2-3*^ when fecundity is allowed to vary. The impact of stochasticity in survival and growth on the pre- dicted population growth rates is unknown. How- ever, given the sensitivity of the model to transition involving growth and survival for juvenile and sub- adult animals, its impact may be substantial. Finally, unlike many traditional fishery models, our demographic model does not take into account den- sity dependence or compensation. However, given the longevity and age to maturity of sandbar sharks, and sharks in general, compensation may not be as signif- 248 Fishery Bulletin 98(2) icant or as observable as that for teleost fish. Sminkey and Musick ( 1995) have suggested three mechanisms of compensation in sharks: decreases in natural mor- tality of younger sharks as the abundance of preda- tory larger sharks is reduced; compensatory increases in fecundity when food is more available or when uter- ine mortality is reduced; and an increase in growth rate and thus a decrease in natural mortality and pos- sibly earlier maturity when food is abundant. In Ches- apeake Bay, they found evidence of a slight increase in growth rate of juvenile sharks after the popula- tion had been depleted but were not able to ascertain if the age of maturity had also been reduced. Late age at maturity due to relatively slow growth rates reduces the probability that small increases in growth or increased neonate survival through density-depen- dent mechanisms will compensate for fishing mortal- ity (Heppell et al., 1999). In summary, the results when F = 0.20 for older stages indicate that sandbar shark stocks are cur- rently being fished above their ability to replace themselves (i.e. r is negative for the best estimate of F). Thus, management action (e.g. time area clo- sures, reduced quota, minimum size) is needed to reduce F to the level where r is zero or positive. Because the model is highly sensitive to juvenile and subadult survival, management actions that reduce the mortality rates of these stages would likely be more effective than nursery closures that protect only neonates and pregnant females. Although our study suggests that the protection of juvenile and subadult sandbar sharks may aid in recovery of sandbar sharks, our method may not work as well on other shark species, because life history traits differ. Sandbar sharks are often con- fused with other shark species such as the dusky shark; therefore, whichever management strategy is chosen, it should work for all large coastal shark species. These problems, combined with a paucity of data on pupping grounds, age at maturity, and other traits, make selection of a conservation method dif- ficult. The model in our study should be viewed as a starting point for looking at the effect of the different options available and for comparing these options among the shark species involved. Acknowledgments We thank Selina Heppell, Pamela Mace, Michael Fogarty, John Musick, Thomas Sminkey, Ed Houde, and Steve Branstetter for comments on an earlier draft of this manuscript. Support for this work comes from the Chesapeake Bay Program USEPA (grant CB-993080-02), the Hudson River Foundation (grant 008/98A), and the University of Maryland Center for Environmental Sciences. Literature cited Anderson, E. D. 1990. Fishery models as applied to elasmobranch fisheries. In Elasmobranchs as living resources: advances in the biol- ogy, ecology, systematic, and the status of the fisheries (H. L. Pratt, S. H. Gruber, T. Taniuchi, eds. ), p. 473-484. U.S. Dep. Commer., NOAA Technical Report NMFS 90. Casey, J. G., and L. J. Natanson. 1992. Revised estimates of age and growth of the sandbar shark (Carcharhinus plumbeua) from the western North Atlantic. C. J. Fish. Aquat. Sci. 49:1474-1477. Casey, J. G., H. L. Pratt Jr., and C. E. Stillwell. 1985. Age and growth of the sandbar shark iCarcharhinus plumbeus) from the western North Atlantic. Can. J. Fish. Aquat. Sci. 42:963-975. Castro, J. 1983. The sharks of North Amencan waters. Texas A&M Univ Press, College Station, TX, 180 p. 1993. The shark nursery of Bulls Bay, South Carolina, with a review of the shark nurseries of the southeastern coast of the United States. Environ. Biol. Fish. 38:37-48. Caswell, H. 1989. Matrix population models. Sinauer Associates. Inc., Sunderland, MA, 328 p. Cortes, E. 1999. A stochastic stage-based population model of the sandbar shark in the western North Atlantic. Am. Fish. Soc. Symp. 23:115-136. Crouse, D. T., L. B. Crowder, and H. Caswell. 1987. Astage-based population model for loggerhead sea tur- tles and implications for conservation. Ecology 68:1412- 1423. Crowder, L. B., D. T. Crouse, S. S. Heppell, and T. H. Martin. 1994. Predicting the impact of turtle excluder devices on loggerhead sea turtle populations. Ecol. Appl. 4:437-445. Gotelli. N. J. 1995. A primer of ecology. Sinauer Associates, Sunder- land, MA, 206 p. Heppell, S. L., L. B. Crowder, and D. T. Crouse. 1996. Models to evaluate headstarting as a management tool for long-lived turtles. Ecol. Appl. 6:556-565. Heppell, S., L. B. Crowder, T. R. Menzel. 1999. Life table analysis of long-lived marine species, with implications for conservation and management. Am. Fish. Soc. Symp. 23:137-148. Hoenig, J. M., and S. H. Gruber. 1990. Life history patterns in the elasmobranchs: impli- cations for fisheries management. In Elasmobranchs as living resources: advances in the biology, ecology, system- atic, and the status of the fisheries (H. L. Pratt, S. H. Gruber, T Taniuchi. eds.), p. 1-16. U.S. Dep. Commer, NOAA, Tech. Report NMFS 90, Kingsland, S. E. 1985. Modeling nature: episodes in the history of popula- tion ecology. Chicago Univ. Press, Chicago, IL, 267 p. Lo, N. C. H., P. E. Smith, and J. L. Butler. 1995. Population growth of northern anchovy and Pacific sardine using stage-specific matrix models. Mar Ecol. Prog. Ser. 127:15-26. Brewster-Geisz and Miller: Management of Carcharhinus plumbeus 249 NMFS (National Marine Fisheries Service). 1993. Fishery management plan for sharks of the Atlantic ocean. Office of Sustainable Fisheries. U.S. Dep. Commer., National Oceanic and Atmospheric Administration (NOAA), NMFS. Silver Spring, MD, 173 p. 1997. Report to Congress: status of fisheries of the United States. U.S. Dep. Commer.. NOAA, Silver Spring, MD, 75 p. Nations, C. S., and M. S. Boyce. 1997. Stochastic demography for conservation biology. In Structured models in marine, terrestrial and freshwater systems (S. Tuljapurkar and H. Caswell, eds.l, p. 450-469. Chapman and Hall. New York, NY, 643 p. Pertierra, J. P.. J. Lleonart. and N. C. H. Lo. 1997. Application of a stage-specific matrix model and length-cohort based analysis to assess the anchovy fishery in Catalan coastal waters (NW Mediterranean Sea). Fish. Res. 30:127-137. Quinlan. J. A., and L. B. Crowder. In press. Searching for sensitivity in the life history of Atlantic menhaden: inferences from a matrix model. Fish. Oceanog. Simpfendorfer, C. A. 1999. Demographic analysis of the dusky shark fishery in southwestern Australia. Am. Fish. Soc. Symp. 23:149-160. Sminkey, T. R., and J. A. Musick. 1995. Age and growth of the sandbar shark, Carcharhinus plumbeus, before and after population depletion. Copeia 4:871-883. 