MBL U.S. Department of Commerce Volume 101 Number 1 January 2003 Fishery Bulletin U.S. Department of Commerce Donald L Evans Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atnnosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries lOOFc 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 C 15700, Seattle, WA 981 15-0070. 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: $45.00 domestic and $.56.25 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Dr. Norman Bartoo Editorial Assistant Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 981 15-0070 Editorial Committee Dr. Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr. Theodore W. Pletsch Dr. Joseph E. Powers Dr. Harald Rosenthal Dr. Fredric M. Serchuk Dr. George Watters University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service 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 101 Number 1 January 2003 Fishery Bulletin JAN 2 1 2003 Contents Articles 1 -9 Blaylock, Reginald B., Leo Margolis, and John C. Holmes The use of parasites in discriminating stocks of Pacific halibut (Hippoglossus stenolepis) in the northeast Pacific 10-21 Comyns, Bruce H., Richard F. Shaw, and Joanne Lyczkowski-Shultz Small-scale spatial and temporal variability in growth and mortality of fish larvae in the subtropical northcentral Gulf of Mexico: implications for assessing recruitment success 22-31 DeMartini, Edward E., Gerard T. DiNardo, and Happy A. Williams Temporal changes in population density, fecundity, and egg size of the Hawaiian spiny lobster (Panulirus marginatus) at Necker Bank, Northwestern Hawaiian Islands The conclusions and opinions expressed in Fishery Bullelm are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service (NOAA) or any other agency or institution- The National Marine Fisheries Service iNMFS) does not approve, recommend, or endorse any propnetary product or pro- pnetary 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 matenal 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. 32-43 Friedlander, Alan M., and David A. Ziemann Impact of hatchery releases on the recreational fishery for Pacific threadfin (Polydactylus sexfilis) in Hawaii 44-57 Hart, Deborah R. Yield- and biomass-per-recruit analysis for rotational fisheries, with an application to the Atlantic sea scallop (Placopecten magellanicus) 58-74 Hearn, William S., and Thomas Polacheck Estimating long-term growth-rate changes of southern bluefin tuna (Thunnus maccoyii) from two periods of tag-return data 75-88 Loefer, Joshua K., and George R. Sedberry Life history of the Atlantic sharpnose shark (Rhizoprionodon terraenovae) (Richardson, 1836) off the southeastern United States Fishery Bulletin 101(1) 89-99 Maunder, Mark N., and George M. Walters A general framework for integrating environmental time series into stock assessment models: model description, simulation testing, and example 100-115 Miller, Michael J., David M. Nemerson, and Kenneth W. Able Seasonal distribution, abundance, and growth of young-of-the-year Atlantic croaker (Micropogonias undulatus) in Delaware Bay and adjacent marshes 116-128 Newman, Stephen J., and lain J. Dunk Age validation, growth, mortality, and additional population parameters of the goldband snapper (Pristlpomoldes multidens) off the Kimberley coast of northwestern Australia 129-146 Ralston, Stephen, James R. Bence, Maxwell B. Eldridge, and William H. Lenarz An approach to estimating rockfish biomass based on larval production, with application to Sebastes jordani 147-167 Winship, Arliss J., and Andrew W. Trites Prey consumption of Steller sea lions (Eumetopias jubatus) off Alaska: How much prey do they require? Notes 168-174 Beerkircher, Lawrence, Mahmood Shivji, and Enric Cortes A Monte Carlo demographic analysis of the silky shark (Carcharhlnus falciformis): implications of gear selectivity 175-182 Gunderson, Donald R., Mark Zimmermann, Daniel G. Nichol, and Katherine Pearson Indirect estimates of natural mortality rate for arrowtooth flounder (Atheresthes stomias) and darkblotched rockfish (Sebastes cramerl) 183-188 Marcogliese, David J., Elaine Albert, Pierre Gagnon, and Jean-Marie Sevigny Use of parasites in stock identification of the deepwater redfish (Sebastes mentella) in the Northwest Atlantic 189-193 Polovina, Jeffrey J., Evan Howell, Denise M. Parker, and George H. Balazs Dive-depth distribution of loggerhead (Caretta caretta) and olive ridley (Lepidochelys olivacea) sea turtles in the central North Pacific: Might deep longline sets catch fewer turtles? 194-198 Smith, Susan E., Robert A. Mitchell, and Dan Fuller Age-validation of a leopard shark (Triakis semifasciata) recaptured after 20 years 199 Subscription form Abstract — The use of parasites as in- dicators of the stock structure of Pacific hahbut {Hippoglossus stenolepis) in the northeast Pacific was investigated by using 328 adult (>55 cm fork length) hahbut from 15 composite localities ranging from northern California to the northern Bering Sea and 96 ju- venile (10-55 cm) halibut from five localities ranging from the northern Queen Charlotte Islands to the Bering Sea. Counts of eight selected parasite species (the juvenile acanthocephalans Corynosoma strumostim and C. vil- losum, the metacestode Nyhelinia sur- menicola, the digenean metacercaria Otodistnmum sp., and the larval nema- todes Anisakis simplex, Pseudoterra- noi'a decipiens, Contracaecum sp., and Spirurid gen. sp.) that produce infec- tions of long duration, do not multiply in the host, and that have a relatively high abundance in at least one geo- graphic locality were subjected to dis- criminant function analysis. Juvenile Pacific halibut showed no separation and, even though they were not heav- ily infected with parasites, the analysis suggested that juveniles could be a mixed stock. Three groups of adults were identified: fish from California to the southern Queen Charlotte Islands, those from the northern Queen Char- lotte Islands to the central Bering Sea, and those from the central and north- ern Bering Sea. These groups suggest that the single stock concept be more thoroughly evaluated. The use of parasites in discriminating stocks of Pacific halibut (Hippogiossus stenolepis) in the northeast Pacific Reginald B. Blaylock Department of Biological Sciences University of Alberta Edmonton, AB T6G 2E9, Canada and Department of Fisheries and Oceans Pacific Biological Station Nanaimo, B.C. V9R 5K6, Canada Present address: College of Manne Sciences The University of Southern Mississippi 703 East Beach Blvd PO Box 7000 Ocean Spnngs, Mississippi 39566-7000 E-mail address: reg blaylock@usm.edu Leo Margolis (deceased) Department of Fishenes and Oceans Pacific Biological Station Nanaimo, B.C. V9R 5K6, Canada John C. Holmes Department of Biological Sciences University of Alberta Edmonton, AB T6G 2E9, Canada Manuscript accepted 10 July 2002. Fish. Bull. 101: 1-9(2003). ' The Pacific halibut (Hippogiossus steno- lepis ) is an Arctic-Boreal Pacific pleuro- nectid flatfish ranging throughout the North Pacific from southern California to northern Japan, but is most abun- dant in the Gulf of Alaska, The halibut supports one of the top five commercial fisheries in North America, with aver- age annual landings of approximately 25,000 metric tons from 1991 to 1995 (IPHC, 1996), and is also widely sought in the sport fishery, thus contributing significantly to west coast economies. The International Pacific Halibut Com- mission (IPHC) is responsible for man- agement of the resource. From the 1930s through the 1950s the IPHC re- cognized at least three stocks of hali- but from tagging experiments, egg and larval drift, anatomical differences, and differences in growth rate: 1) those in the Bering Sea; 2) those from the Gulf of Alaska south to Cape Spencer, Alaska; and 3) those south of Cape Spencer (Skud, 1977). These bound- aries roughly followed the zoogeo- graphic zonation in the North Pacific. Skud (1977) re-analyzed the data and concluded that there was extensive intermingling of fish among areas and that there was no evidence to indicate that fish north and south of Cape Spencer, Alaska, constituted different stocks. Available biochemical evidence (Tsuyuki et al., 1969; Grant et al., 1984), although limited in scope and by sampling effort, suggests little genetic variation throughout the northeast Pacific. As a result, the IPHC manages halibut as a single population, but with statistical divisions for management of data. Parasites have been used successfully to distinguish populations or stocks of fishes and, as a result, provide informa- tion useful in fisheries management ( see Fishery Bulletin 101(1) 180 Figure 1 Sampling localities for 328 adult (circles) and 96 juvenile (squares) Pacific halibut, Hippoglossus stenolepis, in the northeast Pacific. OR = Oregon-northern California, WA = Washington, VI = Van- couver Island, SQC = southern Queen Charlotte Islands, NQC = northern Queen Charlotte Islands, SAl = southeast Alaska site 1, SA2 = southeast Alaska site 2, KP = Kenai Peninsula, KI = Kodiak Island, NI = Nagai Island, UP = Unimak Pass, WAL = western Aleutian Islands, SB = southern Bering Sea, PI = Pribilof Island, SMI = St. Matthew Island. Individual hauls with at least 10 fish (for a total of 202 fish) are shown as solid circles. Other collection sites are shown as stippled circles. See Table 1 for sample sizes. reviews by Lester, 1990;Moser, 1991;Williamset al., 1992). With respect to flatfish, Gibson ( 1972) used parasitological data to distinguish three groups of Platichthys flesus and Krzykawski and Wierzbicka (1982) used parasitological data and other information to distinguish between Bar- ents Sea and Labrador stocks of Greenland hahbut, Rein- hardtius hippoglossoides. Khan et al. (1982) and Arthur and Albert (1993) used parasites to distinguish between Atlantic and Gulf of St. Lawrence stocks of/?, hippoglos- soides, and Boje et al. (1997) used parasites to indicate differences among Greenland stocks of Greenland halibut and stocks from the western Atlantic. No similar work on flatfishes has been done in the Pacific and, with the excep- tion of Krzykawski and Wierzbicka ( 1982) and Boje et al. (1997), there has been no attempt to distinguish between stocks of a species across a significant portion of the spe- cies' range. In this article, we use discriminant analysis on counts of some of the parasites from adult Pacific halibut to deter- mine if they form discrete groups or stocks in the north- east Pacific. We do a similar analysis on the juvenile fish and compare the results to the adult analysis to determine when separation is likely to occur Materials and methods A total of 328 adults ( >55 cm fork length) from 15 composite localities, ranging from northern California to the vicinity of St. Matthew's Island in the Bering Sea and 96 juveniles (10-55 cm) from five localities ranging from the north- ern Queen Charlotte Islands to the Bering Sea (Fig. 1), were caught by staffs of the IPHC and the U.S. National Marine Fisheries Service during the summers of 1990-92 (using longlines and trawls). Most localities (for the adult samples) included fish taken from several hauls; however, 202 fish came from 13 individual hauls, each of which con- tained at least 10 fish. Fish were bagged individually and immediately frozen at sea for later examination. Fish and parasites were processed by using standard parasitological techniques (see Blaylock et al., 1998a). We followed Bush et al.'s (1997) definitions for prevalence, abundance, and intensity. Parasites used in the analyses were chosen according to the guidelines of Arthur and Albert (1993). Only those species with infections of long duration, that do not multiply in the host, and that have a relatively high abundance in at least one geographic locality were used. Of the 59 parasite taxa identified from Blaylock et al : Use of parasites In discriminating stocks of Hippoglossus stenolepis Pacific halibut (Blaylock et al., 1998a), eight taxa met these criteria: the juvenile acanthocephalans Corynosoma strumosum (body cavity) and C. villosum (body cavity), the metacestode A^v6e/;>;(a surmenicola (stomach wall), the di- genean metacercaria Otodistomum sp. (stomach wall), and the larval nematodes Anisakis simplex (body cavity, or- gans, musculature), Pseudoterranova decipiens (body cav- ity, organs, musculature), Contracaecum sp.(body cavity), and Spirurid gen. sp. (stomach wall). A ninth taxon, the larval nematode Hysterothylaciun7 adiincum (body cavity and organs) was included for the analysis of juveniles. Because individual fish varied extensively in size (fork length), and the number of a parasite individuals was strong- ly correlated with fish size (Blaylock et al., 1998a), parasite numbers were corrected for differences in host size. Counts of individual parasites were first log-transformed (ln(A:-i-l)). To adjust for the effect offish length, a regression of the trans- formed parasite numbers on fish length for each species in each locality (and haul) was calculated. This relationship was then used to adjust the number of parasite individuals within each fish in each locality (and haul) to that expected for the average-size fish in the overall sample (80.9 cm for adults, 39.2 cm for juveniles). These data were then used in discrimi- nant function analyses. We applied the most widely used (and available) method of discriminant function analysis, in which the data were divided into training and test sets, and a dis- criminant function calculated on the training set was used to classify the test set. Interpretations were based on patterns in the test sets. To insure that any identified patterns were due to differences among localities rather than simply differ- ences among individual hauls, we performed the same analy- sis on both the locality and the individual haul data. Our training set consisted of six fish randomly selected from each haul ("haul" training set) or these fish plus four from the northern Queen Charlotte Islands and six from Unimak Pass ("locality" training set). Discriminant func- tions calculated from data on these "training" fish were used to classify each of the remaining fish from each haul ("haul" test set) or those fish plus all remaining fish ("local- ity" test set). The test set fish were first classified into one of the 13 hauls or 15 localities. Classification matrices were examined for the degree of misclassification. Hauls or local- ities were then grouped and regrouped into four and three groups based on patterns in the 13 or 15 category analyses and the zoogeographic zones from Blaylock et al. (1998b). Analyses were then repeated. Classifications were exam- ined for misclassification, and boundaries adjusted for re- testing. Results presented are those from the best fit "test" classifications. Statistical analyses were performed in SYS- TAT for Windows version 5.05 (Wilkinson et al., 19921. The entire data set from which the data for this analysis came is available for purchase from the Depository of Unpublished Data, Document Delivery, CISTI, National Research Coun- cil of Canada, Ottawa, ON KIA 0S2, Canada. Results Of the taxa that met the Arthur and Albert ( 1993) criteria, A^. surmenicola was most common and abundant in north- ern localities and fairly common and abundant in central localities. Corynosoma strumosum, although variable in prevalence and abundance, was much more common in the northernmost localities. Corynosoma villosum, although prevalent everywhere, was more abundant in northern fish. Otodistomum sp. and Spirurid gen. sp. were more common and abundant in southern localities. Anisakis simplex, although present in virtually every fish from every locality, was more abundant in southern fish. Pseu- doterranova decipiens and Contracaecum sp. appeared to be more common in central areas (Table 1). In the juve- niles, A. simplex and P. decipiens were more common in central localities, whereas C. villosum, C. strumosum, and Hysterothylacium aduncum were more common in north- ern localities (Table 2). The haul analyses indicated that the majority of fish from some hauls (12/14 Vancouver Island |VI] fish, 3/4 Southeast Alaska 1 |SA1| fish, 3/5 from the Pribilof Islands |PI|, and all 4 from St. Matthew's Island ISMI]) could be correctly classified but that fish from surrounding areas also were incorrectly classified to these hauls. Moreover, the percentage