1996. Demographic analysis of the sandbar shark, Carcha- rhinus plumbeus. in the western North Atlantic. Fish. Bull. 94:341-347. Springer, S. 1960. Natural history of the sandbar shark, Eulamia mil- berti. U.S. Fish Wildl. Serv. Fish. Bull. 61:1-38 Tuljapurkar, S. 1997. Stochastic matrix models. In Structured models in marine, terrestrial and freshwater systems ( S. Tuljapurkar and H. Caswell, eds.), p. 59-87. Chapman and Hall. New York, NY, 643 p. 250 Abstract.— Food habits of the South American sea lion (Otaria flavescens) off Patagonia were studied by means of stomach content analysis. The sam- ples were collected during 1982-1987 and 1990-1998 in northern and central Patagonia. The samples (/!=59) came from individuals found dead on beaches and from animals recovered in inciden- tal catches of the fishery. Forty-one prey species (including fishes, cephalopods, crustaceans, gastropods, polychetes, sponges, and tunicates) were identi- fied; most important were Argentine hake (Merluccius hubbsi), red octopus {Enteroctopus megalocyathus), Argen- tine shortfin squid illlex argentinus), "raneya" iRaneya brasiliensisK Patago- nian squid (Loligogahi), and Argentine anchovy (Engraulis anchoita). Differ- ences in diet were found between sexes but not between geographical area of sampling, period of sampling, or source of samples. Females fed mostly on ben- thic species, whereas males fed mostly on demersal-pelagic species. The differ- ence in diet between sexes was asso- ciated with different feeding grounds or different home ranges and could be produced by different constraints in the feeding behavior of each sex. These different constraints and restric- tions could lead females to feed in more coastal and shallower waters than those waters where males feed. Some of the important prey were commercial spe- cies (Argentine hake, Argentine short- fin squid, Patagonian squid I consumed at both commercial and noncommer- cial sizes by sea lions. The presence of gastroliths was independent of the presence of stomach parasites; how- ever, gastrolith weight was positively correlated with individual sea lion's length, indicating that gastroliths could be involved in buoyancy control. In sum- mary, these stomach content analyses indicate that South American sea lions feed primarily on demersal and ben- thic species and, in general terms, use resources according to their environ- mental availability. Food habits of the South American sea lion, Otan'a flavescens^ off Patagonia, Argentina Mariano Koen Alonso Enrique A. Crespo Susana N. Pedraza Centre Nacional Patagonico (CONICET) and Universidad Nacional de la Patagonia Boulevard Brown 3600 (9120) Puerto Madryn. Chubut, Argentina E-mail address koenigcenpateduar Nestor A. Garcia Centre Nacional Patagonico (CONICET) Boulevard Brown 3600 (9120) Puerto Madryn. Chubut, Argentina Mariano A. Coscarella Centre Nacional Patagonico (CONICET) and Universidad Nacional de la Patagonia Boulevard Brown 3600 (9120) Puerto Madryn, Chubut, Argentina Manuscript accepted 9 August 1999. Fish. Bull. 98:250-263 (20001. The South American sea hon (Otaria flavescens) is one of the most common and abundant marine mammal spe- cies in the southwestern Atlantic and is distributed along the coasts of South America from Peru to south- em Brazil in both the Pacific and Atlantic Oceans ( Vaz Ferreira, 1982; CrespoM. Most investigations on South American sea lions have been focused on social behavior, breeding biology, and population dynamics (Vaz Ferreira, 1982; Campagna and Le Boeuf, 1988; Cappozzo et al., 1991; Crespo and Pedraza, 1991; CrespoM. Recent research has pro- vided new data on the interactions between South American sea lions and fisheries (Crespo et al., 1994, 1997), on population size and trends (Reyes et al., 1999; Dans et al.^), and on diving behavior of lactating females (Werner and Campagna, 1995). South American sea lions are con- sidered opportunistic and broad- spectrum feeders, feeding on fish, squid, crustaceans, and occasionally on sea birds (Vaz Ferreira, 1982; George-Nascimento et al., 1985; Crespo et al., 1997). In Chilean waters, their most important prey are the Patagonian grenadier iMac- ruronus magellanicus ) and the king- clip (Genypterus spp.); no relation- ship has been found between pred- ator and prey sizes (George-Nasci- mento et al., 1985). For Patagonian waters, only preliminary informa- tion has been published about feed- ing (Crespo et al., 1997). Several studies (Hamilton, 1934; Vaz Fer- ' Crespo, E. A. 1988. Dinamica pobla- cional del lobo marino del sur Otaria flavescens (Shaw, 1800) en el norte del lito- ral patagonico. Doctoral thesis, Facultad de Cs. Exactas y Naturales, Universidad Nacional de Buenos Aires, Buenos Aires, Argentina, 298 p. - Dans, S., E. A. Crespo, S. Pedraza, R. Gon- zalez, and N. Garcia. 1996. Estructura y tendencia de los apostaderos de lobos marines de un pelo iOtaria flai-escens) en el norte de Patagonia. Informes Tecnicos del Plan de Manejo Integrado de la Zona Costera Patagonica. Fundacion Patago- nia Natural (Puerto Madryn, Argentina) 1.3:1-21. Koen Alonso et al : Food habits of Otaria flavescens 251 Table 1 Number of South American sea lion stomachs analyzed in our study by period of time sampling. The number of empty stomachs is given in parentheses. geographical area, sex, and source of Period Area Males Females Dead on shore Caught incidentally Dead on shore Caught incidentally 1982-87 1990-98 Northern Patagonia Central Patagonia Northern Patagonia Central Patagonia 6(1) 13(4) 1(1) 8(0) 3(1) 1(0) 20(4) 4(0) 3(0) reira, 1982; George-Nascimento et al., 1985) have reported the presence of gastrohths in this species, and others (Taylor, 1993) have suggested functions for them (food processing, buoyancy control, elimina- tion of internal parasites, and alleviation of hunger). The marine ecosystem in Patagonia supports one of the most intense fisheries in the world, with approximately one million tons of catch per year during the 1990s (Anonymous, 1996). South Ameri- can sea lions are reported to be caught incidentally in the trawl fisheries for Argentine hake iMerliiccius hubbsi) and Argentine red shrimp iPleoticus miiel- lerl) (Crespo et al., 1994, 1997). Some of the target and bycatch species of these and other fisheries, such as the Argentine shortfin squid (Illex argentinus) fishery, are consumed by South American sea lions (Crespo et al., 1997). The development of the fisher- ies may have been one factor that slowed the recov- ery of the South American sea lion population after harvesting of this marine mammal species ended in the mid 1960s (Crespo and Pedraza. 1991). The objectives of this research were to describe the diet of South American sea lions off Patagonia, to evaluate some hypotheses on the function of gastro- hths, and to explore the possibility of trophic compe- tition with commercial fisheries. southwestern Atlantic Ocean • Rookeries o Sampling sites Materials and methods Sample studied The total sample was composed of 59 stomachs from 28 males and 31 females, obtained in the periods 1982-1987 and 1990-1998 (Table 1). Most of the 1982-1987 sample (?! = 10) was collected in the north- ern area of Patagonia (Fig. 1). These animals were found dead on shore. The 1990-1998 sample in^AQ) was collected in the northern and central Patagonian areas (Fig. 1). During this period, the animals were Figure 1 Study area, showing the two geographical areas considered in this work: northern and central Patagonia. The filled circles indicate the location of rookeries of South American sea lion and the empty circles indicate sampling sites. obtained from two sources: shores where animals were found dead ( 14 males and 24 females ) and fisher- ies v/Iiere animals were caught incidentally (8 males and 3 :. 'Table 1). Samples were clumped 252 Fishery Bulletin 98(2) into geographical areas based on the spa- tial distribution of rookeries (Fig.l) and the reported animal movements between rookeries in northern Patagonia (Crespo and Pedraza, 1991; Crespo^; Dans et al.^). Sex and standard length (SL, cm) were recorded when possible. Males ranged from 114 to 243 cm SL, whereas females ranged between 102 to 196 cm SL (Fig. 2). Stomach content analysis Stomach contents were preserved in 70% alcohol or frozen at -20°C. Hard pieces were recovered by using sieves of different mesh sizes (from 0.5 to 10 mm) flushed with water and by using decantation trays. Fish oto- liths and bones, cephalopod beaks, crusta- cean exoskeletons and other hard remains were used to quantify and identify the prey species. Identification was made by using local species reference collections at the Marine Mammal Laboratory, Centro Nacio- nal Patagonico, CONICET, and available catalogues (Clarke, 1986;Mennietal., 1984; Roper et al. , 1984; Boschi et al., 1992; Gosztonyi and Kuba''). Complete and undigested elements (com- plete prey, otoliths and beaks) were measured with digital calipers. When digested and broken hard pieces were found in a stomach, the measurements for these elements were assigned from a random sample of undigested and whole parts of the same species obtained within the same stomach (Koen Alonsoetal., 1998). Size (total length ([TL]) offish and dorsal mantle length (IDML]) of squid, cm) and wet weight (W, g) of prey were estimated from hard pieces by using allometric regressions (Clarke, 1986; George-Nasci- mento et al., 1985; Koen Alonso et al., 1998) (Table 2). In those cases where regressions were not avail- able, regressions of related species were employed (Table 2). When related species regressions did not exist, weight was assigned by direct comparison with measured and weighed individuals of similar size for the same species or by weighing the fragments found in the stomach. The presence of stomach stones and parasites was recorded and all the gastroliths were weighed in each stomach. 110 130 150 170 190 210 Standard length (cm) 230 250 I Males a Females Figure 2 Length-frequency distribution of the total sample of South American sea Hons analyzed in this work. Three males and four females were not included in this figure because their standard lengths could not be obtained. •* Gosztonyi, A., and L. Kuba. 1996. Atlas de huesos craneales y de la cintura escapular de peees eosteros patagonicos. Informes Tecnicos del Plan de Mancjo Integrado de la Zona Costera Patagonica. Fundacion Patagonia Natural i Puerto Madryn. Argentina) 4:1-29. Data analysis The relative importance of prey species was evalu- ated by means of the index of relative importance (IRI) (Pinkas et al., 1971). The IRI was calculated for each prey species as IRI = i'7(N -^^WWcFO, Where %FO = the percent frequency of occurrence; %N = the percentage by number; and %W = the percentage by regression-esti- mated wet weight. This IRI is a modified version of the index where the original term of percentage by volume was replaced by the '/fW term (Koen Alonsoetal., 1998). In order to make easier the interpretation of the IRI, this index was expressed on a percent basis i'JdRI) (Cortes, 19971. Graphical representation of the diet was also employed to present some results (Cortes, 1997). Two overlap indices, the general overlap index (GO) and the specific overlap index (SO) (Petraitis, 1979; Ludwig and Reynolds, 1988), were used to examine dietary differences. These indices were selected because they are based on the same theo- retical framework, have associated statistical tests (Petraitis, 1979), and the GO presents a small bias even when the sample size is small (Smith and Zaret, 1982). Koen Alonso et a\ : Food habits of Otana flavescens 253 Table 2 Regressions used to estimate size and wet weight of prey of the South American sea lion off Patagonia, with their sample size (n). coefficient of determination (r-'). and source. Total length (XL) and dorsal mantle length (DML) are gi ven in centimeters, otolith length (OL), lower rostral length (LRL), and lower hood length (LHL) are given in m limeters, and wet weight (W) is in grams. Prey species Regressions n r^ Source Teleosts Engrauhs anchoita rL=2.36817+3.560L 79 0.70 Koen Alonso et al. 1998 W=0.0025rL 3^" 81 0.93 Koen Alonso et al. 1998 Seriolella punctata rL=-0. 1533+4. 19870L 45 0.8956 present study W=0.0182TL2 8635 22 0.8782 present study Stromateus brasiliensis rL=3.042OLi '59 51 0.98 Koen Alonso et al. 1998 W=0.0006418rL3^'" 63 0.98 Koen Alonso et al. 1998 Nemadactyhis bergi TL=-3.8317+5.67350L 56 0.9079 present study W=0.0144rL 2^301 23 0.9824 present study Merluccius hubbsi TL=1.8230Lio"2ifOZ,<15 447 0.93 Koen Alonso et al. 1998 TL=1.9840L io5ifO£>i5 693 0.91 Koen Alonso et al. 1998 W=0.00476rL '061 if OL< 15 469 0.92 Koen Alonso et al. 1998 W=0.00972rL 2 886 if OL>15 742 0.96 Koen Alonso et al. 1998 Acanthistius brasilianus TL=10.4444+1.8673OL 23 0.7816 present study W=0.0082TL 3 1-13 11 0.9654 present study Pseudopercis semifasciata ' TL=-12.9242+4.64250L 27 0.9789 present study W=0.005TL 3 2066 13 0.9807 present study Genypterus blacodes rL=-18.3696+5.63940L 45 0.7890 present study W=0.0016rL 3 2251 24 0.9783 present study Raneya brasiliensis TL=-0.7671+3.1968OL 52 0.6819 present study W=0.0022rL 3 2685 26 0.9349 present study Patagonotothen spp.- rL=-3. 33204+4. 219360L 121 0.7175 present study W=0.0013TL3 668 75 0.9729 present study Triathalassothm argentina rL=0.8116+2.775OL 4 0.4278 present study W=0.1987rL2ii23 4 0.6161 present study Paralichthys isosceles TL=-0.9035+4.6962OL 17 0.9436 present study W=0.0013TL 3 6036 8 0.9975 present study Agnathans Mixine sp. W=0.003rL 2 81089 7 0.9314 present study Cephalopods Ille.x argentinus DML=-3. 178+5. 617L//L 27 0.93 Koen Alonso et al. 1998 DML=0.08257+6.009LRL 63 0.87 Koen Alonso et al. 1998 W=0.00982OML 3 238 66 0.98 Koen Alonso et al. 1998 Loligo gahi-^ I>ML=-0.712+4.622L//L 98 0.76 Koen Alonso et al. 1998 W=om6DML 2 "^3 102 0.93 Koen Alonso et al. 1998 Octopus vulgaris'' \^- pi 82+3 03Ln(LHLi 108 — Clarke, 1986 Eledone sp. yy_pl 68+2 S.