offish correctly classified by the haul func- tions was, in all cases, within only a few percentage points of that correctly classified by the equivalent locality func- tion. Thus, patterns do not appear to be associated with independent hauls. Therefore, we present only the results of the locality analyses. Fifteen category discriminant analyses revealed severe misclassification in most areas. Only 39% were correctly classified to locality (Table 3). The functions did assign correctly the majority of test fish from two localities ( 19/26 from Vancouver Island [VI] and 14/22 from the southern Bering Sea [SBl ). However, misclassification of fish from surrounding areas to these localities indicated less than accurate discrimination. The clearest indications from these analyses were that localities from the vicinity of the Queen Charlotte Islands south should be grouped together and that there is a suggestion that the two northern Ber- ing Sea locations (PI and SMI) should be grouped. Regrouping the localities into four categories by using boundaries from zoogeographic analyses (Blaylock et al., 1998b) plus the apparent northern Bering Sea grouping (PI-SMI), considerably improved the predictive ability of the functions. The "best fit" four-category grouping gave ap- proximately 62% correct classification at the locality level (Table 4). The four-category functions were good predictors for the California-Oregon (OR) to southern Queen Char- lotte Islands (SQC) fish; over 80% of these southern fish were correctly classified, and only about 6% of the other fish were misclassified to this group. Over 70% of the Pribilof-St. Matthew Island (PI-SMI) fish were correctly classified, and only 7% of the other fish were incorrectly classified to this group. There was much misclassification in the two central groups, and adjustment of the boundary between these two groups did not produce marked improvement (not shown). Grouping into three categories by combining the two central groups resulted in substantial improvement in discrimination (83% correct) (Table 5). Shifting of the boundary between the northern and central group re- vealed that discrimination broke down when the southern Fishery Bulletin 101(1) Table 1 Summary of parasites used for discriminatioi SA1= southeast Alaska 1, SA2 = southeast Al be = body cavity, o = organs, m = musculature. of stocks of adult Pacific halibut by locality. OR = Oregon-northern California, WA = aska 2, KP = Kenai Peninsula, KI = Kodiak Island. NI = Nagai Island. WAI = western sw = stomach wall. Intensity = mean number of parasites per infected host. Parasite Site Stage OR(n=23) WA(/i = 14) % Intensity % Intensity Anisakis simplex be, 0, m larva 100 258.2 ±520.2 100 122.2 ±101.0 Corynosoma villosum be juvenile 74 7.9 ±5.4 71 6.3 ±7.3 Corynosoma strtimosum be juvenile 52 5.6 ±6.4 50 6.3 ±5.2 Nybelinia surmenicola sw metacestode 17 2.8 ±2.9 14 4.0 +4.2 Otodistomum sp. sw metacercaria 44 14.3 ±13.8 36 32 ±55.7 Pseudoterranoua decipiens be, 0, m larva 44 2.5 ±2.1 21 2.3 ±1.5 Contracaecum sp. be larva 0 0 0 0 Spirurid gen. sp. sw larva 22 1.4+0.5 50 11 ±24.7 Parasite KP(n=21) KI(«=26) NI(?! = 131 % Intensity % Intensity % Intensity Anisakis simplex 100 3.3.6+22.1 100 29.5 ±27.3 100 80.3 ±62.0 Corynosoma villosum 95 11.1 ±12.4 100 11.2 ±18.4 85 12.6 ±15.0 Corynosoma strumosum 19 1 ±0.0 27 1.3 ±0.5 8 2.0 ±0.0 Nybelinia surmenicola 43 4.2 ±3.9 65 29.5 +106.6 39 42.2 ±89.0 Otodistomum sp. 14 2.0 ±1.7 4 1±0 8 1±0.0 Pseudoterranoua decipiens 29 1.5+1.2 50 1.9 ±1.4 61 2.5 ±2.0 Contracaecum sp. 62 3.0 +3.08 50 3.2 ±3.2 0 0 Spirurid gen. sp. 9 1.0 ±0 4 1 ±0.0 0 0 Table 2 Summary of parasites used for discrimination of stocks of juvenile Pacific halibut. NQC = northern Queen Charlotte Islands, NI = Nagai Island, UP = Unimak Pass, PI = Nunivak Island (central Bering Sea), be = body cavity, o = organs, m = musculature, sw = stomach wall. Intensity = mean number of parasites per infected host. Parasite Site Stage NQC (n=20) KI(f!=13) NI (n= 20) UP(fi=20) PI (n=23) % Intensity % Intensity % Intensity '7f Intensity 9, Intensity Anisakis simplex be, 0, m larva 25 1.2 8 1.0 ±0 65 3.1 ±1.9 70 3.6 ±3.1 30 1.7 ±0.8 Corynosoma villosum be juvenile 5 1 ±0.0 46 1.3 ±2.4 75 3.4 ±2.6 80 5.8 ±7.4 9 1.5 ±0.7 Corynosoma strumosum be juvenile 0 0 8 1.0 ±0 0 0 5 1 ±0.0 48 1.2 ±0.4 Hysterolhylacium aduncum be, 0 juvenile 15 1 ±0.0 92 4.4 ±3.2 95 8.7 ±10.7 85 5.4 ±8.8 74 2.8 ±1.5 Nybelinia surmenicola sw metacestode 0 0 0 0 0 0 5 1 ±0.0 26 1.3 ±0.8 Pseudoterranova be, 0, m larva 0 0 8 1.0 ±0 25 1.4 ±0.9 15 1 ±0.0 4 6 ±0.0 decipiens Contracaecum sp. be larva 0 0 15 1.0 ±0 10 5 ±5.7 0 0 0 0 Spirurid gen. sp. sw larva 0 0 0 0 5 2.0 ±0.0 0 0 0 0 Blaylock et al.: Use of parasites in discriminating stocks of Hippog/ossus slenolepis Table 1 Washington, VI = southern Vancouver Island, SQC = Aleutian Islands. SB = southern Bering Sea, PI, Pribi southern Queen Charlotte Islands, NQC = northern Queen Charlotte Islands, of Island (central Bering Sea), SMI = St. Matthew Island (northern Bering Seal. VI(/i=32) 3QC(n=31) NQC(n=8) SAl (/i=20) SA2 (n=29) % Intensity % Intensity % Intensity '7, Intensity % Intensity 100 381.4 ±357.1 100 167.8 ±101.4 100 76.1 ±47.6 100 81.8±141.1 100 44.0 ±57.0 94 8.8 ±7.4 94 13.7 ±24.3 100 5.9 ±7.4 90 10.9 ±16.6 93 16.1 ±32.0 44 3.6 ±5.2 39 1.8 ±1.5 25 2.0 ±0.9 15 1.7 ±0.6 52 1.9 ±1.0 19 16.0 ±31.9 3 1.0 ±0.0 50 13.3 ±23.8 30 11.8 ±25.1 52 4.1 ±2.0 9.4 18.3 ±30.0 39 8.3 ±13.9 38 4.7 ±2.3 35 13.0 ±11.0 21 8.3 ±11.0 16 1.2 ±0.4 13 1.5 ±0.6 25 2.0 ±0.0 25 3.2 ±4.4 69 3.1+3.0 3 1.0+0 0 0 50 3.8 ±4.9 40 3.5 ±2.2 45 2.0 ±2.0 3 1.0 ±0 0 0 0 0 15 2.3 ±1.2 10 2.8 ±2.0 UP(n=20) WAI(«=20) SB(n=29) PI(n = 14l SMI (n=28) % Intensity % Intensity % Intensity 7, Intensity % Intensity 100 53.3 ±43.8 100 41.5 ±55.0 100 40.6 ±33.6 100 21.5 ±22.3 84 10.9 ±10.0 85 29.6 ±57.4 95 13.8 ±14.0 97 16.8 ±18.3 86 34.4 ±37.0 90 34.4 ±55.9 30 2.0 ±1.5 40 2.1 ±2.0 38 8.5 ±20.2 79 9.0 ±9.3 77 23.2 ±25.4 55 2.7 ±2.1 35 16.1 ±33.8 52 6.7 ±13.1 57 34.6 ±69.2 58 25.0 +69.8 0 0 5 1±0 0 0 0 0 0 0 45 1.4 ±0.7 45 2.3 ±2.0 38 2.3 ±1.5 29 2.3 ±1.3 23 2.4 ±1.8 15 2.0 ±1.0 30 1.3 ±1.0 38 3.0 ±4.5 14 1.0 ±0 7 4.0 ±3.5 5 1.0 ±0.0 0 0 4 1 ±0.0 0 0 0 0 Bering Sea (SB) was included in the northern group (not shown). Inclusion of the northern Queen Charlotte Islands (NQC) in the southern group had little effect (81% cor- rect classification) (not shown). These analyses indicated a southern (OR-SQC) group, a central (NQC-SB) group, and a northern (PI-SMI) group. Classification into two categories (with SQC as the di- viding line) provided no substantial improvement (87'^ correct) (not shown). Inclusion of NQC in the southern group had little effect (88% correct classification). Discrimination of juveniles was poor with any organi- zation of localities. The "best fit" classification correctly classified only 66% of the fish and there was substantial misclassification among the localities (Tables 6 and 7). Fish from the northern Queen Charlotte Islands (NQC) through Nagai Island (NI) separated reasonably well, but the majority of fish from the northernmost locality were also misclassified to this group. Note that parasite num- bers and prevalences were low in the juveniles (Table 2). Discussion Our results show four things: 1) parasites clearly differen- tiate a group of southern adults; 2) parasites provide some evidence for a separation of the northernmost adults; 3) the differentiation is not always unequivocal; and 4) para- sites do not differentiate groups of juvenile fish. Skud (1977) concluded that southern and northern groups mixed extensively at all ages of their life history and that, although populations of adults may be largely discrete in the summer, any such discreteness was tempo- rary because tagging evidence suggested more extensive winter migrations associated with spawning. Our data, on the other hand, suggest that there is some merit to the IPHC's early recognition of three stocks of adult halibut. Parasite data support the existence of two major groups of halibut and suggest the possibility of a third group in the central and northern Bering Sea. The high proportion of correct classifications based on parasites suggest that these differences are well established. Recognition of three such groups is also supported by several of Skuds (1977) observations. He presented data suggesting that after fish home to spawning areas, southern and northern fish maintain reasonably separate migration circuits between feeding and spawning grounds. Data from Skud (1977) and more recent tagging data (Geernaert, 1996) also suggest that southern fish move less than their northern counterparts. Skud also recog- nized a resident population in the Bering Sea. These con- Fishery Bulletin 101(1) Table 3 Cross validation results of a 15-category discriminant function classification of adult Pacific halibut in the northeast Pacific based on parasite data. Numbers offish assigned to each locality and the corresponding percentage of the sample assigned to that cat- egory are shown. See Table 1 legend for key to abbreviations. Correct classifications are shown in bold ('289r of 240). True category Assigned category OR WA VI SQC NQC SAl SA2 KP KI NI UP WAL SB PI SMI OR 4 17% 2 9% 1 4% 2 9% 2 9% 3 13% WA 2 25% 2 3% 1 13% 1 13% 2 25% VI 1 4% 19 73% 2 8% 3 12% 1 4% SQC 1 4% 1 4% 10 40% 11 44% 2 8% NQC 2 50% 2 50% SAl 1 7% 1 7% 4 29% 1 7% 1 7% 7% 3 21% SA2 1 4% 1 4% 2 9% 1 4% 3 13% 3 13% 4% 1 4% 8 35% 1 4% 1 4% KP 7 50% 1 7%: 1 7% 1 7% 7% 2 14% 1 7% KI 7 35% 2 10% 1 5% 2 10% 5% 2 10% 3 15% 2 10% Nl 1 14% 1 14% 2 29% 1 14% 1 14% UP 2 14% 1 7% 2 14% 2 14% 1 7% 2 14% 3 21% 1 7% WAI 3 21% 1 7%' 1 7% 2 14% 2 14% 0 5 36% SB 1 7% 1 7% 4 29% 3 21% 1 7% 2 14% 1 7% 2 14% 5 36% 1 7% 2 14% PI 1 13% 3 38% 4 50% SMI 1 5% 3 14% 3 14% 1 5% 14 64% elusions pose two questions. First, do fish from different groups mix extensively? Second, do such groups represent reproductive units or stocks? Our analysis was based on a small set of larval para- sites, all of which are known to be long-lived and do not multiply in the host. Other long-lived parasites such as the myxosporeans have been used in stock discrimination but were not included here because of a lack of abundance data. However, the decreased ability to detect differences because of the small data set was offset by an increased ability to detect the host's past activities. Most of these parasites live for at least several years; therefore, the presence and abundance of these parasites may indicate where the host has been over that time period. At least some of the individuals of each of the parasite species, however, were probably short-term acquisitions (lasting a few years); thus, there may be some bias in the data of the recent past. Our data suggest less extensive movement of Pacific halibut in southern areas. Because parasites are generally more abundant in the south, southern fish may be more easily classified. Nevertheless, if the southern fish mingle extensively with more northern fish, there should be more similarity in the parasite faunas. In particular, central area fish should develop characteristics of southern fish. This did not happen, as is shown by the very low propor- tion of central fish misclassified as southern fish (Table 5). Our information cannot completely rule out the move- ment of southern fish to central areas during the spawn- ing season, and then back to southern areas for the feed- ing season. Their long-lived parasite fauna, having been established in the distinct southern areas, would probably Blaylock et al : Use of parasites in discriminating stocks of Htppoglossus stenolepis Table 4 Cross validation results of a four-category discriminant function classification of adult Pacific halibut in the northeast Pacific based on parasite data. Numbers of fish assigned to a category and the corresponding percentage of the sample in that category are shown. Correct classifica- tions are shown in bold (63% of 240). OR-SQC = Oregon- northern California to southern Queen Charlotte Islands, NQC-KP = northern Queen Charlotte Islands to Kenai Peninsula, KI-SB = Kodiak Island to southern Bering Sea, PI-SMI = Pnbilof Islands to St. Matthew Island. True category OR-SQC NQC-KP KI-SB PI-SMI Assigned category OR-SQC NQC-KP KI-SB PI-SMI 60 79% 3 5% 7 9% 3 4% 28 50% 23 30% 1 3% 7 9% 21 42% 41 53% 6 20% 6 8% 4 7% 7 9%. 23 77% Table 6 Cross validation results of five-category discriminant function classification for juvenile Pacific halibut in the northeast Pacific based on parasite data. Numbers of fish assigned to a category and the corresponding percentage of the sample in that category are shown. Correct classi- fications are shown in bold (44% of 62). NQC = northern Queen Charlotte Islands, NI = Nagai Island, UP = Unimak Pass, and PI = Nunivak Island (central Bering Sea). True category NQC KI NGI UP PI Assigned category NQC KI NI 9 69% 1 14% 5 39% 5 39% 6 38% 4 31% 5 71% 3 23% 4 25% 3 23% 2 15% UP 2 15% 5 39% 1 6%' PI 1 8% 5 31% Table 5 Cross validation results of a three-category discriminant function classification for adult Pacific halibut in the northeast Pacific based on parasite data. Numbers offish assigned to a category and the corresponding percentage of the sample in that category are shown. Correct clas- sifications are shown in bold (83% of 240). OR-SQC = Oregon-northern California to southern Queen Charlotte Islands, NQC-SB = southeast Alaska to southern Bering Sea, PI-SMI = Pribilof Islands to St. Matthew Island. Assigned category True category OR-SQC NQC-SB PI-SMI OR-SQC NQC-SB PI-SMI 63 83% 10 7% 7 9% 112 84% 5 17% 12 9% 25 83% Table 7 Cross validation results of a three-category discriminant function classification for juvenile Pacific halibut in the northeast Pacific based on parasite data. Numbers of fish assigned to a category and the corresponding percentage of the sample in that category are shown. Correct classi- fications are shown in bold (66% of 62). NQC = northern Queen Charlotte Islands, NI = Nagai Island, UP = Unimak Pass, PI = Nunivak Island (central Bering Sea). True category NQC-NI UP PI Assigned category NQC-NI UP PI 30 91% 6 46% 10 63% 6 46% 1 6% 1 8% 5 31% not lose their southern character. Winter sampling could potentially determine if this is the case. With respect to the Bering Sea, we suggest that the majority of the mixing occurs in the southern Bering Sea because classification breaks down when the southern Bering Sea is included in the northern region. This mix- ing is consistent with larval studies that show that larvae enter the Bering Sea through the Aleutian chain. Those fish may not disperse far into the Bering Sea. Rather, they either remain in the southern Bering Sea or migrate back to the Gulf of Alaska area (Skud [1977] believed that both occurred). A migration may explain why fish tagged in the Bering Sea tend to be recovered at greater distances from the tagging site than those tagged elsewhere (Geernaert, 1996). Migrations of the central and northern Bering Sea group appear to be in a more northerly direction (Skud, Fishery Bulletin 101(1) 1977), which would preclude mixing in the Aleutians and the Gulf of Alaska. Zoogeographic analysis with patterns of prevalence showed that Bering Sea parasites are rarely found outside the Bering Sea (Blaylock et al., 1998b). The patterns identified in our analysis agree only in part with zoogeographic analyses (Blaylock et al., 1998b). The southern boundaries in both studies are in the vicinity of the Queen Charlotte Islands, providing additional support for the existence of a southern group of halibut. However, this analysis, unlike the zoogeographic analyses, indicated no sign of a division in the vicinity of Kodiak Island, sug- gesting that the division near Kodiak Island depends on short-lived species not included in this analysis. The evi- dence for the existence of a northern Bering Sea group is equivocal; it was supported by the clustering of localities by using prevalences and, to some degree, the clustering of individuals, but was not supported by any other analyses (Blaylock etal., 1998b). With respect to juveniles, Skud's (1977) analysis clearly indicates compensatory movement from the Gulf of Alaska and southern Bering Sea to southern areas, and, as such, predicts that juveniles should have more similar parasite faunas among areas. Our data show this similarity, but there are