^Lni LHL i 214 — Clarke, 1986 ' These regressions were also used fo - Pmgmpes brasilianus. - These regressions were used ior Patagonotothen cornucola. ^ These regressions were also used forLoligo sanpaulensis. ■" This regression was used for Enteroctopus megalncyathiis and Octopuf! lehuelchu The GO evaluates the probabiUty of obtaining the utihzation curve of each group from the common uti- lization curve of all groups (Petraitis, 1979). This index has a minimum value which depends on the sample size and the number of prey species con- sidered. To compare the index obtained from dif- ferent samples, it was adjusted to vary between and 1 (GO^) (Ludwig and Reynolds, 1988). The null 254 Fishery Bulletin 98<2) hypothesis of a complete overlap {G0=1) can be sta- tistically tested by using the V-statistic (Ludwig and Reynolds, 1988). The SO is a pairwise nonsymmetric index which evaluates the probability of obtaining the utilization curve of one group from the utilization curve of the other (Petraitis, 1979; Ludwig and Reynolds, 1988). The probability of obtaining the utilization curve of the group / from the utilization curve of group k is denoted by SO,^. The null hypothesis of a complete overlap (S0^.=1) can be tested with the fZ-statistic (Ludwig and Reynolds, 1988). The comparisons made were 1 > geographical area of sampling (between northern Patagonia and cen- tral Patagonia), 2) period of sampling (between 1982-1987 and 1990-1998), 3) source of sampling (between dead animals on shore and entangled ani- mals in the fishery), and 4) sex (between males and females). Because entangled individuals were mostly males and were obtained only during the 1990-1998 period (Table 1), comparisons were made between entangled and nonentangled males in the period 1990-1998. The data employed for this analysis were those data on the occurrences of prey species that pre- sented an %IRI greater than 2'7c in the pooled sample. Inherent in the use of occurrences of prey species is the assumption that each prey species in a stom- ach was consumed independently. For this reason, the correlations between the prey species used in our analyses were evaluated with the Spearman rank correlation coefficient (r^) (Siegel and Castel- lan, 1995). The use of occurrences of prey species as data also increases the sample size for these com- parisons because the stomach of one sea lion usually contained more than one prey species. Differences in prey sizes consumed were tested by using the nonparametric two-sample Mann-Whitney U test in those cases where differences were detected (Siegel and Castellan, 1995). The relationship between mean length of prey in each stomach and predator SL was evaluated using the Tj, (Siegel and Castellan, 1995). This analysis was performed by using the pooled sample and analyzed by sex of predator. The function of gastroliths The role of gastroliths in eliminating stomach par- asites and the potential function of gastroliths in buoyancy control were investigated. The indepen- dence between the presence of gastroliths and the presence of parasites in the stomachs was tested with Fisher's exact test. The relationship between SL and total weight of gastroliths found in the stomach (GW) was evaluated with the Spearman rank cor- relation coefficient (r^) (Siegel and Castellan, 1995). This relationship was analyzed by considering each sex and the pooled sample. Results Prey species Forty-eight of 59 stomachs analyzed contained food remains (Table 1). Aproximately 37 prey species were identified, mostly fishes and cephalopods (Table 3). Additionally, the stomach of one female found dead on the beach contained two sponge species, tube polychetes, nudibranchs and hagfish (Mixine sp.). Because this specimen was considered sick and anomalous, it was excluded from the analysis. The collection analyzed was composed of 1449 individual prey, and the total estimated weight was 209.9 kg. Males consumed a broader trophic spectrum of 32 prey species (Table 3), dominated by Argentine hake, followed by Patagonian squid, Loligogahi, Argentine shortfin squid, "raneya," Raneya brasiliensis, and red octopus, Enteroctopus megalocyathus. Only the Argentine hake had a %IRI greater than 10%. The total number of prey found in male sea lions stom- achs was 738 and the total estimated biomass was 91.4 kg. The five most important prey represented 74.0% by number and 74.6% by weight. Twenty-nine species were found in female sea lion stomachs. The important prey were red octopus, Argentine shortfin squid, Argentine hake, "raneya," and Argentine anchovy (Engraulis anchoita) (Table 3). Only the first three species had %IRIs greater than 10%. The total number of prey found in the female stomachs was 711 and the estimated weight of this collection was 118.5 kg. The five most impor- tant prey represented 75.5%> by number and 91.3% by weight. Homogeneity of the sample Six species (Argentine hake, red octopus, Argentine shortfin squid, Patagonian squid, "raneya," and Argentine anchovy) had an %IRI greater than 2% in the pooled sample (Fig. 3). All pairwise correlations for these prey species were nonsignificant (P>0.05). Data on the occurrences of these species were used in the overlap analysis. Considering the GO, no differences in diet were found between geographical areas and between peri- ods of sampling (Table 4). However, the SO indicates differences in diet between the periods 1990-1998 and Koen Alonso et al : Food habits of Otaria flavescens 255 Table 3 Number (/! ), percent frequency of occurrence (%FO), percent number C/t A'l, percent e.stimated wet wei ght C/fW), and percent index of relative importance i'ilRI) of prey of the South American sea lion oft Patagonia. The ecological group for each species is shown in parentheses (P=pelagie, B=benthic, DP= demersal pelagic, DB=demersal benthic, NA=not assigned). Ecological groups were assigned following Angelescu ( 1982 ), Menni ( 1983 ), Menni et al. (1984), Angelescu and Prenski ( 1987 , and Bosch i et al. (1992). Prey Females Males n %N %W %F0 %mi n %N %W '7rFO %IRI Teleosts Merluccius hubbsi (DPI 101 14.2 9.0 34.6 11.9 286 38.8 52.4 72.7 68.6 Raneya brasiliensis (DB) 151 21.2 3.0 19.2 6.9 67 9.1 2.8 40.9 5.0 Engraulis anchoita (Pi 123 17.3 2.0 15.4 4.4 33 4.5 0.8 27.3 1.5 Patagonotnthen cornucola (DBl 34 4.8 0.3 7.7 0.6 40 5.4 0.4 22.7 1.4 Paralichthys isosceles (B) 18 2.5 0.7 7.7 0.4 8 1.1 2.6 18.2 0.7 Triathalassothia argentina (Bl 13 1.8 0.4 15.4 0.5 5 0.7 0.2 9.1 0.1 Genypterus blacodes (DBl 12 1.7 0.7 3.8 0.1 3 0.4 3.4 9.1 0.4 Pseudopercis semifasciata (DB) 3 0.4 0.9 3.8 0.1 3 0.4 11.4 9.1 1.1 Stromateus brasiliensis (DP) 5 0.7 0.7 3.8 0.1 5 0.7 1.4 9.1 0.2 Acanthistius brasilianus (DB) 1 0.1 0.4 3.8 <0.1 2 0.3 1.4 9.1 0.2 Seriotella punctata (DP) — — — — — 9 1.2 0.5 4.5 0.1 Iluocoetes fimbriatus (DBl — — — — — 7 0.9 0.2 4.5 0.1 Percophis brasiliensis (DBl 1 0.1 0.8 3.8 0.1 1 0.1 0.6 4.5 <0.1 Pinguipes brasilianus (DB) 2 0.3 <0.1 3.8 <0.1 1 0.1 0.1 4.5 <0.1 Unidentified fish (NA) 1 0.1 <0.1 3.8 <0.1 1 0.1 0.1 4.5 <0.1 Nemadactylus bergi (DBl — — — — — 3 0.4 0.2 4.5 <0.1 Prionotus punctatus (Bl — — — — — 1 0.1 0.1 4.5 <0.1 Trachurus picturatus (P) 1 0.1 0.1 3.8 <0.1 — — — — — Elasmobranchs (DB) 6 0.8 3.0 7.7 0.4 1 0.1 0.5 4.5 <0.1 Agnathans Mixine sp. (B) — — — — — 18 2.4 0.7 13.6 0.4 Cephalopods Enteroctopus megalocyathus (B) 54 7.6 61.7 53.8 55.0 21 2.8 9.3 27.3 3.4 Illex argentinus (DP) 108 15.2 15.7 38.5 17.5 27 3.7 8.2 54.5 6.7 Loligogahi(T)P) 18 2.5 0.2 26.9 1.1 145 19.6 1.9 40.9 9.1 Octopus tehuelchus (B) 6 0.8 0.1 11.5 0.2 9 1.2 0.4 22.7 0.4 Eledone sp. (B) 6 0.8 0.1 7.7 0.1 18 2.4 0.3 13.6 0.4 Loligo sanpaulensis (DP) 3 0.4 <0.1 7.7 0.1 1 0.1 <0.1 4.5 <0.1 Semirossia tenera (DBl 2 0.3 <0.1 3.8 <0.1 3 0.4 <0.1 4.5 <0.1 Crustaceans Crabs (Bl 9 1.3 <0.1 15.4 0.3 4 0.5 <0.1 13.6 0.1 Pleoticus muellen (DBl 3 0.4 <0.1 3.8 <0.1 6 0.8 0.1 9.1 0.1 Amphipods (Pi 12 1.7 <0.1 3.8 0.1 — — — — — Munida subrugosa (Bl 2 0.3 <0.1 3.8 <0.1 1 0.1 <0.1 4.5 <0.1 Muni da gregana ( B 1 — — — — — 2 0.3 <0.1 9.1 <0.1 Sero/issp. (Bl — — — — — 2 0.3 <0.1 9.1 <0.1 Peisos petrunkevitchi (DB) — — — — — 5 0.7 <0.1 4.5 <0.1 Heterosquilla platensis (B) 1 0.1 <0.1 3.8 <0.1 — — — — — Polychetes Eunice argentinensis (B) 8 1.1 <0.1 3.8 0.1 — — — — — Tunicates Pedicle tunicate (Bl 7 1.0 <0.1 7.7 0.1 — — — — — 1982-1987 (Table 4). The Argentine anchovy was not found in the small sample for the period 1982-1987, and was the least important of the six prey species selected for the overlap analysis. When the analysis was performed excluding this prey species, no differ- ences were found between periods in either of the 256 Fishery Bulletin 98(2) two overlap indices (Table 4 ). No differences were also found between entangled and nonentangled males in the period 1990-1998 (Table 4). When the difference in feeding between sexes was analyzed, no differences were found with the GO, but the SO indicated significant differ- ences in diet (Table 4). Taking into account that the GO analyzes the differences between the uti- lization curves of the groups with reference to a common utilization curve, whereas the SO ana- lyzes one utilization curve with respect to the other one, our data suggest that there are some differences in the diet between sexes. Prey size No differences were found in the sizes of Ar- gentine hake (^7=13,785.5; ",„„,,,=286; n/,,„„/„= 101; P=0.496), Argentine shortfin squid ([/= 1,331.5; n,„„,,,=27; n/„„„,^.,= 108; P=0.486), and Argentine anchovy ({7=1,627; /3„,„,„=33; "/„„„,,,= 123; P=0.080) consumed by the two sexes. Sev- enty-four percent of the Argentine hake eaten by sea lions were less than 30 cm TL. Argentine shortfin squid consumed by sea lions had a DML greater than 15 cm, whereas Argentine anchovy consumed by sea lions were mostly between 12 and 17 cm of TL (Fig. 4). Red octopus consumed by males weighed significantly less than those consumed by females ({7=194. 5; n„,„,,. =21; /!,-„,„., =54; •^ ' /?? a / (.'.s ^females ' P<0.0001). Patagonian squids consumed by males were larger than those consumed by females ( {7=880; ",„«/.,= 145; /v,„,„,,,= 18; P=0.024), but the range of DMLs of squid consumed by females was gi-eater than that of squid eaten by males. Larger "raneya" was consumed by male sea lions than "raneya" eaten by female sea lions ( {7=2,681; '!,„„/,.,=67; 'ife,„„i,,s=^^^< P<0.0001)(Fig. 5). No relationships were found between mean length of prey and predator SL with the pooled sample (r,=0.007; n=37; P=0.964), with males only (r^=0. 004; n = 19; P=0.985) or with females only (r =0.118; n = 18; P=0.641). Gastroliths Of the stomachs analyzed, 60.4Vf had gastroliths and 87. 97^ had parasites. The parasites found in the stomachs were mostly nematodes. The presence of parasites and gastroliths was independent (Fisher exact test; P=0.999 ). A positive correlation was found between the SL of South American sea lions and GW (r^.=0.572; n=45; P<0.0001) (Fig. 6). Gastroliths were found in 56.7'7f of females sea lions and 90.07r of females had parasites. The pres- -T"" i 1 j, 1 '■!•■-.. 100 r"'\ J ......| f 1 ■■■••■)•..... J "\-...,^ ..--'"" : ^-i : ,.-< - ^''■. ..<■••""' : '■■•. : : '•■-■. : "^ 80 f ...i "' i >.._ : '•■.. ; ; ..■■•-""1 ...■■-*'. f"-. "'"■. 60 1 \ "n.j ■■•••■... ^ ' -5*.. : :"-.. '■■■., O 40 - ; 1 ■■•..! : '■■••,.. 1 'S 1 i H ..X 1 ""k I "H.._ \ . ■■! 1 ...•■<•' N 1 ■•■■•4 20 i....- T-.. .5..--;: K ■■""'■■ ..■'■■.■''"# .-.■■""'■••- "U ""■% •ijaN. ;:i;;;- "f--^.... .'■■•<■■■■"' "y tf>\ i '■:>.vr:i^r--.. ....■•■■■< \,^^-^^^ X-' ■•;, ^:. ,--•■: >>-'-''^<*?5 ^^ "^ X f 1' ^^-"''^ ^ «> Figure 3 The diet of South American sea Hons. The axes are '72'^^r, and they were H = Argentme hake, = red octopus, S = Argentine shortfin squid, P = Pata- gonian squid. R = "Raneya," and A = Argentine anchovy. ence of gastroliths and parasites was independent (Fisher exact test; P=0.5645). A positive correlation was found between SL and GW (r^-0.66; n=23; P=0.0006). Gastroliths were found in 85.7% of the male sea lion stomachs, and 64.3'^ of males had parasites. The independence between the presence of gastro- liths and parasites could not be rejected (Fisher exact test; P=l ), and a positive correlation was found between SL and GW (/•^,=0.599; «=22; P=0.0032). Discussion The samples analyzed was collected over a broad range of time and space and were derived from two sampling sources. For that reason several subsam- ples were tested for homogeneity before a sample could be considered to be representative of the diet of South American sea lions. Those subsamples were tested by means of the GO and SO indices because they have associated statistical tests. The GO has only a small bias related to the difference in sample sizes even when the total sample size is relatively small (Smith and Zaret, 1982), and both overlap measures do not change if the resources are divided Koen Alonso et al : Food habits of Otana flavescens 257 Table 4 Diet overlap analyses jetween the major sources of variation in the sample studied. GO = genera! overlap index; G0„ = adjusted general overlap index V = the statistic to test the null hypothesis that GO = 1; df = degrees of freedom; P = probability of the statistic; SO,^ = specifi c overlap of group / onto group k: U = statistic to test the nu 11 hypothesis that SO,^ = 1. The num ber of prey occurrences in each category are indicated in parentheses. Source of variation General overlap index GO G0„ V df P Period of time 0.987 0.953 2.870 5 0.720 Period of time' 0.997 0.990 0.585 4 0.965 Geographic area 0.986 0.971 2.989 5 0.702 Source of sampHng 0.983 0.966 1.715 5 0.887 Sex 0.970 0.940 6.514 5 0.259 Source of variation Specific overlap index I k SO,, U df P Period of time 1982-1987(11) 1990-1998(96) 0.869 3.078 5 0.688 1990-1998(96) 1982-1987(11) 0.197 311.532 5 <0.001 1982-1987' (11) 1990-1998' (86) 0.971 0.658 4 0.956 1990-1998' (86) 1982-1987' (11) 0.969 5.367 4 0.252 Geographic area central Patagonia (38) northern Patagonia (69) 0.942 4.545 5 0.474 northern Patagonia (69) central Patagonia (38) 0.937 8.951 5 0.111 Source of sampling nonentangled males (29) entangled males (22) 0.932 4.066 5 0.540 entangled males (22) nonentangled males (29 0.933 3.037 5 0.694 Sex females (49) males (58) 0.870 13.606 5 0.018 males (58) females (49) 0.894 13.029 5 0.023 ' Analysis performed excluding the Argentine anchovy. into classes below the organism's level of discrimi- nation (Petraitis, 1979). However, one cannot eval- uate the interaction between sources of variation with these analyses. For this reason, as in any sta- tistical test, the lack of a difference for a particular source of variation does not indicate the absence of a difference, it means only that there is not enough information to affirm that this source of variation is significant. The marine communities for both areas in this study comprise the same species and are considered to be relatively homogeneous along the entire geograph- ical areas range covered by the study area (Fig. 