significant caveats. First, our samples of juve- niles came from areas that form a single group in the clas- sification of adults. The sample from the northern Queen Charlottes is near the southern boundary of that group, and the sample from Nunivak Island is near the northern boundary. Samples of juveniles from other areas, particu- larly the southern area, should be examined to help clarify this issue. Second, and maybe more important, in these smaller fish, prevalences and intensities are low and per- haps hinder separation. However, because halibut at this stage are susceptible to bycatch in other fisheries (IPHC, 1996), management should probably consider juveniles a mixed stock to prevent impacts on future halibut popula- tions in distant localities. Overall, our analysis provides a less clear picture than that of Arthur and Albert (1993) for Greenland halibut in the northwest Atlantic. Part of the lack of clarity may be due to our use of the training and test set method rather than the bootstrapping method used by Arthur and Albert, which would increase the likelihood of correctly classify- ing similar fish. Also, Arthur and Albert were dealing with a very different system. Geological and oceanographic conditions around the Gulf of St. Lawrence are quite com- plex and create great potential for the isolation of stocks. The northeast Pacific is more open and has fewer isolating mechanisms than the northwest Atlantic. Further, the system is clinal (Blaylock et al., 1998b) and Pacific hali- but are quite capable of migrating along the entire Pacific coast; therefore, less clear cut divisions are expected. Nev- ertheless, wc successfully identified groups of fish, some with a high degree of accuracy. Skud (1977) suggested that juveniles will, as adults, homo to the areas in which they were spawned, making the existence of reproductive stocks at least possible. Modern molecular methods could address the issue. For example, molecular methods could potentially address the existence of separate stocks in the south and in the northern Bering Sea. The limited molecular studies done to date, however, have not elucidated any indentifiable stock structure be- cause of limited sampling localities, the limited number of loci examined, and the use of juveniles only. Tsuyuki et al. (1969) examined a single serum hemoglobin transferrin lo- cus in halibut from ten sites from Vancouver Island to the Bering Sea and found that only one southeast Alaska local- ity was different. Grant et al. (1984) found no differences between Gulf of Alaska and Bering Sea halibut at five loci but were able to distinguish northeast Pacific halibut from Japanese halibut. However, it is important to note that biochemical and genetic information measures differentia- tion at a different time scale than that reflected in parasite data (Lester et al.. 1988). According to Grant (1984), move- ment of only a few Atlantic herring (Clupea harengus) may be sufficient to obscure true differences between different breeding stocks. Thus, even limited gene flow could obscure any differences in the loci examined. Parasite or tagging information alone, however, can not determine whether or not the groups we identified are reproductive stocks. Therefore, all potential factors that might refine the halibut stock concept should be consid- ered. The parasite data suggest a conservative approach to management that recognizes a mixed stock of juveniles and three potential stocks of adults — one in the south, an- other in the northern Bering Sea, and a third and largest centered in the Gulf of Alaska. Acknowledgments We thank the International Pacific Halibut Commission, Seattle, WA, for coordinating sampling and for financial support. Mark Higgins and John Quintero provided invaluable assistance in the laboratory. Tom McDonald and Dave Whitaker provided technical assistance. We also thank Al Shostak and Jeff Lotz for advice and comments. Literature cited Arthur, J. R., and E. Albert. 1993. Use of parasites for separating stocks of Greenland halibut (Reinhardtius hippoglossoides) in the Canadian northwest Atlantic. Can. J. Fish. Aquat. Sci. 50:2175- 2181. Blaylock, R. B., J. C. Holmes, and L. Margolis. 1998a. The parasites of Pacific halibut iHippoglos- siis stenolepis) in the eastern North Pacific; host-level influences. Can. J. Zool. 76:536-547. Blaylock. R. B.. L. Margolis. and J. C. Holmes. 1998b. Zoogeography of the parasites of Pacific halibut [Hippoglossus ulenolepis) in the northeast Pacific. Can. J. Zool. 76:2262-2273. Boje. J., F. Riget, and M. Koie. 1997. Helminth parasites as biological tags in population studies of Greenland halibut tRciuhariltiiia hippoglossoi- dex (Walbauml). in the north-west Atlantic. ICES J. Mar Sci. 54:886-895. Bush. A. O.. K. D. LafTcrty, J. M. Lotz. and A. W. Shostak. 1997. Parasitology meets ecology on its own terms: Margo- lis et al. revisited. J. Parasitol. 83:575-583. Blaylock et a\. Use of parasites in discriminating stocks of Htppoglossus stenolepis Geemaert, T. 1996. Tagging studies. In Report of assessment and re- search activities, 1995, p. 277-288. International Pacific Halibut Commission, Seattle, WA. Gibson, D. I. 1972. Flounder parasites as biological tags. J. Fish Biol. 4:1-9. Grant, W. S. 1984. Biochemical population genetics of Atlantic herring, Clupea harengus. Copeia 1984:357-364. Grant, W. S., D. J. Teel, T. Kokayashi, and C. Schmitt. 1984. Biochemical population genetics of Pacific halibut {Hippoglossus stenolepis) and comparison with Atlantic halibut (H. hippoglossus). Can. J. Fish. Aquat. Sci. 41: 1083-1088. IPHC (International Pacific Halibut Commission). 1996. Report of assessment and research activities, 1995, p. 121-172. IPHC, Seattle. WA. Khan, R. A., M. Dawe. R. Bowering, and R. K. Misra. 1982. Blood protozoa as an aid for separating stocks of Green- land halibut, /ft'(>i/iard/(Hs hippoglossoides. in the northwest Atlantic. Can. J. Fish. Aquat. Sci. 39:1317-1322. Krzykawski, S. and J. Wierzbicka. 1982. An attempt to determine the systematic position of Greenland halibut. Reinhardtius hippoglossoides (Wal- baum. 1792), from Labrador region and Barents Sea on the basis of morphometric, biologic, and parasitological studies. Acta Ichthyol. Piscat. 22:59-75. Lester, R. J. G. 1990. Reappraisal of the use of parasites for fish stock identification. Aust. J. Mar. Freshwater Res. 41:855-864. Lester, R. J. G., K. B. Sewell, A. Barnes, and K. Evans. 1988. Stock discrimination of orange roughy, Hoploslelhus atlanticus, by parasite analysis. Mar Biol. 99:137-143. Moser. M. 1991. Parasites as biological tags. Parasitol. Today 7:183- 186. Skud, B. E. 1977. Drift, migi-ation, and intermingling of Pacific halibut stocks. International Pacific Halibut Commission, Scien- tific Rep. No. 63, 42 p. IPHC, Seattle, WA. Tsuyuki, H., E. Roberts, and E. A. Best. 1969. Serum transferrin systems and the hemoglobins of the Pacific halibut iHippoglossus stenolepis). J. Fish. Res. Board Can. 26:2351-2362. Wilkinson, L., M. Hill, J. Welna, and G. Birkenbeuel. 1992. SYSTAT for Windows: statistics, version 5 edition, 750 p. SYSTAT Inc., Evanston, IL. Williams, H. H., K. MacKenzie, and A. McCarthy. 1992. Parasites as biological indicators of the population biology, migrations, diet, and phylogenetics of fish. Rev. Fish Biol. Fish. 2:144-176. 10 Abstract— Extensive plankton collec- tions were taken during seven Septem- ber cruises (1990-93) along the inner continental shelf of the northcentral Gulf of Mexico ( GOM ). Despite the high productivity and availability of food during these cruises, significant small- scale spatial variability was found in larval growth rates for both Atlantic bumper (Chloroscombrus chrysurus, Carangidae) and vermilion snapper {Rhomboplites aurorubens, Lutjani- dael. The observed variability in larval growth rates was not correlated with changes in water temperature or asso- ciated with conspicuous hydrographic features and suggested the existence of less-recognizable regions where condi- tions for growth vary. Cruise estimates of mortality coefficients (Z) for larval Atlantic bumper (n=32,241 larvae from six cruises) and vermilion snapper (n = 2581 larvae from four cruises) ranged from 0.20 to 0.37 and 0.19 to 0.29, re- spectively. Even in a subtropical cli- mate like the GOM, where larval-stage durations may be as short as two weeks, observed variability in growth rates, particularly when combined with small changes in mortality rates, can cause order-of-magnitude differences in cumulative larval survival. To what extent the observed differences in growth rates at small spatial scales are fine-scale "noise" that ultimately is smoothed by larger-scale processes is not known. Future research is needed to further characterize the small-scale variability in growth rates of larvae, particularly with regard to microzoo- plankton patchiness and the temporal and spatial pattern of potential preda- tors. Small-scale spatial variability in larval growth rates may in fact be the norm, and understanding the implica- tions of this subtle mosaic may help us to better evaluate our ability to partition the causes of recruitment variability. Small-scale spatial and temporal variability in growth and mortality of fish larvae in the subtropical northcentral Gulf of Mexico: implications for assessing recruitment success Bruce H. Comyns Department of Coastal Sciences College of Marine Sciences The University of Southern fVlississippi 703 East Beach Drive Ocean Springs, Mississippi 39566 E mail address bruce.comyns(g)usm.edu Richard F. Shaw Department of Oceanography and Coastal Sciences School of The Coast and Environment Louisiana State University Baton Rouge, Louisiana 70803 Joanne Lyczkowski-Shultz Southeast Fisheries Science Center National Manne Fishenes Service P,0- Drawer 1207 Pascagoula, Mississippi 39568 Manuscript accepted 1 1 July 2002. Fish. Bull. 101(2):10-21 (2003). For many marine fishes year-class strength undergoes large fluctuations because of the inherent variability in larval, postlarval, and juvenile survi- vorship (Hjort, 1914; Gushing, 1975; Lasker, 1975; Hunter, 1982; Houde, 1987; Goshorn and Epifanio, 1991; Pepin and Myers, 1991). Understanding and quantifying recruitment variability remains one of the greatest challenges in fisheries science today (Fritz et al., 1990; Gushing and Horwood, 1994; Leggett and Deblois, 1994; Mertz and Myers, 1995). Early survival rates are influenced not only by predation pres- sure but also by gi-owth rate which can alter the duration of the larval stage when larvae are exposed to accumula- tive high mortality rates (Houde, 1987; Ghambers and Leggett, 1987; Ander- son, 1988; Bailey and Houde, 1989). Pepin (1991) formalized this concept by depicting the cumulative mortality (C) of a population from stage a to older stage b as the direct function of the instantaneous growth ig{x\) and mor- tality (AfUl) rates such that ■mx] g\x] dx, where .v are factors that influence the vital rates (M andg) such as food avail- ability, temperature, and abundance of potential predators. Many questions remain concerning the causes of recruitment variabil- ity. Reasons for variability include the following: the inherent variability in growth and mortality rates and result- ing survivorship; difficulties in estimat- ing mortality rates with sufficient ac- curacy and precision; and the complex interrelationships among factors that affect survivorship of larvae (Parrish, 1973; Laurence, 1979; Houde, 1987; Beyer, 1989; Pepin, 1991 ). Houde ( 1989) hypothesized that cohort survivorship is more sensitive to small changes in vital rates in high latitude systems than in tropical or subtropical systems because the colder temperatures cause slower growth rates and longer larval stage durations, i.e. up to 100 days. Comyns et al.: Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 11 88" / MISSISSIPPI SOUND • 59 •SS ,/ 9 -eo »58 • 57 > • 56 \. • 55'. u • 54 • 53 • 52 51 •46 / • 50 •49 ,47 •AS BRETON SOUND ^ •AS ,•44 • 36 • 37, • 38 • 39 • 4Q • 41,: 7 *4?- •43 J;^ • 34 • 33 . • 32 • 31 • 30 • 29 • 28 ■^n^ T.rH.^^ j^f4 .-^ tlUkm Figure 1 Station locations (•) of plankton collections in the northcentral Gulf of Mexico, September 1990 to 1993. Pepin (1991) questioned this conclusion because he found no net effect of temperature on postlarval stage-specific mortality rates, although his study was based mainly on interspecific variation in mortality (Francis, 1994). The objectives of our study were to determine if growth rates of Atlantic bumper (Chloroscombrus chiysurus, Carangidae) and vermilion snapper (RhombopUtes auro- riibens, Lutjanidae) varied over small spatial scales in the northcentral Gulf of Mexico (GOM); determine the magni- tude and variability of cruise estimates of larval mortal- ity; and determine the potential influence of variability in these vital rates on cohort survivorship in a region where summer water temperatures approach 30°C and larval stage durations are as short as two to three weeks. Vermilion snapper is the most abundant species of snap- per in the northern GOM (Goodyear and Schirripa'), and Atlantic bumper is the most abundant carangid. Materials and methods Sampling location and shipboard procedures Seven, three-day cruises were conducted in inner-shelf waters of east Louisiana, Mississippi, and Alabama during ' Goodyear, C. P., and M. J. Schirripa. 1991. A biological profile for vermilion snapper with a description of the fishery in the Gulf of Mexico. Unpublished report CRD 87/88-16, 53 p. " South- east Fisheries Science Center, National Marine Fisheries Ser- vice, 75 Virginia Beach Drive, Miami FL 33149. September 1990-93 (Fig. 1). Cruise estimates of larval mortality were determined by using data from all cruises during the four-year period. Specimens used for age and growth analyses were collected during 14-16 September 1991 when larvae of both vermilion snapper and Atlantic bumper were abundant. Larvae were collected with a 1 m x 1.4 m Tucker trawl fitted with a 333-pm mesh nitex net and a mechanical flowmeter. Oblique tows were taken from the surface to within a few meters of the bottom and back to the surface at a speed of approximately two knots (1.0 m/s). Samples were concentrated and stored in 95'7f ethanol. At each sampling location surface, midwater and bottom measure- ments of temperature and salinity were obtained with water-bottle casts. Laboratory procedures Lengths of larvae were measured to the nearest 0.1 mm by using a stereomicroscope (12x or 25x) fitted with an ocular micrometer and the larvae were sorted into 0.5-mm size classes. Measurements were taken from the tip of the snout to the end of the notochord in preflexion larvae (notochord length), and from the tip of the snout to the end of the urostyle or hypural plate (whichever was more distal) in flexion or postflexion larvae (standard length). Larval shrinkage was not accounted for because between- station and between-cruise comparisons of growth rates were made with larvae that were preserved in the same concentration of ethanol and stored for approximately the same length of time. Shrinkage of ethanol-preserved 12 Fishery Bulletin 101(1) lan'ae is not large, e.g. 0 to 7% (Theilacker, 1980; Fowler and Smith, 1983; Kruse and Dalley, 1990). It is unlikely that size-related shrinkage effects would have biased our estimates of growth rate because these estimates were based on larvae in similar size classes. Addition- ally, Theilacker (1980) found that preserving northern anchovy larvae after they had died during net capture caused additional shrinkage, but this shrinkage was at a constant rate that was proportional to fish length. Catches of larvae were standardized to account for sampling effort and expressed as number of larvae under 10 m- of sea sur- face. This method of expressing the abundance of larvae more accurately reflects station differences in abundance than a mean density (number/ni'^) when fish larvae are not homogeneously distributed throughout the water column, as has been shown with other species from this area (Lyc- zkowski-Shultz and Steen, 1991), and when sampling (sta- tion) depths are variable, as they were in our study. Dry weights of larvae were determined by rinsing speci- mens with distilled water, drying for 24 h at 60°C, and weighing to the nearest 0.1 pg. Both sagittal otoliths were removed following rehydration for 12 h. Otoliths were mounted convex side up on a glass microscope slide with a drop of Pro-Texx mounting medium and a cover slip. Oto- lith growth increments were counted in the sagittal plane under oil immersion (12.50x). A total of 140 Atlantic bumper larvae and 119 vermilion snapper larvae were selected for age analyses. Specimens were selected from stations where a wide size range of larvae were collected. Daily otolith increment formation has been validated for larval Atlantic bumper (Leffler and Shaw, 1992). Daily increment formation has not been vali- dated for vermilion snapper; however, otolith increments observed in larval vermilion snapper were very similar in width and spacing to validated daily increments found in red snapper from this region (Szedlmayer, 1998; Lycz- kowski-Shultz and Comyns^). Slopes of age-length regres- sions for larval vermilion snapper (n=ll) and red snapper (n=25) collected during July 1992 in our study area were not significantly different, further indicating that vermil- ion snapper, like red snapper, form daily otolith growth increments. Otolith growth increments were counted by using the sagitta (right or left) that provided the most distinct incre- mental zones. Paired /-test analyses showed no significant difference (P<0.05) in diameters of left and right sagittae in both vermilion snapper (/j = ll) and Atlantic bumper (n=20). Daily increments were counted along the longest axis of the otolith from the core to the outer edge. Otoliths were read once by a single reader, and a random subsam- ple of otoliths from vermilion snapper («=30) and Atlantic bumper (n=30) was read a second time to examine within- Lyczkowski-Shultz, J., and B. H. ("omyns. 1992. Karly Hfi- history of snappers in coastal and sholf waters of tho north- central Gulf of Mexico late summer/fall months, 1983-1989, 12 p. + 9 tables. 17 figures. Technical Report submitted to the National Marine Fisheries Service, Southeast Regional Office, 9721 Executive Center Drive North, St. Petersberg, FL. 33702. reader variability. Otolith increment counts differed by one day for only two of the 30 otoliths during the second reading for both species. Data analysis Age-length and age-weight relationships were described by using the exponential equation L or IV = exp(a -i- bt), where, in its linearized form, L = notochord or standard length in mm; W = dry weight in mg; a = y-intercept; b = slopeof regression line (instantaneous growth rate); and t = age of larvae in days. Values of a and h were calculated from the linearized form of the growth equation after the length or weight data were transformed to their natural logarithms. The instantaneous growth rate (b), i.e. the slope of the log- transformed, age-length or age-weight relationship, is also referred to as the growth coefficient. Caution must be exercised when making dry-weight comparisons because of preservation-induced weight loss. Kristoffersen and Salvanes (1998) found that body weight loss was as high as 37-39% in small ethanol-preserved mesopelagic fishes. Dry weight data were used only to determine whether relative changes in weight tracked trends found in age- length relationships. Analysis of covariance (ANCOVA) was used to determine if differences existed among station estimates of instantaneous growth coefficients (Sokal and Rohlf 1969; SigmaStat, 1995). If differences were found ( a=0.05 ), the simultaneous test procedure ( STP; Sokal and Rohlf 1969) was used as an a posteriori test to determine station differences. Cruise estimates of total larval abundance for each size class (catch curves) were developed for Atlantic bumper and vermilion snapper by summing the abundance esti- mates of each size class under 10 m^ of sea surface from each station. Length-frequency distributions were con- verted to age-frequency distributions by assigning ages to mid-points of the 0.5-mm size classes with the age-length relationship previously described. Age-class abundances were corrected for stage duration by dividing the abun- dance estimate of each age class by their respective dura- tions (Houde, 1977). It is necessary to correct for stage du- rations of age classes if growth rates are nonlinear. Stage durations of age classes were determined by assigning ages based on previously determined growth equations to end-points of the 0.5 mm size classes. This customary method for constructing catch curves relies on the rarely examined assumption that larvae at different sampling locations are growing at similar rates. The high r~ values of the age-length relationships (0.92 for Atlantic bumper; 0.84 for vermilion snapper) that resulted when aged lar- vae from all stations were combined indicated that growth Comyns et al : Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 13 rates over the study area were similar enough to justify use of this technique. Cruise estimates of larval mortality rates for Atlantic bumper and vermilion snapper larvae were estimated from catch curve analyses (e.g. Houde, 1977; Essig and Cole, 1986; Watanabe and Lo, 1988; Deegan. 1990; Co- myns et al., 1991 ). The instantaneous mortality rate (Z) was estimated by the slope of the exponential function relating duration-corrected larval abundance and age (Ricker, 1975): D, = D^ exp i-Zt), where D, = total abundance of larvae at time /; Dy= total abundance of individuals at time 0; Z = instantaneous mortality rate; and t = age of size class in days since spawning. Age and abundance of size classes were fitted to this exponential function with a nonlinear least squares routine, and only the descending limb of the regression was used to estimate mortality rates. To reduce poten- tial biases associated with 1) any trend of increasing variance in the length-at-age distribution with increas- ing age, and 2) net avoidance by larger larvae, only Atlantic bumper and vermilion snapper larvae smaller than 6.1 mm and 6.0 mm, respectively, were used to estimate mortality rates. Kolmogorov-Smirnov two- sample tests showed no significant differences (P<0.05) between size-frequency distributions for day versus night catches within this size range for vermilion snap- per and Atlantic bumper larvae. Results 0 221 0.2 0 18 0.16 0.14 0.12 g 0.1 n=19 1=26 „.,, n=22 T f T T ~ _L n=23 I I — I — I — I I — I — 12 23 24 34 37 41 42 44 47 B i 0.7- 0.6- 0.5- 0.4- " n= 8 — 25 " 11 n= n= 13 22 n=19 -I- n J_ n=23 n= -8 10 12 23 24 34 37 41 Station 42 44 47 Figure 2 Growth coefficients (horizontal bars) for Atlantic bumper larvae collected at nine stations in the northcentral Gulf of Mexico, during 14-16 September 1991. Vertical lines repre- sent 95% confidence intervals around the growth coefficients, and numbers above bars depict sample sizes; (A) shows age versus In length growth coefficients and (B) shows age versus In dry-weight growth coefficients. Age and growth Atlantic bumper larvae, which were commonly found throughout the study area, ranged from 2 to 14 days old, 1.4 mm to 8.1 mm in length, and 0.003 mg to 1.446 mg in dry weight. Estimates of age versus length growth coefficients were not similar for all stations (ANCOVA; P=0.001). The STP revealed no overlap in 95% confidence intervals around growth coefficients for larvae collected at station 42 and larvae collected at stations 41, 23, and 47 (Figs. 2A and 3). According to their respective growth equations, Atlantic bumper larvae at station 42 grew at approximately 0.43 mm/d and reached a length of 6 mm in approximately 13.3 days. Larvae collected at adjacent sta- tion 41 grew faster, approximately 0.63 mm/d, and reached a length of 6 mm in 10.4 days. Similarly significant differences in station estimates (n=9) of age-dry-weight growth coefficients were also found (ANCOVA; P=0.01), and growth coefficients for larvae collected at station 42 were significantly different from larvae collected at stations 41 and 23 (STP; Fig. 2B). By 11 days, the estimated dry weight of an Atlantic bum- per larva at station 42 was 0.38 mg, whereas at station 41 larvae gained weight faster and the estimated dry weight of an 11-d-old larva was 0.58 mg. Adjacent stations 41 and 42 were 10 km apart, and water temperatures at these two locations were very similar. Surface temperatures varied by only 0.1°C (28.7°-28.8°C), and surface and midwater temperatures varied by only 0.5°C. Daily surface water temperatures recorded at a weather bouy within the study area showed that temperatures varied by less than 2°C during the 31-d period prior to our study. Significant differences in station (n=7) growth rates of vermilion snapper larvae were also found in our 14-16 September 1991 cruise (ANCOVA; P=0.03). Vermilion snapper larvae ranged from 4 to 16 days old, 2.5 mm to 6.5 mm in length, and 0.014 mg to 0.696 mg in dry weight. Growth coefficients for larvae collected at stations 15 and 25 were significantly different (STP; Figs. 4A and 5). Ac- cording to their respective growth equations, vermilion snapper larvae collected at station 15 reached a length of 5 mm in 10.7 days, whereas larvae collected at station 25 grew more slowly and did not reach a length of 5 mm until 12.6 days. Stations 15 and 25 were located 17 km apart on the inner shelf at water depths of 29-30 m. Surface water temperatures at these stations varied by 2.2°C, and both surface and midwater station temperatures differed by less than 2°C. 14 Fishery Bulletin 101(1) 10 ~ I 6 en £ 4 Larvae from station 41 L = exp(-0.052 + 0.1 68t) = 0.95, n= Larvae from station 42 L = exp(0.147 + 0.1 24t; r^= 0.91, n= 23 n r 7 9 Age (days) 11 13 Figure 3 Age versus length data for Atlantic bumper larvae (/! = 140) collected at nine stations in the northcentral Gulf of Mexico during 14-16 September 1991. High- lighted are the age-length relationships at two adjacent stations where growth rates differed. L = notochord or standard length in mm; t = larval age in days. A 0.151 0.13 0 11 009 0.07 1 0 05 n 12 n=12 - T T _ — n =5 n=15 - - - — ~ __ n=23 n=16 11 15 16 T 1 r 24 25 30 31 B 0.5- n=12 0 45- f)=ii 0 4 - — — n=8 n =5 0 35- - - - n= 16 n= 15 ^ ^M n=23 ^ ^^ ^^ ^^ 0.3- L - - - 0.2b- 11 15 16 24 Station 25 30 31 Figure 4 (irowth coofTicients (horizontal bars) for vermilion snapper larvae collected at seven stations in the northcentral Gulf of Mexico during 14-16 September 1991. Vortical lines represent 9.')7( confidence intervals around the growth coefficients, and numbers above bars depict sample sizes. (Al shows age versus In length growth coefficients, and iBl shows age versus In dry- weight growth coefficients. Differences in age versus dry-weight growth coefficients were also significantly different (ANCOVA; P=0.03) and once again stations 15 and 25 (Fig. 4B) were significantly different (STP). Vermilion snapper larvae gained weight faster at station 15 where an 11.0-d-old larva had an esti- mated dry weight of 0.28 mg. At station 25 the estimated dry weight of this same larva was only 0.17 mg. Although vermilion snapper larvae were collected at most of the sta- tions within the study area (Fig. 1), abundances were low at shallow (12-14 m depth) stations immediately south of the Mississippi-Alabama coast, and larvae were never collected at stations within Chandeleur Sound. These sta- tions were very shallow (4-9 m). Although our study did not assess microzooplankton prey availability, macrozooplankton dry-weight estimates varied widely over space and time. At the 33 stations east of Chandeleur Sound where larvae of vermilion snapper and Atlantic bumper used in our study were captured, macro- zooplankton dry-weight estimates at 20 stations exceeded 3g/100 m'*, and at eight of those stations values exceeded 5g/100 m'^. Seven days later only at five of the 33 stations were macrozooplankton dry-weight estimates >3g/100 m' and at no station did estimates exceed 5g/100 m-^. Mortality estimates Atlantic bumper was generally the most abundant spe- cies in plankton collections; 32,241 larvae were collected during six cruises conducted in September of 1990, 1991, and 1993. Mortality rates were not estimated for Atlantic bumper larvae collected during the two cruises conducted in September 1992 because abundances of larvae were very low. When station abundance data were pooled for each of the six cruises, size-frequency distributions gener- Comyns et al,: Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 15 8-1 Larvae from station 15 L = exp(0.368 -H oust) ^/ 6 - ^ °''"'' \^^^^ ? E £ 4- C ^^^^^^ o -1 ^^g^^^^^^^^»°°^ Larvae from station 25 ^^^^^^""^"^"^ L = exp(0,512 -(- 0,087t) 2 - n - r^= 0.91, n=23 U 1 1 1 1 1 4 6 8 10 12 14 Age (mm) Figure 5 Age-length data for vermilion snapper larvae in =89 ) collected at seven stations in the northcentral Gulf of Mexico during 14-15 September 1991. L = body length in mm; t = larval age in days. Highlighted are the age-length relation- ships at two stations where growth rates differed. ally exhibited a similar decrease in abundance with suc- cessively larger size classes; however, the smallest size class (1.1-1.6 mm) was the most abundant in only three of the six cruises (Fig. 6, A, B, and F). This under repre- sentation of the smallest size class in several cruises was likely influenced by several potential factors, including a possible decrease in spawning prior to sampling and patchiness of eggs and newly hatched larvae caused by the aggregation of spawning adults. Cruise-estimates of mortality coefficients, which were derived by pooling data from all stations sampled during a cruise and omitting the smallest size class, ranged from 0.20 to 0.37 (Fig. 6). It is likely that mortality rates varied between stations, but as previously mentioned, an average cruise-estimate of mor- tality was determined to ascertain a realistic level about which the effects of small variations in growth rates could be assessed on the cumulative survival of larvae. Standard errors of Z estimates were low, ranging from 0.02 to 0.05. Size-frequency distributions were derived for vermilion snapper lar\'ae (n=2581) taken during two September 1991 cruises, and single late-September cruises in 1992 and 1993 (Fig. 7) when vermilion snapper larvae were abundant. Mortality estimates could not be estimated for five September cruises during the period 1990-93 because relatively few larvae were collected. Larvae collected dur- ing three of the four cruises when they were abundant showed a steady decrease in abundance of successively larger size classes (Fig. 7, A, C, and D). During the fourth cruise (late September 1991; Fig. 7B), the size-frequency distribution showed a distinct peak in abundance of in- termediate-size larvae (4.0-mm size class). Mortality coef- ficients (Z) from the four cruises ranged from 0.19 to 0.30 and standard errors for the mortality coefficients were relatively low ranging from 0.02 to 0.05. Discussion Plankton collections taken in the northcentral GOM during September showed that growth and mortality rates did vary in time and space for Atlantic bumper and vermilion snapper larvae, and that these differences were great enough to significantly impact the cumulative sur- vival of larvae in a subtropical climate where larval-stage durations are short (i.e. two weeks). Growth and mortal- ity estimates of vermilion snapper larvae were previously unknown. Two previous studies of growth and mortality of Atlantic bumper larvae (Leffler and Shaw, 1992; Sanchez- Ramirez and Flores-Coto, 1998) provided no information on variability in growth rates at small spatial scales and no estimates of mortality during the period when our study was conducted. Highly significant between-station differences in growth rates were observed for both Atlantic bumper and vermilion snapper larvae. The largest difference in age versus length growth coefficients for Atlantic bumper larvae was found at adjacent, inner-shelf stations located approximately 10 km apart. According to growth equations, the faster grow- ing larvae grew to a length of 6 mm 2.9 days sooner than larvae at the adjacent station, and differences in larval weight gain as expressed by dry weight of 11-d-old larvae varied by over 30^?