1) (Menni, 1983; Menni and Lopez, 1984; Angelescu and Prenski, 1987). In addition, the trend of sea lion stocks in northern and central Patagonia has fol- lowed a pattern of decline and recovery during recent decades. The northern Patagonia stock declined to one fifth of its original population size between 1930 and the mid-1960s as a consequence of intense har- vesting. Then it remained stable between 1972 and 1990, when it began to recover (Crespo and Pedraza, 1991 ). The central Patagonia stock had not been har- vested but declined in a similar proportion during the mid-century and has shown evidence of popula- tion increase in recent years — a fact that suggests a strong connection and interchange of individuals between these regions (Reyes et al., 1999). In addi- tion, preliminary results based on electrophoresis analysis of blood enzymes and proteins codified by 10 loci have indicated that nine of these loci are mono- morphic, suggesting that South American sea lions from both areas have had a strong gene flux between them and possibly constitute an unique biological population (Crespo"*). Regarding the two sampling periods, the differ- ences in the diet detected with the SO are surely associated with the absence of Argentine anchovy in the 1982-1987 samples. The SO indicated that the utilization curve of food resources of the period 1990-1998 could not be drawn from the utilization ^ Crespo, E. A. 1998. Laboratorio de Mami'feros Marines, Centre Nacional Patagonico ( CONICET). Boulevard Brown 3600, (9120) Puerto Madryn, Chubut, Argentina. Personal commun. 258 Fishery Bulletin 98(2) 3 •g > •5 c 140 130 120 110 100 90 80 70 60 50 40 30 20 10 u c u 3 IT V Argentine hake I commercial size ^ 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Total length (cm) 80 Argentine shortfin squid 70 1 commercial size — ► 60 50 40 30 10 , . J n 1 15 20 25 30 35 Dorsal mantle length (cm) Argentine anchovy 40 45 50 45 40 35 30 25 20 15 10 5 1^ 1 tr-TVT ^ 10 11 12 13 14 15 16 Total length (cm) 18 19 20 Figure 4 Repression-estimated size-frequency distribution of Argentine hake (Mer- luccius hubbsn. Argentine shortfin squid illlex argentinus), and Argentine anchovy iEngraulis anchoita) consumed by South American sea lions off Patagonia collected in our study. The arrows indicate commercial sizes for Argentine hake and Argentine shortfin squid. All the sizes for Argentine anchovv are commercial. curve in the period 1982-1987. Never- theless, when the Argentine anchovy is excluded from the analysis, no differ- ences were found between the two peri- ods in either at the two overlap indices. Probably, the absence of Argentine anchovy in the period 1982-1987 is related to the small size of the sample collected and to the fact that Argen- tine anchovy was the least important prey species among the six prey spe- cies selected for the overlap analysis (Fig. 3) because the available infor- mation indicates that the abundance of Argentine anchovy was high and roughly constant during the entire study period (Ciechomski and San- chez, 1988; Pajaro et al.5). No differences in diet were detected when shore and entangled individu- als were analyzed, even though sev- eral sources of bias could be operating at the same time. The diet informa- tion obtained from individuals found dead onshore could be biased, depend- ing on the degree of digestion of the stomach contents. In our study, the use of several (and complementary) hard remains, such as otoliths and bones, allowed us to reduce this source of bias. Even when a small otolith was totally digested, the fish bones (mostly skull bones) permitted us to identify and quantify these prey. Thus, estimat- ing prey size by regressions avoided underestimating the importance of small or highly digested prey. In some cases, when we used the regression of a related species, the results were likely partially biased. Samples from entangled sea lions probably did not hold any of these biases because all the stomach contents were presum- ably composed of fresh materials. The source of bias in this case would have been due to individuals that fed inside the net. Nevertheless, a comparison ^ Pajaro, M., R. Sanchez and G. Macchi. 1997. Evaluacion de la biomasa de adultos de.sov- antes de la poblacion nortefia de anchoita {Engraulis anchoita I en el periodo 199.3-1996. Abstracts of the XII Simposio Cientifico- Tecnologico de la Comision Tecnica Mixta del Frente Maritimo, Montevideo, Uruguay, November 12-14. 1997, 4 p. Koen Alonso et al : Food habits of Otaria flavescens 259 of dead, beached males with entan- gled males did not show any signifi- cant difference. The only difference in diet found between subsamples was that shown by sexes. In this case differences in behavior and feeding habits could reflect differences in diet. Even if each sample source may have had differ- ent potential biases and the overlap analysis between them did not detect differences, we consider our sample to be, even with its limitations, a reason- able approximation of the diet of South American sea lions in Patagonia. The diversity of prey species (Table 3) found in the diet of the South American sea lion indicates that it is a broad-spectrum predator. Some of these prey species (Argentine hake, Argentine shortfin squid, and Ai'gen- tine anchovy) are abundant key spe- cies in the Patagonian continental shelf ecosystem and have commercial value (Angelescu, 1982 ;Angelescu and Prenski, 1987; Brunetti, 1990; Bezzi etal., 1994). Argentine hake and Argentine short- fin squid are the two major target spe- cies of the Argentine fleet (Anonymous, 1996), and Patagonian squid is also exploited in the Falkland (Malvinas) Islands (Hatfield, 1996). Sea lions ate these prey species at both commercial and noncommercial sizes (Fig. 4). The fishery catches Argentine hake with length modes between 35 and 40 cm TL (Caiiete et al., 1986), and the minimum commercial size of this species is 30 cm TL. Mostly noncom- mercial sizes of Argentine hake (less than 30 cm TL) were consumed by South American sea lions (Fig. 4). Estimates of stock number by age by using virtual population analysis indi- cated that the most abundant Argen- tine hakes are those of age-1 and age-2 year classes (approximately 30 cm or less in TL), which represent around 56% in number of the estimated hake stock (Bezzi et al., 1994). These data indicate that sea lions are feeding on this species according to prey-size dis- tribution and availability in the envi- ronment. Hakes smaller than 10 cm Red octopus 8000 7000 6000 3 5000 f" 4000 - 3000 u > ^ 2000 1000 20 a 16 I 12 I 8 <§ ' 32 28 a 24 20 % '6 12 8 I Male Sea lions Female Sea lions Patagonian squid Male Sea lions Female Sea lions Raneya Male Sea lions Female Sea lions Min-Max 25%-75% Median value Figure 5 Box plots of regression-estimated size of red octopus (Enteroctopuf; megalo- cvathiis). Patagonian squid iLoligo gahi), and "raneya" (Raneya hrasilwn- sis) consumed by male and female South American sea lions collected in our study. 