^. Water temperatures at these two sta- tions were extremely similar; surface temperatures varied by only 0.1°C, midwater temperatures varied by 0.4°C, and surface and midwater temperatures varied by 0.5°C. It is likely that a similarly small temperature differential was present during the two-week period prior to this cruise, i.e. throughout the life of larvae used in our study because dai- ly surface water temperatures recorded at a weather bouy within the study area during the previous month showed 16 Fishery Bulletin 101(1) 3000 2500 2000 1500 1000 500 0 Sep. 7-9 1990 49/52 stations n=5,531 Z=0-37 r^=0.97 SE=0.05 2 1 2.6 3,1 3 6 4 1 4.6 5,1 5 6 700 600 500 400 300 200 100 0 11 16 2.1 2.6 3.1 3.6 4.1 4.6 5.1 5.6 Sep. 14-16 1991 41/50 Stations n=3,611 Z=0.20 r^=0.82 SE=0.03 1400 1200 1000 800 600 400 f 200 J J Sep. 9-11 1993 33/47 stations n =7,282 Z=0.30 1^=0.93 SE=0.04 -fi^JZL 1 1 1 6 2 1 26 3 1 3.6 4,1 4,6 5.1 5 6 Sep. 14-16 1990 49/54 stations n=9,948 Z=0.30 r=0.98 SE=0.02 ^ wn Wl m^ _ 3 1 36 4 1 46 5 1 5 6 Sep. 21-23 1991 45/51 stations n =4,379 Z=0.28 r^=0.97 SE=0.02 B H PI PI PI - 1.1 16 2.1 2,6 3,1 3.6 4.1 4.6 5 1 56 Sep. 19-21 1993 25/32 stations n = 1,490 Z=0.32 r^=0.90 SE=0.05 11 16 2 1 2 6 3.1 36 4 1 46 5 1 5 6 Size class (mm) Figure 6 Size-frequency distributions of Atlantic bumper larvae collected during six cruises in the northcentral Gulf of Mexico during September 1990. 1991, and 1993. Values of Z, SE, and r'^ refer to mortality curves produced from the duration-corrected, age-frequency distributions (omitting the smallest size class). The fraction listed for each cruise refers to positive (larvae were collected) stations/total stations sampled. Abundance of each size class is pooled estimates of station abundances. that temperatures varied by less than 2°C. Significant dif- ferences in both age versus length and age versus weight relationships were also found for vermilion snapper larvae collected at relatively close stations (i.e. 17 km apart). Wa- ter temperatures at these two stations were similar but differed by as much as 2°C. Faster growing larvae reached a length of 5 mm approximately 2 days sooner than larvae growing in nearby areas. Significant differences were also found in larval weight-gain; dry weight of 11-d-old larvae from different stations varyicd by as much as 657f . The variability in growth rates that we observed was likely caused by station differences in food availability and size-selective mortality, and to a lesser degree by wa- ter temperature. Unfortunately our data did not allow us to determine the individual effects of these factors on observed growth rates. At least for Atlantic bumper, the effects of temperature changes were probably minimal. Larval survival is generally more influenced by factors other than temperature. Morse (19891 found a positive cor- relation between length-dependent mortality and surface water temperature for 26 larval fish taxa and attributed this to increased predator consumption rates (caused by increased metabolic rates) at higher temperatures. He also concluded that increased growth due to increases in temperature alone would generally impart no advantage to reduce larval mortality because of the concomitant increased predatory consumption rates. Increased larv-al- stage duration at cooler temperatures is not necessarily associated with increased cumulative larval mortality be- cause predation rates decline with decreasing temperature (Pepin, 1991; FVancis, 1994). Methot ( 1981 ) concluded that after correcting for the effect of temperature on growth rates, the mean growth rate of larval fish is an indicator of the degree to which lar\'al growth, and presumably sur- vival, is food limited. We acknowledge that size-selective predation, i.e. "cull- ing out" the slowest (or fastest! growing lai-vae, could have produced the differences in size-at-age structure among Comyns et a\: Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 17 i 0) 100 - o CO 50 ■ > B Sep, 14-16 1991 24/36 stations 938 larvae Z=0 25 r^=097 SE=0.03 71 5 55 180 160 140 120 100 80 60 40 20 0 all II Sep 21-23 1991 29/35 stations 975 larvae Z=0.29 71 r-=0.95 SE=0.05 L II 200 . 180 1 ^ 160 H ^ 120^ ^ 100 ^ y 80 0 '^"' 60 > 40 i ^ 20 J K 0 25 lis ^j / / D Sep. 27-29 1992 50 29/37 stations 45 458 larvae Z=0-18 "o ^ pj 35 . < r=0.87 SE=0,03 30- :j > 25 /■ K 20 15 10 0 1 1/1 , g^ Sep. 19-21 1993 11/32 stations 210 larvae Z=0.25 3 5 4 4.5 5 55 25 3 Size class (mm) Figure 7 Size-frequency distributions of vermilion snapper larvae collected during four cruises conducted in the northcentral Gulf of Mexico during September 1991, 1992. and 1993. Estimates of Z, SE, and r- refer to mortality curves produced from duration-corrected age-frequency distributions. The fraction listed for each cruise refers to positive (larvae were collected) stations/total stations sampled. Abundance of each size class was pooled from estimates of station abundances. stations that we observed. However, predation pressure seems unlikely to have been the primary cause of this variability. If among-station variability in size-selective mortality was largely responsible for the differences in larval growth rates, one would expect the variability in size-at-age at each station to be quite variable and this was not the case. Stations where the effects of size-selec- tive mortality were minimal (or less) should have had both fast and slow growing larvae present; yet coefficients of determination (r^) were >0.90 for age versus length regressions at all stations. Furthermore, there was no correlation between observed growth rates and r- values which would be expected if size selective predation was largely responsible for the variability in growth rates that we observed. Many studies have shown that food availability has a large influence on growth rates of larvae (e.g. Houde and Schekter, 1981; Buckley et al., 1987; Pepin, 1991i and it is likely that station differences in food availability influ- enced our observed differences in larval growth rates. We did not collect the small size-fraction of prey eaten by fish larvae, but our data did reveal extensive spatial and tem- poral variability in the abundance of macrozooplankton. Macrozooplankton biomass at station 42, where relatively slow growth of Atlantic bumper occurred, was 2.6 mg/100 m-' whereas at station 41, where larvae were growing faster, macrozooplankton dry weight (3.9 g/100 m') was bWc higher. When all stations were considered, there was no correlation between macrozooplankton dry weight and growth coefficients of lan'ae, but macrozooplankton biomass was certainly very patchily distributed. For most stations there was at least a 509^ difference in macrozoo- plankton dry weight between one of the adjacent sta- tions. It is equally likely that the smaller size fraction of zooplankton that fish lai-vae eat were also very patchily distributed. In addition, several other studies have shown that primary production in the northern GOM is dynamic and spatially heterogenous (Lohrenz et al., 1990, 1994; Re- dalje et al., 1994), although these studies have focused on regions influenced by discharge from the Mississippi and Atchafalaya rivers. In many studies the significant spatial variability in growth rates of field-caught lai-\'ae cannot be explained by changes in water temperature. These reported differences in growth rates have often been associated with factors such as storm events (Lasker, 1975; Maillet and Checkley, 1991). different geographical locations (Mokness, 1992; Nixon and Jones, 1997; Allman and Grimes, 1998;), or distinct hydrographic features such as tidal fronts (Munk, 1993) and riverine discharge plumes (Govoni et al., 1985; DeVries et al., 1990; Lang et al., 1994). All studies in the GOM that have reported spatial differences in larval 18 Fishery Bulletin 101(1) growth rates have involved comparisons in the vicinity of the Mississippi River discharge plume (Govoni et al., 1985; DeVries et al., 1990; Lang et al., 1994; Allman and Grimes, 1998). The observed variability in larval Atlantic bumper and vermilion snapper growth rates reported in our study was not associated with conspicuous hydrographic fea- tures (e.g. hydrographic convergence zones) and suggests the existence of less-recognizable regions where condi- tions for growth vary. Cruise estimates of mortality were determined to as- certain a realistic level about which the effects of small variations in growth rates on the cumulative survival of larvae could be assessed. In order to do this, data from all stations sampled during a cruise were pooled. This provided the most reliable general estimate of mortality for each cruise despite likely site-specific differences in mortality rates that are extremely difficult to measure. Such pooling of data is not unusual; in fact Morse (1989) suggested that samples should be summed over the larval production cycle. Essig and Cole (1986) estimated mortal- ity rates of larval alewives iAlosa pseudoharengus) by us- ing both converted length-frequency distributions, as we did, and actual age-frequency distributions. They found no statistical difference between the two methods. Pepin and Miller ( 1993 ), however, warned that because variability in observed length-at-age increases with larval age (Cham- bers et al., 1988), analyses that use size in older fish to rep- resent age may yield biased estimates of mortality rates. Yet, Pepin and Miller (1993) observed that their mortality rates, which were estimated by using size as a proxy for age, were consistent with mortality rates reported from other environments and species. Ideally, all fish would be aged, but for our study this was not possible because of the large sample sizes, multiple cruises, and the labor-inten- sive nature of otolith preparation for age determination. Atlantic bumper lai-vae were extremely abundant in = 32,241 for six cruises), and cruise estimates of age-fre- quency distributions showed consistent, well-defined de- scending limbs. Estimates of mortality coefficients (Z) for Atlantic bumper larvae were similar for September cruis- es conducted in the same year. For example, in 1990 the two cruise estimates of Z were 0.37 and 0.30, in 1991 the two Z estimates were 0.20 and 0.28, and in 1993 estimates of Z were 0.30 and 0.32. These mortality rates are similar to estimates reported by Leffier and Shaw (1992) during four September cruises in the same area during 1986-87 (Z=0. 17-0.35) and by Sanchez-Ramirez and Flores-Coto (1998) in the southern Gulf (0.15-0.30). In addition, stan- dard errors of the mortality estimates from our study were low, ranging from 0.02 to 0.05. Cruise estimates of mortality rates for vermilion snap- per were determined during four cruises when larvae were relatively abundant («=2581). The descending limbs of three of the size-frequency distributions uniformly spanned all seven size classes, but during one cruise the middle size class was most abundant and the descending limb of this size-frequency distribution was restricted to four size classes. However, mortality rates were quite simi- lar during all cruises (Z=0.19 to 0.30) and each had a low standard error {SE=0.02 to 0.05). Collections of Atlantic bumper and vermilion snapper larvae were taken when water temperatures ranged from 25° to 30°C, and the mortality coefficients estimated from these collections were similar to those reported for other species under similar temperature regimes. Houde (1989) summarized vital rates of six species of larval fish as reported in seven studies where the mid-points of water temperatures at the time of collection ranged from 26° to 28°C. Most of these studies generated a range of mortality estimates, and the mid-points of the ranges reported in six of these studies varied from 0.21 to 0.38, values that are consistent with the mortality estimates (0.19 to 0.39) that we observed. Our primary reason for estimating mortality rates was to ascertain a realistic level about which small variations could be assessed for potential effects on the cumulative survival of larvae, particularly in conjunction with vari- ability in larval growth rates. Our method assumes a con- stant birth rate, or recruitment rate into the population, and assumes that fish leave the population only through death. There is clearly some expected variability in the de- gree to which these assumptions were met; however, based on the similarity of mortality estimates, not only between cruises but also to previously published estimates, it is concluded that our mortality estimates are biologically meaningful. The well-accepted fisheries paradigm holds that changes in year-class strength are determined by variability in mor- tality during early life stages (Sissenwine, 1984; Houde, 1987; Bailey and Houde, 1989; Gushing and Horwood, 1994). Despite extensive efforts to understand the causes of recruitment variability, significant questions remain because the operant factors are likely to be interrelated parts of the ecosystem dynamics that comprise a multidi- mensional system (Ellersten et al., 1995). For example, it is not the mortality or growth rate alone that determines survival during the early life-stages, but the ratio MIG, the stage-specific mortality rate (Pepin, 1991). Examin- ing previously published information, Houde (1989) found an exponential increase in predicted lai-val-stage dura- tion with decreasing water temperature for 26 species of larval fishes and surmised that when temperature is low, small changes in growth rates can induce large changes in larval-stage duration that may significantly affect the recruitment process. To determine the potential effects that variability in vi- tal rates might have on the cumulative survival of larvae, hypothetical numbers of newly hatched Atlantic bumper were projected to a size of 6 mm under the influence of the growth and mortality rates we observed in the study area (Table 1). According to these vital rates, and a hypo- thetical initial cohort size of 1 x lO*" individuals, 124,930 larvae survive to a length of 6 mm under the scenario of relatively fast growth and low mortality (G=0.61 mm/d; Z=0.20). If the growth rate is slowed (G=0.45 mm/d) and it takes approximately three days longer to reach a length of 6 mm, the number of larvae that sui-vive to this length is reduced by 44*7^. If the slower growing lai-vae are exposed to the higher mortality rate (Z=0.37), cumulative survival of larvae decreases by an order of magnitude, and only Comyns et al : Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 19 Table 1 Hypothetical survival of Atlantic bumper larvae to a size of 6 mm un the study area. der the infl uence of growth and mortality rates observed in Initial number in cohort Instantaneous mortality coefficient (per day) Age of 6-mm larva (d) Number of 6-mm larvae 1 X 10« 0.20 10.4 124,930 1 X 106 0.20 13.3 69,948 1 X 106 0.30 10.4 44,157 1 X 106 0.30 13.3 18,499 1 X 106 0.37 10.4 21,322 1 X 106 0.37 13.3 7292 7292 larvae survive to a length of 6 mm. Houde (1987) published a projection on the mortality of larvae exposed to hypothetical levels of mortality and growth rates, but the theoretical exercise used relatively long larval-stage durations (45-56 days). Results of our study show that even in a subtropical climate where larval stage durations may be as short as two weeks, relatively small changes in observed larval growth rates, particularly when com- bined with small differences in mortality, can have a large impact on cumulative larval survival. To what extent the observed differences in growth rates at small spatial scales are fine-scale "noise" that is ultimately smoothed by larger-scale processes is not known. Future research is needed to further characterize the small-scale variability in growth rates of larvae, particularly with regard to mi- crozooplankton patchiness and the temporal and spatial pattern of potential predators. Small-scale spatial vari- ability in larval growth rates may in fact be the norm, and understanding the implications of this subtle mosaic may help us to better evaluate our ability to partition the causes of recruitment variability. Acknowledgments Collections serving as the basis of this research were sup- ported by the SEAMAP program (Southeast Area Moni- toring and Assessment Program) and the NOAA/NMFS MARFIN program (Marine Fisheries Initiative). Several cruises were also conducted by personnel from National Marine Fisheries Service in Pascagoula, Mississippi. Sorting of these plankton samples was made possible by funding provided by the U.S Fish and Wildlife Service through the Wallop Breaux program. This program is administered in Mississippi by the Department of Marine Resources (DMR) whose personnel must be thanked for providing support. Sorting of the plankton samples was made possible by the efforts of several people, including Mae Blake, Cindy Gavins, Pam Bond, Dianne Scott, Ngoc Bui, and Jean Bennett. We also thank Pam Bond for many contributions, including acting as field party leader during cruises, for larval identifications, much of the data entry and management, and for preparing otoliths. We are also grateful to Chet Rakocinski for providing help with statis- tical analyses. Literature cited Allman, R. J., and C. B. Grimes. 1998. Growth and mortality of little tunny (Euthynnus alletteratus) larvae off the Mississippi River plume and Panama City, Florida. Bull. Mar. Sci. 62:189-197. Anderson, J. T. 1988. A review of size dependent survival during pre-recruit stages of fishes in relation to recruitment. J. Northwest Atl. Fish. Sci. 8:55-66. Bailey, K. M., and E. D. Houde. 1989. Predation on eggs and larvae of marine fishes and the recruitment problem. Adv Mar Biol. 25 (ISBN 0-12- 026125-1): 1-83. Beyer, J. 1989. Recruitment stability and survival: simple size-spe- cific theory with examples from the early life dynamics of marine fish. Dana 7:45-147. Buckley, L. J.. T. A. Halavik, A. S. Smigielski, and G. C. Laurence. 1987. Growth and survival of the larvae of three species of temperate marine fishes reared at discrete prey densities. Trans. Am. Fish. Soc. Symp. 2:82-92. Chambers, R. C, and W. C. Leggett. 1987. Size and age at metamorphosis in marine fishes: an analysis of laboratory-reared winter flounder iPseudopleu- ronectes americanum with a review of variation in other species. Can. J. Fish. Aquat. Sci. 44:1936-1947. Chambers, R. C, W. C. Leggett, and J. A. Brown. 1988. Variation in and among life history traits of labora- tory-reared winter flounder iPseudopleuronectes ameri- canus). Mar. Ecol. Prog. Sen 47: 1-15. Comyns, B. H., J. Lyczkowski-Shultz, D. L.Nieland, and C.A.Wilson. 1991. Reproduction of red drum in the north-central Gulf of Mexico: seasonality and spawnor biomass. In Larval fish recruitment and research in the Americas; proceedings of the 13th annual larval fish Conference, Merida, Mexico, 21-26 May, 1989, p. 17-26. U.S. Dep. Commer. NOAA Tech. Rept. MNSS 95. Gushing, D. H. 1975. Marine ecology and fisheries, 278 p. Cambridge Univ. Press, Cambridge. 20 Fishery Bulletin 101(1) Gushing, D. H., and J. W. Hoi-wood. 1994. The gT'owth and death of fish larvae. J. Plank. Res. 16:291-300. Deegan, L. A. 1990. Effects of estuarine environmental conditions on population dynamics of young-of-the-year gulf menhaden. Mar Ecol. Prog. Ser 68:195-205. De Vries. D. A., C. B. Grimes, K. L. Lang, and D. B. Wliite. 1990. Age and growth of king and Spanish mackerel larvae and juveniles from the Gulf of Mexico and U.S. South Atlantic Bight. Environ. Biol. Fish. 29:135-143. Ellersten, B., P. Possum, P Solemdal, and S. Sundby. 1995. The 'critical period' concept — a century of recruitment research. Mar. Ecol. Prog. Ser 128:306-308. Essig,R. J, and C.F.Cole 1986. Methods of estimating larval fish mortality from daily increments in otoliths. Trans. Am. Fish. Soc. 115:34-40. Fowler, G. M., and S. J. Smith. 1983. Length changes in silver hake (Merluccius bilinearis) larvae: effects of formalin, ethanol and freezing. Can. J. Fish. Aquat. Sci. 40:866-870. Francis, M. P. 1994. Duration of larval and spawning periods in Pagrus auritus (Sparidael determined from otolith daily incre- ments. Environ. Biol. Fish. 39:137-152. Fritz, E. S., L. B. Crowder, and R.C. Francis. 1990. The National Oceanic and Atmospheric Administra- tion plan for recruitment fisheries oceanography research. Fisheries 15:25-31. Goshom, D. M., and C. E. Epifanio. 1991. Development, survival, and growth of larval weakfish at different prey abundances. Trans. Am. Fish. Soc. 120: 69.3-700. Govoni, J. J., A. J. Chester, D. E. Hoss, and P. B. Ortner 1985. An observation of episodic feeding and growth of larval Leiostomus xanthurus in the northern Gulf of Mexico. J. Plank. Res. 7:137-146. Hjort, J. 1914. Fluctuations in the great fisheries of northern Europe. Rapp. P.-V. Reun. Cons. Int. Explor Mer 20:1-13. Houde, E. D. 1977. Abundance and potential yield of the round herring, Etrumeus teres, and aspects of its early life history in the eastern Gulf of Mexico. Fish. Bull. 75:61-89. 1987. Fish early life dynamics and recruitment variability. Trans. Am. Fish. Soc. Symp. 2:17-29. 1989. Comparative growth, mortality, and energetics of marine fish larvae: temperature and implied latitudinal effects. Fish. Bull. 87:471-495. Houde, E. D., and R. C. Schecter 1981. Growth rates, rations and cohort consumption of marine fish larvae in relation to prey concentrations. Rapp. R-V. Reun. Con.s. Int. Kxplor Mer 178:441-453. Hunter, J. R. 1982. Feeding ecology and predation of marine fish larvae. In Marine fish larvae — morphology, ecology, and relation to fisheries (R. Lasker, ed.), p. 34-77. Univ. Washington Press, Seattle, WA. KristolTersen, J. B., and A. G. V. Salvanes. 1998. Effects of formaldehyde and ethanol pre.servation on body and otoliths ui Maurolicus MuvUen and Benthosenui glaciate. Sarsia 83:95-102. Kruse, G. H., and E. L. Dalley 1990. Length changes in capelin, Mallotus villosus (Muller), larvae due to preservation in formalin and anhydrous alcohol. J. Fish. Biol. ,36:619-621. Lang, K. L., C. B. Grimes, and R. F Shaw. 1994. Variations in age and growth of yellowfin tuna larvae, Thunnus albacores, collected about the Mississippi River plume. Environ. Biol. Fish 39:259-270. Lasker, R. 1975. Field criteria for survival of anchovy larvae: the rela- tion between inshore chlorophyll maximum layers and suc- cessful first feeding. Fish. Bull. 73:453-462. Laurence, G. C. 1979. Larval length-weight relations for seven species of northwest Atlantic fishes reared in the laboratory. Fish. Bull. 76:890-895. Leffler, D. L., and R. F. Shaw. 1992. Age validation, growth, and mortality of larval Atlan- tic bumper (Carangidae: Chloroscombrus chrysurus) in the northern Gulf of Me.xico. Fish. Bull. 90:711-719. Leggett, W. C, and E. Deblois. 1994. Recruitment in marine fishes: is it regulated by star- vation and predation in the egg and larval stages? Neth. J. Sea Res. 32:119-134. Lohrenz, S. E., M. J. Dagg, and T. E. Whitledge. 1990. Enhanced primary production at the plume/oceanic interface of the Mississippi River. Cent. Shelf Res. 10: 639-664. Lohrenz, S. E., G. L. Fahnenstiel, and D. G. Redalje. 1994. Spatial and temporal variations of photosynthetic parameters in relation to environmental conditions in coastal waters of the northern Gulf of Mexico. Estuaries 17:779-795. Lyczkowski-Shultz, J., and J. P. Steen Jr 1991. Diel vertical distribution of red drum Sciaenops ocel- latus larvae in the northcentral Gulf of Mexico. Fish. Bull. 89:631-641. Maillet, G., and D. M. Checkley Jr 1991. Storm-related variation in the growth rate of otoliths of larval Atlantic menhaden Brevoortia tyrannus: a time series analysis of biological and physical variables and implications for larva growth and mortality. Mar Ecol. Prog. Ser 79:1-16. Mertz, G., and R. A. Myers. 1995. Estimating the predictability of recruitment. Fish. Bull. 93:657-665. Methot, R. D., Jr 1981. Spatial covariation of daily growth rates of larval northern anchovy, Engraulis mordax. and northern lamp- fish, Stenobrachius leucopsarus. Rapp. P.-V. Reun. Cons. Int. Explon Mer 178:424-431. Morse, W. W. 1989. Catchability, growth, and mortality of lai-val fishes. Fish. Bull. 87:417-446. Mokness, E. 1992. Differences in otolith microstrueture and body growth rate of North Sea herring [Cliipea harengus L.) larvae in the period 1987-1989. ICES J. Mar Sci., 49:223-230. Munk, P 1993. Differential growth of larval sprat Sprattus spratlus across a tidal front in the eastern North Sea. Mar Ecol. Prog. Ser 99:17-27. Nixon, S. W., and C. M. Jones. 1997. Age and growth of larval and juvenile Atlantic croaker, Micropogimias undulatus. from the Middle Atlan- tic Bight and estuarine waters of Virginia. Fish. Bull. 95: 773-784. Parrish, B. B. 1973. Foreward, fish stocks and recruitment. Rapp. P.-V. Reun Cons. Int. Explor Mer 164:1-3. Comyns et al.: Spatial and temporal variability in growth and mortality of fish larvae in the Gulf of Mexico 21 Pepin, P. 1991. Effect of temperature and size on development, mor- tality, and survival rates of the pelagic early life history stages of marine fish. Can. J. Fish. Aquat. Sci. 48:503-518 Pepin, P., and R. A. Myers. 1991. Significance of egg and larval size to recruitment variability of temperate marine fish. Can. J. Fish. Aquat. Sci. 48:1820-1828. Pepin, P, and T.J. Miller. 1993. Potential use and abuse of general empirical models of early life history processes in fish. Can. J. Fish. Aquat. Sci. 50:1343-1345. Redalje, D. G., S. E. Lohrenz, and G. L. Fahnensteil. 1994. The relationship between primary production and the vertical export of particulate organic matter in a river impacted coastal ecosystem. Estuaries 17:829-838. Ricker, W. E. 1975. Computation and interpretation of biological statistics offish populations. Bull. Fish. Res. Board Can. 191:382. Sanchez-Ramirez, M., and C. Flores-Coto. 1998. Growth and mortality of larval Atlantic bumper Chlo- roscombrus chrysurus (Pisces: Carangidae) in the southern Gulf of Mexico. Bull. Mar. Sci. 63: 295-303. SigmaStat. 1995. Statistical software version 2.0 for Windows 95, NT and 3.1. Jandel Scientific, San Rafael, CA. Sissenwine, M. P. 1984. Why do fish populations vary? In Exploitation of marine communities (R. M. May, ed.), p. 59-94. Springer- Verlag, Berlin. Sokal, R. R.,andFJ. Rohlf 1969. Biometry. 776 p. W.H. Freeman and Company, San Francisco, CA. Szedlmayer, S. T. 1998. Comparison of growth rate and formation of otolith increments in age-0 red snapper. J. Fish. Biol. 53:58-65. Theilacker, G. A. 1980. Changes in body measurements of larval northern anchovy, Engraitlis mordax, and other fishes due to han- dling and preservation. Fish. Bull. 78:685-692. Watanabe, Y, and N. C. H. Lo. 1988. Larval production and mortality of Pacific saury, Cololabis saira, in the northwestern Pacific ocean. Fish. Bull.78:601-613. 22 Abstract— Fecundity (F, number of bixHwied eggs ) and egg size were estimated for Hawaiian spiny lobster iPanulirus margirmtus) at Necker Bank, North- western Hawaiian Islands (NWHI), in June 1999, and compared with previous (1978-81, 1991) estimates. Fecundity in 1999 was best described by the power equations F = 7.995 CL ^■""", where CL is carapace length in mm (f-^'=0.900), and F = 5.174 TW ^ ^ss^ ^here TW is tail width in mm (r2=0.889) (both n=40; P< 0.001). Based on a log-linear model ANCOVA, size-specific fecundity in 1999 was 18% greater than in 1991, which in turn was 16% greater than during 1978-81. The additional increase in size- specific fecundity observed in 1999 is interpreted as evidence for flirther com- pensatory response to decreased lobster densities and increased per capita food resources that have resulted either from natural cyclic declines in productivity, high levels of harvest by the commercial lobster trap fishery, or both. The density decline is well-documented by a fivefold decrease in commercial catch-per-trap- haul (CPUE) during the late 1980s to early 1990s and by a similar decrease m research CPUE for all-sized (includ- ing juvenile) P. marginatus through the 1990s. Fecundity increases are consis- tent with decreases in median body size at sexual maturity, first described from comparisons of 1977-81 and 1986-87 specimens and consistently observed thereafter during the 1990s. Egg size covaried with fecundity; in 1999, indi- vidual eggs within broods had a 11% greater mass (15% greater volume) than eggs brooded in 1991. Implications of these obsei-vations are discussed in rela- tion to possible future management mea- sures for a commercial lobster fishery in the NWHI. More generally, our findings argue for the need to routinely reevalu- ate compensatory responses in e.xploited stocks of lobsters and other resources. Temporal changes in population density, fecundity, and egg size of the Hawaiian spiny lobster iPanulirus marginatus) at Necker Bank, Northwestern Hawaiian Islands Edward E. DeMartini Gerard T. DiNardo Happy A. Williams Honolulu Laboratoi7 Southwest Fisheries Science Center National Manne Fishenes Service 2570 Dole Street Honolulu, Hawaii 96822-2396 E-mail address (for E. E DeMartini): Edward DeMartini@nooa gov Manuscript accepted 20 September 2002. Fish. Bull. 101:22-31 (2003). The endemic Hawaiian spiny lobster (Panulirus marginatus) has been the principal target of the Northwestern Hawaiian Island (NWHI) commercial trap fishery since the mid- to late 1970s (Uchida and Tagami, 1984; Polovina, 2000). Landings and exvessel (whole- sale) value have fluctuated greatly over the years, in part because of annual variations in trapping effort and a 1-yr fishery closure in 1993, but have been generally lower during the 1990s because of declines in oceanic produc- tivity and recruitment and increased exploitation (Polovina and Moffitt, 1995; Polovina et al., 1995). The fishery was closed in 2000 because of increasing uncertainty in the population models used to assess stock status. In Decem- ber 2000 President Clinton, through Executive Order (EO) 13178 and later through EO 13196, established the Northwestern Hawaiian Islands Coral Reef Ecosystem Reserve which may prohibit commercial lobster fishing in the NWHI for at least 10 years. Ajinual research surveys of the National Marine Fisheries Service (NMFS), Honolulu Laboratory, have demonstrated a decline (Fig. 1) in spiny lobster density (CPUE, catch-per-trap-haul) at Necker Bank, NWHI, one of the sites at which spiny lobsters have been consistently targeted since about the mid-1970s. Polovina (1989) first described a den- sity-dependent decrease in median body size at sexual maturity and an increase in asymptotic body size for spiny lobster at Necker Bank, based on a contrast between specimens collected during 1977-81 and 1986-87. DeMartini et al. (1993) observed an increase in size-spe- cific fecundity for specimens collected in 1991, used to further characterize the Necker Bank population's status after- exploitation. Compensatory increases in juvenile growth and survival and in- creases in size at maturity as responses to decreased density following increased fishery exploitation have been observed for other spiny lobster stocks (e.g. see Pollock, 1995a, 1995b). In this article, our objectives were to estimate recent (1999) berried female fecundity and egg size for the Hawai- ian spiny lobster at Necker Bank and to relate these to prior, analogous esti- mates for lobsters collected in 1991 and 1978-81, analyzed by DeMartini et al. (1993). We then use the 1999 fecundity estimates and 1999 commercial catch data to characterize recent egg produc- tion by the Necker Bank population. We conclude with a brief discussion of the management implications of com- pensatory reproductive responses by the population. Methods and materials All specimens used in this study were trapped from Necker Bank surround- ing Necker Island (23°34'N, 164°42'W), NWHI, during the species' mid-spring to mid-summer peak period of egg brooding at mid-archipelago latitudes (Uchida and Tagami, 1984). Specimens DeMartini et al.: Population density, fecundity, and egg size of Panulirus marginatus 23 for 1999 were collected on a cruise of the NOAA ship Townsend Cromwell. Details of specimen collection and processing of the 1978-81 and 1991 samples are described by DeMartini et al. (1993). The 1999 samples were collected during 9-22 June 1999 from the bank terrace at a median 27-m depth by using molded plastic ("Fathom Plus") traps baited with 1 kg of mackerel (Scomber Japonicus) and fished with a standard (over- night) soak. Shipboard processing Specimens were processed identically to those collected in June 1991. All specimens were processed alive within minutes of trap retrieval. Both carapace length (CL: defined as the straight line distance between the anterior edge of the supraorbital ridge and the posterior edge of the carapace along the dorsal midline) and tail width (TW: defined as the straight line distance across the abdo- men at the widest spot between the first and second abdominal segments ) of each specimen were measured to 0.1 mm with dial calipers. TW is the present metric of choice for lobster management in the NWHI trap fishery. CL was the metric used to characterize body size in many prior research and management studies of the species, and its measurement was needed for comparison with results of studies made prior to the mid-1980s. Ber- ried (ovigerous) females were scored for egg developmental stage by using a gross visual proxy (brooded eggs noted as either orange or brown in color to the unaided eye). Berried specimens were individually flash-frozen for laboratory evaluation ashore. Laboratory analyses 700 ^ 600 o o 500 300 - 100 - 0 - 1982 1986 1988 1998 2000 B Commercial CPUE ^TV / \ A: / » * I V \ \ \ %s^ 1 1 1 1 1 1982 1984 1986 1988 1990 1992 Year 1994 1996 1998 2000 Fecundity, here defined in the limited sense of a single brooded egg mass (see Chubb, 2000), was estimated for 5-10 females per 5-mm TW class in order to provide at least 40 total specimens spanning the entire size range for analyses. Except for sample sizes, procedures were identical to those used for the 1991 collection. Only females bearing orange egg clusters with embryos lacking visible melanin pigment (early embryonic development) were considered in order to minimize the probability of physical damage, egg loss, and fecundity underestima- tion during capture and handling, which is an apparent problem only for broods of heavily pigmented (brown), late- development eggs with soft capsules (DeMartini, unpubl. data). Frozen specimens were thawed overnight at 3°C. All four pairs of egg-bearing pleopods were then removed from the abdomen, gently blotted (damp-dry) on a paper towel, and weighed individually to 0.1 mg on an electric Figure 1 Time series plots of (A) the Northwestern Hawaiian Islands (NWHI) commercial trap catch and landings of Panulirus marginatus (no. of lobsters x 1000) and effort (no. of trap-hauls x 1000); and (B) total P. marginatus catch-per-trap-haul (CPUE) at Necker Bank, NWHI, during the 1983-99 commercial fishing seasons and as assessed on 1988-1999 lobster research cruises. (Research CPUE data are lacking for years prior to 1988.) Dashed lines framing the research CPUE curve in B represent bootstrapped 95'7f confidence intervals (DiNardo et al.