260 Fishery Bulletin 98(2) TL are not caught by either sea Hons or the fishery because of the pelagic behavior offish in this size range (Angelescu and Prenski, 1987). The commercial squid species found in the diet of sea lions (Argentine shortfin and Pata- gonian squid) form schools of restricted size range ( Brunetti and Ivanovic, 1992; Hatfield, 1996). Therefore, it is difficult to determine if the consumed sizes represent the environ- mental availability of these prey species. The commercial size of shortfin squid is approxi- mately 15 cm DML, and this species was con- sumed by sea lions at commercial sizes (Fig. 4). Patagonian squid was consumed mostly at noncommercial sizes (less than 10 cm DML) because that most of the squid catches in the Falkland fishery were between 10 and 15 cm DML (Hatfield, 1996). These results indicate some overlap be- tween the South American sea lion diet and fishery catches but not enough to conclude that competition exists with the fishery. The population of South American sea lions in northern Patagonia has been increasing during recent years at a rate of increase greater than 39( (Crespo and Pedraza, 1991; Dans et al.'). There is no indication that fishery catches affect the availability of food for sea lions at the present time. More detailed estimates of food consumption by South American sea lions are needed. Also, estimates of the catch and bycatch of the fishery are needed to evaluate conclusively the existence of ecological competition. Differences in the diet between sexes are probably associated with different utilization of common and frequent food resources, suggesting some kind of dif- ferential feeding behavior between the sexes. The South American sea lion is a dimorphic and poly- gamous species. Therefore each sex must have dif- ferent ecological constraints. Adult female feeding trips last for about three days during reproductive (Cappozzo et al., 1991) and nonreproductive (Reyes and Crespo^) seasons. Nursing pups may limit the distance that females can travel to feeding ground. Males are not restricted by nursing pups and their feeding trips seem to be less constant (Reyes and Crespo^). There is also some evidence obtained from sightings from fishing vessels that males move far- ther offshore than do females which remain closer 50 100 150 200 250 Standard length of sea lions(cm) 300 • Maks o Females Figure 6 Scatterplot of total weight of gastroliths versus standard length for South American sea lions, by se.xes. * Reyes, L. M,, and E. A. Crespo. 1993. Variaciones diarias y lunares y viajes de alimentacion en el lobo marino del sur Otarin ftavescens en el norte de Patagonia. Abstracts of the Jornadas Nacionales de Ciencias del Mar "93," 19-25 September 1993, Puerto Madryn, Argentina, 156 p. to the coast (Crespo et al., 1997). Thus, differences in the diet could be associated with different feeding grounds or different home ranges between the sexes. The prey of female South American sea lions were more evenly distributed within ecological groups than the prey of males (Fig. 7). The prey of females were mostly benthic and demersal-pelagic species. The mean dive depth recorded by lactating females in Patagonia was 60.9 m, and GdVc of the dives were flat-bottomed and U-shaped ( Werner and Campagna, 1995) — data that agree with the bottom and coastal feeding behavior suggested by the stomach contents. On the other hand, the most important prey of males were demersal-pelagic species (Fig. 7). The Patagonian squid spawns in shallow waters, and the new generation migrates offshore to feed, grow, and mature (Hatfield, 1996). This migration pattern implies that small Patagonian squid must be more abundant in shallow, coastal waters than in deeper, offshore areas, but their size range in the coastal area must be broader than that in offshore areas because mature (and large) squids return to shallow waters to spawn. The consumption of larger Patagonian squid by male sea lions and the broader range size of Patagonian squid eaten by female sea lions agree with the hypothesis that females feed in more coastal and shallower waters than do males. The red octopus lives mostly in caves on rocky bot- toms (Re, 1998) and is the most important prey spe- cies of female South American sea lions. Red octopus reach maturity around 120 mm DML and 850 g of total weight (Re, 1998); male sea lions consumed Koen Alonso et al : Food habits of Otaria fiovescens 261 ..-'■. .•■■■"!' '>■■.. .., " '■■ ■■>•■. .-■■f' ..4-.. i ■■■■•.. ...•f*'' ...■■■!•■■ ...r'Df .1 1 i '>■. ..•,■■■■■ i .I--- ■■■" 1 "i--...i "■■>••■•.. ,0« ••■ ,.....-i-- [.....f-- ■■■•■v.. ""'!-•■. i "i ,00 ..••'"* .••■? -i .••*-. : ■'<.. : ...••I"" ...j-"' .■•^•. :' '"' ■•. : '*•: '- 8" " ..•••■••f" l.--t"" '"■■••-.. "''--. ''! »» : .- ***"* : ' \ '■•■. 'i--. '^■ «» ■■■■•■■\....|--''1',...p'"-----.... '■ '"*■ 6» ■•■"■' ..'■•••"Bi . ■■"'\x-.. "■■•■•I. i i« ..■■•I"" \ ...k:' pft^-.. ■•-.. i l» ;; ■■•■"i....r" >... r-i.. •-. .•■■•1' !i ..-'x i >:. i ■■■>-. 1 ...-K. ..^^K >-.. ■■■■•. ''* jO ■fi ...••<.. i X. \ :iMC >.. '■■•^ <.. ...>C .!>H ^-^ !>-. .... -k. .>< ;;::4::; « ;><■ '•> ■^ \.-' •■•■■'NA-v-'''^*' 'P'^^J.''^ -iP"\i^^,X'-g> ° o « o Males Females Figure 7 Ecological groups in the diet of male and femal ? South American sea lions in our study. '7fF0 = percent frequency of occurrence, "^JN = percentage by number, and "rW = percentage by regression | estimated wet weight. P = pelagic. B = benthic. DP = demersal pelagic, DB = demersal benthic, NA = not assigned. mostly immature individuals whereas female sea lions consumed mature ones. If females search for prey on rocky bottoms, they will catch the mature and larger red octopus. Instead, if male South Amer- ican sea lions feed mostly in the water column near the bottom, they could catch the younger red octopus when they are actively moving on the bottom. Younger octopus could be more vagrant than adults, and in some species, posthatching octopuses do exhibit pelagic behavior (Boletzky, 1977). The difference in the size of red octopus that South American sea lion consumed could be associated with this characteris- tic in prey biology, thus supporting the hyphothesis of different feeding behavior between the sexes. In regard to gastrolith function, the hypothesis of buoyancy control has been postulated with the evi- dence of the presence of gastroliths in several living and extinct tetrapods that swim using their limbs in the form of an underwater fly (Taylor, 1993). Taylor (1993), in an extensive comparative study, demon- strated that there is no correlation between the pres- ence of gastroliths and diet, but he found a correlation between gastroliths and underwater flying. Another explanation for the ingestion of stones is that gastro- liths "grind up" parasitic worms that usually infest seals (Riedman, 1990). The role of stomach stones seems to be better explained as "buoyancy control" than as "elimination of stomach parasites" because of the independence between the presence of para- sites and gastroliths, and the significant correlations between predator size and gastrolith weight. Gas- troliths found in the South American sea lions could be considered "ballasts" that allow the sea lions to regulate their buoyancy. Moreover, the gastroliths can also be quickly swallowed and vomited, allow- ing sea lions to change buoyancy according to their needs (Harrison and Kooyman, 1968). In summary, these stomach content