-). microbalance. Eggs were then carefully teased off pleopod setae with jeweler's forceps and stored after being wrapped in cool, damp paper towels to minimize evaporative weight loss. Individual pleopods were then rew-eighed and the weight of each pleopod's egg complement was calculated by difference. Three subsamples of 0.1-0.2 g, each compris- ing about 700-1000 eggs total (about 100 eggs per pleopod, pooled over all 8 pleopods), were next weighed to 0.1 mg, their component eggs counted, and relative fecundity (RF, number of eggs per gram of brooded eggs) was calculated as a simple ratio, with the three subsamples used to calcu- late a mean and standard error of RF. Fecundity (F, defined 24 Fishery Bulletin 101(1) Table 1 (A) Summary catch statistics {Townseiid Cromwell research cruise, Necker Bank, June 19991 and (B-D) fundamental linear-mass interrelationships for body and egg sizes of female Hawaiian spiny lobster (Pa/iulirus marginatus) at Necker Bank, Northwestern Hawaiian Islands. A Female catch statistics Total Berried TW„ TW,, median range median range 834 54.6 350 42.0 50.1 24-72 51.1 38-72 B Relation of tail width to carapace length and vice versa; model: Y - aX + b TW = 0.6087 CL + AAA and CL = 1.5772 TW - 4.00, where TW = tail width in mm, CL = carapace length in mm, and a and h are fitted constants. [H-'=0.963, ?i=825,P<0.0011 C Relation of body weight to carapace length; model: Y = aX'' SW = 0.00090 CL 2 9^=2 [range: 51.9-114.7 mm CL,n= 197,P<0.0011 where BW = total body weight in g, and CL = carapace length in mm, for unbemed females (source: Uchida and Tagami [1984] ). D Relation of egg weight to egg diameter; model: Y = aX'' EW = 0.3985 ED2 2472 where EW = egg weight in 0.001 mg, and ED =egg diameter in mm. lr-'=0.833,;!=40, P<0.001] as the total number of pleopod-brooded eggs) was calcu- lated as the product of mean RF and total weight of the brooded egg mass. Pilot tests indicated that this procedure estimated F with coefficients of variation (CV, SD/mean x 100%) consistently <5'7f. Subsamples of 25 eggs were ran- domly taken from each female's total egg complement and the diameter of individual eggs were measured (random axis) at 500x magnification by using a dissecting micro- scope and an optical micrometer. Average individual egg weight was also independently derived as the ratio of the weight to numbers of eggs present in the parent sample. Statistical analyses Relations of female body size to fecundity and body size to egg size were evaluated for the 1999 samples by using both linear and nonlinear least squares procedures (proc REG, proc NLIN) of PC SAS for Windows v 6.12 (SAS Institute, 1990a, 1990b). Analysis of covariance (ANCOVA; proc GLM; SAS Institute, 1990c; Chubb, 2000) was used to compare size-specific fecundity estimates of Necker Bank P. ryjarginaliis among the three exploitation periods: 1978-81, 1991, and 1999. Subseasonal variation within spawning seasons was controlled by the aforementioned restriction on month of specimen collection, and single collections were assumed to provide accurate character- izations within exploitation periods. Fecundity data used to characterize the 1978-81 and 1991 periods at Necker Bank are listed in Appendix A of DeMartini et al. (1993). Analogous comparisons of size-specific egg sizes were lim- ited to the 1991 and 1999 periods because no data on vari- ance of egg sizes were available for tiic 1978-81 samples (DeMartini et al., 1993). Body-size-fecundity relations were allometric, hence log-linear (see Somers, 1991); natu- ral logarithms were used for ANCOVAs and regressions of log-linear relations. An index of reproductive potential (IRP; Kanciruk and Herrnkind, 1976) was computed for the 1999 specimens in order to determine the size classes of females that con- tributed most to population egg production. The IRP was constructed by using data for female P. marginatus caught by the commercial fishery at Necker Bank during 1999, collected by several contracted fishery observers. Results Fecundity and egg size of lobsters in 1999 Fecundity A number of the female Panulirus marginatus trapped on the research cruise at Necker Bank during June 1999 were berried (Table 1). The estimated fecundity of 40 females, spannmg 54.3 to 105.4 mm CL (39.3-67.4 mm TW, see CL-TW relation; Table 1), ranged more than fivefold, from 109,865 to 590,530 eggs (Appendix A). Fecundity was positively and nonlinearly related to TW (Fig. 2) and CL and best described by the power equations F = a TW' and F = a CL'', respectively, as F= 5.1743 rW''^ 7580^ |7-=0.889| and F= 7.9952 CL-^-^'i' lr2=0.900. both n=40. P<0.001. DeMartini et al.: Population density, fecundity, and egg size of Panulirus marginatus 25 600 r • • / • / 500 y 1 / • (numbers of eggs ■ o o o o ••/ * • 1 LL 200 • y^ ^ • 100 >* • ■ 1 1 1 1 1 1 35 40 45 50 55 60 65 70 Tail width (mm) Figure 2 Scatterplot and fitted power curve describing the relation between fecundity (no. of brooded eggsxlOOO) and tail width (TW, in mm) for Panulirus marginatus collected at Necker Bank. 1999. 0.80 • • • 075 • • • • • E, 1 0.70 •D O) UJ , , • • • • . •• • • • • • 065 . • • • • • • 060 • • ■ t ■ 1 1 1 1 35 40 45 50 55 60 65 70 Tail width (mm) Figure 3 Scatterplot describing the relation between egg diameter (mm) and tail width (TW, in mm) for Panulirus marginatus collected at Necker Bank. 1999. The standard errors of 6 were 0.1787 and 0.1472, respec- tively. A log-linear fit of the F-CL data (LaF=2.5533 LnCL -hi. 3977) was nominally inferior (r-=0.886) to the curvilin- ear fit but was required for the general linear model used in the ANCOVA comparisons that follow below. Fecundity subsamples averaged 1.5 ±0.14(SE)'7f of total brooded egg mass weight. Brooded egg masses weighed an average 51.4 g and ranged from 15.8 to 109.2 g. CVs of the three replicate estimates of RF averaged 1.2 ±0.1V'/(. Mean RF was 5882 ±160(SE) eggs per g of brooded eggs and ranged nearly twofold from 4030 to 7930 eggs per g of eggs among the 40 females. Based on the aforedescribed nonlinear best fit, the fecundity of the median-size (53.8 mm TW, 80.9 mm CD female caught at Necker Bank in 1999 by the com- mercial trap fishery was an estimated 306,400 ±90,200 (95% CI) eggs. Egg size The mean (±SE) diameter of early-stage eggs carried by the 40 berried females was 0.69 ±0.007 mm, with a range of 0.61 to 0.79 mm. Analogous median (25th, 75th percentile) diameters were 0.70 (0.65, 0.72) mm. The corresponding egg weights were 0.17 ±0.005 and 0.18 (0.15, 0.20) mg (range: 0.13-0.25 mg). Individual egg size (diameter: Fig. 3; weight) was unrelated (both P>0.63) to female body size (TW). Individual egg weight was a power function of egg diameter (Table 1). Temporal comparisons of fecundity and egg size Fecundity Size-specific fecundities differed among the three periods, and body-size-adjusted means differed for each period (Table 2, Fig. 4). Size-adjusted mean fecun- dity in 1999 was 18% greater than in 1991, which in turn was 16% greater than during 1978-81. Lobster in 1999 thus exhibited a cumulative 36% increase in size-specific fecundity over that described for lobster collected during 1978-81. The statistical power ( 1 minus /3, where ji is the probability of making a type-II error) to detect an effect size equal in magnitude to the changes observed between 1978-81, 1991, and 1999 was estimated as >97% at a = 0.05. Egg size Brooded eggs on average were about 5% greater in diameter (equivalent to 15% greater volume assuming the volume of a sphere, V=4/3 nr') and were 11%^ heavier in 1999 compared to 1991 (Table 3). The precision of our 26 Fishery Bulletin 101(1) Table 2 (A) ANCOVA and (B) component least squares regression statistics for log-linear (LnV = Ina + 6LnX) relations of fecundity (F) to carapace length {CD for P. margmatus caught at Necker Bank during three periods: 1978-81, 1991, and 1999. Natural logs are used throughout. Under- lines illustrate that the least-square means for each period differ from one another. MSE = mean square error A ANCOVA model: LaF = = LnCL -1- period (r-'= 0.855; root MSE= =0.1843) Factor df MS F P Model 3 6.82 199.6 0.0001 LnCL 1 17.16 502.1 0.0001 Period 2 0.87 25.5 0.0001 LnCL X period (P=0.27 — ns; not included in final model ) Error 105 0.03 Total 108 12.426 > 12.284 > 12.119 1999 > 1991 > 1978-81 B Regression models: LnY = Lna -i- 6LnA' 1978-81 LnF= 2.7994 + 2. 1569 LnCL, r2=0.708, /!=35, P<0.001, SEa=1.0367, SE 6=0.2412 1991 LnF = 1.5859 + 2.4778 LnCL, r2=0.881, n=34, P<0.001, SE a=0.6934, SE 6=0.1607 1 999 LoF = 1 .3977 -i- 2.5533 LnCL, r-'=0.886, /I =40, P<0.001, SE a=0.6452, SE 6=0.1485 measurements was sufficient for the observed change in egg diameter to have had an 87'* chance of being detected at a = 0.05. Individual and population egg production Based on the IRP of Kanciruk and lierrnkind (1976), most egg production by the Necker Bank population of P. margincitus in 1999 was by small adults (<60 mm TW) that now dominate the population (Table 4). Large adults (>60 mm TW), although highly fecund, are now too rare to contribute substantially to total population egg produc- tion (Table 4). Table 3 (A) ANCOVA and (B) component least squares regression statistics for linear relations of egg diameter (ED, in eye- piece units, X 0.020=mm) to carapace length (CL, mm) of P. marginatus caught on research cruises to Necker Bank during 1991 and 1999. See Table 2 caption for additional details. A ANCOVA model: ED = CL + period (r2=0.142; root MSE=2. 193) Factor df MS F P Model 2 27.36 5.69 0.005 CL 1 3.33 0.69 0.41 Period 1 46.85 9.74 0.003 CL X period iP = 0.81 — ns; not included in final model) Error 69 4.81 Total 71 0.691 > 0.658 1999 > 1991 B Regression models: ED = a + b CL 1991 i5:L> = 31.1646 -I- 0.0230 CL, r2=0.021. n=32. P<0.001, SEa=2.1224, SE 6=0.0282 1999 ED = 33.5760 -i- 0.0128 CL, r2=0.006. ?!=40, P<0.001, SEa=2.1768, SE 6=0.0274 Discussion Size-speciFic fecundity and egg size Fecundity The initial 16% increase in body size-specific fecundity between 1978-81 and 1991 occurred while commercial CPUE decreased fivefold. Unlike commercial data, research CPUE data were collected at fixed sta- tions (including juvenile habitat), were uninfluenced by increased catchability (the targeting of larger adult lobster in more productive habitats by commercial fishermen ), and continued to show a decline of similar magnitude during the 1991-99 period when size-specific fecundity increased an additional 18''i (Fig. IB). Thus both observed fecundity responses occurred simultaneously with declining lobster densities. The cumulative 36% increase in size-specific fecundity observed for Necker Bank P. marginatus over a >20-yr period of exploitation is not unreasonable given the evidence for concurrent, compensatory declines in body size at sexual maturity in this population (Polovina, DeMartini et al : Population density, fecundity, and egg size of Panulirus marginatus 27 1989; DeMartini et al.^). Density-dependent changes in somatic growth, survival rates, and body sizes at sexual maturity have been described for numerous other palinurid spe- cies (Pollock, 1995a, 1995b). At least one case study provides further evidence for reproduc- tive compensation. Chittleborough (1979) doc- umented a decreased interval between broods as a response to increased exploitation in the Western Australian rock lobster, P. cygnus. Prior to the present study, the study of De- Martini et al. (1993) was the only published record of changes in size-specific fecundity in a spiny lobster, perhaps attributable to density declines resulting from exploitation, although perhaps only reflecting natural interannual variation independent of fishing (Pollock, 1995b). The fecundity update for 1999 in this ar- ticle further supports DeMartini et al.'s (1993) original interpretation of an increase in size- specific fecundity as a density-dependent re- sponse at lower population densities. Other data on body size at sexual maturity for the period 1988-99, to be reported elsewhere (De- Martini et al.'), extend the temporal pattern of smaller body size at sexual maturity first documented during 1986-87, after 10 years of exploitation, by Polovina (1989). The observed decrease in size at maturity could have been caused by slower growth (Pollock, 1995a) re- sulting from lower levels of oceanic productivity (Polovina et al., 1994, 1995). However, if smaller size at maturity has been a proximal response to decreased rather than in- creased per capita food availability, it is inconsistent with the simultaneous increases in size-specific fecundity and egg size which have occurred. Evidence for changes in the nutritional status of P marginatus at Necker Bank dur- ing 1991-95 is equivocal (Parrish and Martinelli-Liedtke, 1999). Resolution of whether the lower densities of spiny lobsters at Necker Bank have resulted from natural de- clines in productivity, increased fishery exploitation (or both) would require comparative evaluations for lobsters collected from fished as well as unfished control areas at the bank; unfortunately, as of 1999 unfished lobster habi- tat at Necker Bank does not exist. The observed increase in size-specific reproductive output of P. marginatus probably has been a phenotypic response to lower densities and higher per capita food availabilities at Necker Bank. It is unlikely, given the 3-yr generation time of P. marginatus (Uchida and Tagami, 1984) and relatively short (20-1- yr) period over which the responses have occurred, that a genetic, rather than phenotypic, dynamic has been involved. More extensive comparisons of the egg productions of P. marginatus popu- lations among Necker and other NWHI banks differing in natural and fishery-induced densities would be necessary • 1999 ^ O 1991 ® 1978-81 J^ 13 - 73 7 2 0.11 0.09 782 0.01 0.08 242.3 Total 6226 2340 100.00 100.00 100.00 for all-sized females has been observed in two Caribbean species, Panulirus inflatus (Gracia, 1985) and P. argiis (Fonseca-Larios and Briones-Fourzan, 1998). Briones- Fourzan and Contreras-Ortiz ( 1999), however, could detect no difference in egg size among P. guttatus sampled during three consecutive years. Mean egg size declined within the spawning season, but this decline reflected a loss