analyses indi- cate that South American sea lions feed primarily on demersal and benthic species and, in general terms, use resources according to their environmen- tal availability. Males and females appear to have different constraints in their feeding behavior and these restrictions could lead females to feed in more coastal and shallower waters than those where males feed. These potential differences in feeding grounds or home ranges, or both, could explain the observed differences in diet between the sexes. Acknowledgments The authors wish to thank Pablo Mariotti, Barbara Heron Vera, Nancy Mora, Pablo Nepomnaschy and Laura Reyes for their help with the stomach contents analysis; Silvana L. Dans, Pablo Yorio and Guillermo Harris for their critical reading and useful comments on earlier versions of the manuscript; and all of the 262 Fishery Bulletin 98(2) fishermen and wildlife wardens for their help aboard ship and in the field. Three anonymous reviewers (especially no. 817) and Sarah Shoffler made impor- tant and useful comments that enhanced the anal- ysis of results and the expression of the ideas. As always, Sharyn Matriotti gave us important help with this article. Institutional support was provided by Centro Nacional Patagonico (CONICET), Univer- sidad Nacional de la Patagonia, Fundacion Patago- nia Natural, Prefectura Naval Argentina, and the government of Chubut Province. The fishing compa- nies Harengus S.A. and Alpesca S.A. collaborated with the authors in the retrieval of entangled sea lions and gave their permission for work onboard their fishing vessels. This work was carried out with the financial support of the National Geographic Soci- ety (Grant 5548/95 to E.A. Crespo and A.C.M. Schia- vini), the Whale and Dolphin Conservation Society, the Patagonian Coastal Zone Management Plan (GEF/UNEP FPNAVCS) and the Programa de Coo- peracion Cientifica con Iberoamerica (1996-1998). Literature cited Angelescu, V. 1982. Ecologia trofica de la anchoita del Mar Argentino (Engraulidae, Engraulis anchoita). Parte II. AJimentacion, comportamiento y relaciones troficas en el ecosistema. INIDEP ( Institute Nacional de Investigacion y Desarrollo Pesquerol, Contribution 409, 83 p. Angelescu, V., and L. B. Prenski. 1987. Ecologia trofica de la merluza comiin del Mar Argen- tino (Merlucciidae, Merluccius hubbsi). Parte 2. Dinamica de la alimentacion analizada sobre la base de las condicio- nes ambientales, la estructura y las evaluaciones de los efectivos en su area de distribucion. INIDEP. Contribu- tion 561, 205 p. Anonymous. 1996. Captura 95. Redes 88:24-40. Bezzi, S., G. Canete, M. Perez, M. Renzi, and H. Lassen. 1994. Report of the INIDEP working group on assessment of hake (Merluccius hubbsi) north of 48'S (Southwest Atlan- tic Ocean). INIDEP Doc. Cient. 3:5-28. Boletzky, S. V. 1977. Post-hatching behavior and mode of life in cephalo- pods. Symp. Zool. Soc. Lond. 38:557-567 Boschi, E. E., C. E. Fischbach, and M. I. lorio. 1992. Catalogo ilustrado de los crustaceos estomatopodos y decapodos marinos de Argentina. Frente Maritimo 10 (sec. A):7-94. Brunetti, N. E. 1990. Evolucion de la pesqueria de lUe.x nrgentinus (Castel- lanos, 19601. Inf. Teen. Inv. Pesq. 1.55. 19 p. Brunetti, N. E., and M. L. Ivanovic. 1992. Distribution and abundance of early life stages of squid illlex argentmus} \n the south-west Atlantic. ICES .] Mar. Sci. 49:17.5-183. Campagna, C, and B. J. Le Boeuf. 1988. Reproductivebehaviorof Southern sea lions. Behav- iour 104:2.3.3-261. Canete, G. R., R. G. Perrotta, and J. A. Perez Comas. 1986. Aspectos comparativos entre muestreos de desem- barque y muestreos en alta mar de la especie merluza (Mer- luccius hubbsi), considerando el descarte efectuado por la flota marplatense en el periodo julio 1979-Setiembre 1980. Publ. Com. Tec. Mix. Fr. Mar. 1:447-453. Cappozzo, H. L., C. Campagna, and J. Monserrat. 1991. Sexual dimorphism in newborn Southern sea lions. Mar. Mamm. Sci. 7(41:385-394. Clarke, M. R. 1986. A handbook for the identification of cephalopods beaks. Clarendon Press, Oxford, 273 p. Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elas- mobranch fishes. Can. J. Fish. Aquat. Sci. 54:726-738. Crespo, E. A., J. F. Corcuera, and A. Lopez Cazorla. 1994. Interactions between marine mammals and fisheries in some coastal fishing areas of Argentina. In Gillnets and cetaceans (W. F Perrin, G. P. Donovan, and J. Barlow, eds.), p. 269-281. Rep. Int. Whal. Comm. (special issue 15). Crespo, E. A., and S. N. Pedraza. 1991. Estado actual y tendencia de la poblacion de lobos marinos de un pelo lOtaria flavescens) en el litoral norpa- tagonico. Ecologia Austral 1:87-95. Crespo, E. A., S. N. Pedraza, S. L. Dans, M. Koen Alonso, L. M. Reyes, N. A. Garcia, M. Coscarella, and A. C. M. Schiavini. 1997. Direct and indirect effects on the high seas fisheries on the marine mammal populations in the northern and central patagonian coast. -J. Northwest Atl. Fish. Sci., 22:189-207. de Ciechomski, J. D., and R. P. Sanchez. 1988. Analisis comparativo de las estimaciones de biomasa de la anchoita (Engraulis anchoita) en el Atlantico Sudoc- cidental en diferentes aiios y con distintas metodologias. Publ. Com. Tec. Mix. Fr. Mar. 4:117-132. George-Nascimento, M. F., R. A. Bustamante, and R. C. Oyarzun. 1985. Feeding ecology of the Southern sea lion Otaria flave- scens Shaw. 1800: food contents and food electivity. Mar. Biol.Prog.Ser. 21:135-143. Hamilton, J. E. 1934. The Southern sea lion Otaria byronia (de Blainville). Discovery Rep. 8:269-318. Harrison, R. J., and G. L. Kooyman. 1968. General physiology of the Pinmpedia. In The behav- ior and physiology of pinnipeds (R. J. Harrison. R. C. Hubbard, R. S. Peterson, C. E. Rice, and R. J. Schus- terman, eds.), p 221-296. Appleton-Century-Crofts, New York. NY. Hatfield, E. M. C. 1996. Towards resolving multiple recruitment into loliginid fisheries: Loligogahi in the Falkland Islands fishery. ICES J. Mar Sci. .53:565-575. Koen Alonso, M., E. A. Crespo, N. A. Garcia, S. N. Pedraza, and M. Coscarella. 1998. Diet of dusky dolphins, Lagenorhynchus obscurus. in waters off Patagonia, Argentina. Fish. Bull. 96:366-374. Ludwig, J. A., and J. F. Reynolds. 1988. Statistical ecology. John Wiley and Sons, New York, NY, 337 p. Menni, R. C. 1983. Lospecesenel mediomarino. Estudio Sigma S.R.L., Buenos Aires, Argentina, 169 p. Koen Alonso et al : Food habits of Otana flavescens 263 Menni, R., and H. L. Lopez. 1984. Distributional patterns of Argentine marine fishes. Physis 42 (issue 103, sec. A):71-85. Menni, R. C, R. A. Ringuelet, and R. A. Aramburu. 1984. Feces marinos de la Argentina y Uruguay. Editorial Hemisferio Sur S.A., Buenos Aires. Argentina, 359 p. Petraitis, P. S. 1979. Likehood measures of niche breadth and overlap. Ecology 60(41:703-710. Pinkas, L., M. S. Oliphant, and I. L. K. Iverson. 1971. Food habits of albacore, bluefin tuna and bonito in California waters. U.S. Dep. Fish. Game, Fish Bull. 152:1-105. Re, M. E. 1998. Pulpos octopodidos (Cephalopoda, Octopodidae). In El mar argentino y sus recursos pesqueros (E. Boschi, ed.), tomo 2: Los moluscos de interes pesquero. Cultivos y estrategias reproductivas de bivalvos y equinoideos, p 99-114. Instituto Nacional de Investigacion y Desarrollo