in mass caused by embryonic development within individual eggs rather than the production of smaller eggs later in the breeding season (Briones-Fourzan and Contreras-Ortiz, DeMartini et al.; Population density, fecundity, and egg size of Panulirus marginatus 29 1999). Pollock {1995c) noted that P. guttatiis produces unusually few, but large eggs for a shallow-water tropical palinurid. Using the CL-to-body-weight regression listed in Table 1, we estimated an inverse index of egg size (Pollock, 1997) for Necker Bank P. marginatus in 1999 that was 660 eggs per g total body weight. Such small eggs are typical within the derived lineage of shallow-water, subtropical and tropical members of the genus (Pollock, 1997). We could find no other studies documenting changes in egg size as a response to density fluctuation in palinurid lobsters. Prior to the mid-1990s, information on temporal and size-related patterns of fecundity and egg size were largely restricted to cold- and warm-temperate members of the genus Jasi/s and Panulinif; (Pollock, 1995c, 1997). Per- haps egg size, like size-specific fecundity, is phenotypically labile in tropical reef species of the genus Panulirus for which high and variable predation pressure makes such plastic responses adaptive. More research on size-specific, individual reproductive output is needed for P. marginatus and other tropical reef species of spiny lobsters. It is unknown whether egg size lability in P. marginatus has a genetic or environmental basis. One could perhaps evaluate this for individual females by repetitively mea- suring egg subsamples from successively brooded egg masses of berried tagged and recaptured females. Fixed but differing egg sizes among individual females would be consistent with a genetic basis. On the other hand, changes in the size of eggs produced by the same individual female in successive broods would suggest that environmental factors are involved. Management implications One of our observations has major relevance to the man- agement of P. marginatus in the NWHI lobster fishery. Based on the IRP of Kanciruk and Herrnkind (1976), egg production by the Necker Bank population of P. margin- atus was dominated by the 50-57 mm TW classes in 1999, which together contributed >43'7f to population egg pro- duction (Table 4). Even though each large (>60 mm TWi individual produces a disproportionately great number of eggs, large females are now so poorly represented in the population that they no longer drive population egg production (Table 4). The eggs produced by smaller (50- 57 mm TW) females are more important to the population now than before exploitation. In 1996 a "retain all" size policy was established for the commercial fishery, replac- ing a 50-mm-TW minimum size limit used previously, in part because of the high mortality of discarded lobsters (DiNardo et al., 2002). If a commercial lobster fishery with a minimum size limit were to be reinstated in the NWHI, a minimum size larger than the previous (50 mm TW) should be considered. Our findings on the size distribution of population egg production indicate that smaller adult females, which now produce most of the population's eggs, should be further protected, perhaps by using larger escape vents in traps. Doing so would increase total population egg production and might assist in countering recruitment overfishing (Botsford, 1991; Pollock, 1993). Panulirus mar- ginatus production at Necker Bank historically and pres- ently dominates archipelago-wide production by the spe- cies; this production is supported by empirical catch data (DiNardo et al.'^) as well as modeling of its recruitment dynamics (Polovina et al., 1999). Augmenting egg produc- tion by the Necker Bank population might significantly bolster stock-wide productivity. The body size distribution of egg production by P. marginatus at other NWHI banks is presently unknown, however, and egg production by large females elsewhere possibly could partly offset the deficit in production at Necker. Our observations on the size distribution of egg production at Necker Bank none- theless merit important consideration for setting size limits for spiny lobster management. By necessity we calculated the IRP assuming that all size classes produced the same (single) brood per spawn- ing period because data on size-specific spawning frequen- cy were lacking. We caution that, if females >60 mm TW (whose size-specific egg production is greatest) produce broods more frequently than smaller females (Lipcius, 1985), we have proportionately underestimated the con- tribution of larger females to population egg production. Individual Panulirus marginatus of all sizes likely produce multiple broods per individual spawning season, based on the protracted period during which females are berried (Uchida and Tagami, 1984; Polovina and Moffitt, 1995) and the occasional presence of new, intact (unused) spermatophore plates on spent females (unpubl. data, Honolulu Laboratory, NMFS). (The latter observation in fact suggests that Necker Bank P. marginatus can pro- duce more than one brood per molt [like P. argus; Sutcliffe, 1953].) There are no time-series growth-rate data avail- able with which to evaluate whether females of a given body size might now be producing larger broods at more frequent intervals than previously. If females are now growing faster, it is likely that the rates of both molting and brood production are now greater Accurate estimates of individual spawning frequencies and how these might differ among females of varying body sizes, would be needed to fully describe the compensatory increase in reproduction which has occurred for the Necker Bank population of P. marginatus. Acknowledgments We thank several anonymous fishery observers for collec- tion of invaluable commercial catch data and R. Moffitt, J. Polovina, and an anonymous reviewer for constructive criticisms of the manuscript. - DiNardo, G. T, W. R. Haight, and J. A. Wetherall. 1998. Sta- tus of lobster stocks in tfie Northwestern Hawaiian Islands, 1995-97, and outlook for 1998. Southwest Fish. Sci. Cent. Admin. Rep. H-98-05, 35 p. Honolulu Laboratory, Southwest Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Honolulu, HI 96822-2.396. 30 Fishery Bulletin 101(1) Literature cited Annala, J. H. 1991. Factors influencing fecundity and population egg production of Jasus species. In Crustacean egg produc- tion, Crustacean issues 7 (A. Wenner and A. Kuris, eds.), p. 301-315. Balkema, Rotterdam. Botsford, L. W. 1991. Crustacean egg production and fisheries manage- ment In Crustacean egg production. Crustacean issues 7 (A. Wenner and A. Kuris, eds.), p. 379-394. Balkema, Rotterdam. Briones-Fourzan, P.. and G. Contreras-Ortiz. 1999. Reproduction of the spiny lobster Panulirus gitttatus (Decapoda: Palinuridae) on the Caribbean coast of Mexico. J. Crust. Biol. 19:171-179. Chittleborough, R. G. 1979. Natural regulation of the population of Panulirus longipes cvgnus George and responses to fishing pressure. Rapp. P-V. Reun. Cons. Perm. Int. Explor. Mer 175:217- 221. Chubb, C. F 2000 Reproductive biology: issues for management. In Spiny lobsters: fisheries and culture, 2"0.05; 1999: x-=^.2A, df=3, P>0.05) and for Kailua Bay (1998: ^-=5.74, df=3, P>0.05; 1999: ;f-=7.14, df=3, P>0.05). Angling (pole and line fishing) accounted for the great- est number of hatchery-reared and wild Pacific threadfin acquired in the reward program but had the lowest CPUE among gear types (Table 5 ). Over both survey years, for wild and hatchery-reared fish combined, angling accounted for 63.0% of the total Pacific threadfin catch, followed by gill- nets ( 19.2% ), thrownets ( 13.9% ), and surround nets (3.9% ). Length of Pacific threadfin caught varied among gear types, between years, and between hatchery-reared and wild Pacific threadfin. For 1998, mean size of Pacific threadfin captured was significantly different among gear types (P=6.378, df=3, 632, P<0.001) and between hatch- ery-reared and wild fish (P=11.833, df=l, 632, P<0.001). Gillnets tended to catch larger fish, and surround nets captured the smallest fish (Tukey multiple comparison test results — gillnets: 255.5 mm > angling: 235.1 mm > thrownet: 214.0 mm > surround nets: 193.7 mm). Mean length for hatchery-reared Pacific threadfin pooled over all gear types was 207.0 mm (SD=28.6) and mean length for wild Pacific threadfin was 242.2 mm (SD=52.7). 38 Fishery Bulletin 101(1) A 1998 N Kahana Bay release site <-, Kaneohe Bay B 1999 Maleakahana (, Kahana Bay release site \ Waimanalo Figure 3 Movement patterns of hatchery-reared Pacific Ihreadfin released in Kahana Bay from the reward fishery program during the lAi 1998, n - 20 and (B) 1999. n = 40 fishing seasons. Friedlander and Zlemann: Impact of hatchery releases on recruitment of Polydactylus sexfilis 39 A 1998 B 1999 N ^\ Kallua Bay release site r^ Lanikai Waimanalo Sandy Beach Maleakahana Kailua Bay release site Figure 4 Movement patterns of hatchery-reared Pacific threadfin released in Kailua Bay from the reward fishery program during the (A) 1998, n = 43 and (B) 1999, n = 57 fishing seasons. Information on sex of Pacific threadfin obtained in the reward program was available only during 1999 because whole fish were not acquired in 1998. The ratio of male to hermaphrodite to female was 56:17:27 for wild Pacific threadfin and 81:15:4 for hatchery-reared fish (Fig. 5). Three hatchery-reared fish that had changed from males 40 Fishery Bulletin 101(1) Table 5 (\atch per unit of effort (CPUE) for Pacific Mean rank computed for Kruskal-Wallis dure. Gear types with the same letter (A, threadfin caught by three different gear types used in rank sum test (//=19.303, df=2, P<0.001). Results of E B) are not significantly different. the recreational-artisanal fishery, unn's multiple comparison proce- Gear type n Total hours Total no. of threadfin No hatchery-reared Mean CPUE SD CPUE Dunn's multiple comparisons Thrownet 16 55.0 166 7 5.49 9.34 A Gillnet 31 83.5 221 20 3.52 4.89 AB Angling 113 592.5 817 56 1.63 1.44 B Grand total 160 731.0 1204 83 2.38 3.98 Comparison of size (mm PL) and sex for h 1999. Table 6 atchery-reared and wild catch Pacific threadfin acquired from the reward fishery in Hatchery-reared Wild Sex Mean Min. Max. n Mean Min. Max. n Males 257.1 Hermaphrodites 267.2 Females 311.3 210 220 247 307 323 356 61 11 3 250.0 275.0 315.6 174 200 249 360 372 380 112 462 179 to females were recovered during 1999; one was released during summer 1996, the other two were released during summer 1997. Mean size for males was 250.0 mm (SD=32.1) for wild fish and 257.1 mm (SD=24.8) for hatchery-reared fish; mean size for hermaphrodites was 275.0 mm (SD=27.5) for wild fish and 267.2 mm (SD=43.3) for hatchery-reared fish; mean size for females was 315.6 mm (SD=39.2) for wild Pacific threadfin and 311.3 mm (SD=57.1) for hatchery-reared threadfin (Table 6). The GSI for hatchery- reared males (.v=0.629, SD=0.728) was not significantly different (7=6524, P=0.073) than the GSI for wild males (.v=0.619, SD=0.842), likely because of the larger size of hatchery-reared males during 1999. Number of hatchery- reared females and hermaphrodites was too low for statis- tical comparisons. Condition factor for hatchery-reared Pacific threadfin during 1999 was not significantly different between re- lease sites (F=0.074, df=l, 79, P=0.786) or among release sizes (F=1.488, df=3, 79, P=0.224). Therefore, condition factors for ail hatchery-reared fish were pooled and com- pared to condition factors for all wild fish recovered in 1999. No significant difference was found in condition fac- tors between the.se two groups ( 7=47733.0, P=0.087). Discussion Cultured Pacific threadfin juveniles released into the ocean survived and recruited successfullv into the recreational Sex Figure 5 Sex ratio for wild and hatchery-reared Pacific threadfin returned for the 1999 reward fishing sea.son. Values are total number offish in each sex category. fishery, accounting for Wf and 8*^; of the catch on the windward side of the island of Oahu in two years ( 1998 and 1999, respectively). Hatchery fish from the 1997 release constituted the majority of the hatchery fish returns to the recreational fishery in 1998 (89.4':^ ) and 1999 (95.9'^; ). Few of the hatchery fish released in years prior to 1997 have been recovered from the recreational fishery. The large Fne(dlander and Ziemann: Impact of hatchery releases on recruitment of Polydactylus sexftlts 41 Comparisons of length of juveniles, males for wild fish from the 1999 reward fishery Table 7 hermaphrodites, and females for Pacific threadfin around Oahu from 1962 to 1968 and program. One standard deviation of mean fork length is shown in parentheses. Sex 1962-68, n = 1651 1999 reward program, n=1105 Mann Whitney T-value Fork length (mm) Percentage of total Fork length (mm) Percentage of total P Juveniles Males Hermaphrodites Females 227(30) 268(29) 317(33) 378(45) 6.4 52.3 17.8 23.5 191(24) 249(35) 275(27) 316(39) 39.7 33.8 10.3 16.2 44864 184831 11952 26200 A A A A O O O O b b b b o o o o impact of a relatively small number of released fish on the recreational fishery shows that hatchery releases of limited numbers of fish have the potential to impact both the number of fish taken in the fishery and the rate at which the fishery can recover. The differences in contribu- tion rates for different release years suggest that natural factors affecting the survival of juveniles, as well as early larval stages, vary between years. Hatchery-reared and released fish collected in the rec- reational fishery showed growth rates, condition factors, and gonadosomal indices similar to those of wild fish, suggesting that hatchery-reared fish are able to adapt to the natural environment and integrate into the wild population. Our data (unpubl.) for wild and hatchery fish collected in nursery habitats showed no significant differ- ences in growth rates. The mean size of hatchery-reared fish collected in 1998 was smaller than in 1999 (over 95% of the fish collected in 1998 and 1999 came from the same releases in 1997). Mean size for hatchery-reared fish in 1999 was not different from the mean size of wild fish for both years, which suggests that size of hatchery fish in 1999 represents the approximate size of 2-3 year-old threadfin and mean age of fish in the recreational fish- ery is also 2-3 years. The size-frequency distributions of hatchery and wild fish in 1998 and 1999 suggest that a significant portion of the wild fish in the fishery is younger than two years. Small hatchery fish at release made a higher relative contribution to the recreational fishery than did the larger size group (but not significantly so, except for fish taken in Kailua Bay in 1999), and the nursery habitat sampling conducted after the 1997 releases showed the same (Leber et al., 1998; Ziemann et al.'-). This pattern is in contrast to that observed for mullet in Hawaii (Leber, 19951 and Pacific threadfin for other years (Ziemann et al.'^) Hatchery fish disperse slowly from the point of release along the windward coast of Oahu. In nursery habitats three months after release (Ziemann et al.-i, hatchery fish represented in excess of 70'7( of the threadfin and they decreased within nine months to 10% or less. Some decrease is due to predation, but some is due to dispersal because in 1998, after 1 year at large, fish were caught in the recreational fishery a mean distance of 11.2 km from the release point, and after two years, mean distance had increased to 15.2 km. Dispersal from the two release sites differed: after one year mean distance for Kahana Bay releases was 14.6 km, whereas mean distance for fish releases in Kailua Bay was 9.6 km. The 1999 reward sample contained 16% females, 44% males and hermaphrodites, and 40% immature fish. The life cycle of Pacific threadfin (protandric hermaphrodites) makes this skewed sex ratio even more problematic be- cause individuals do not become functional females until about 30 cm FL and these larger fish are selectively re- moved from the population by fishing. For protogynous species, size-selective fishing mortality may result in differential loss of larger males (Sadovy, 1996; Beets and Friedlander, 1999). The percentage of juveniles in the catch was high. Mean size of Pacific threadfin in all sexual categories was significantly smaller than that reported by Kanayama* in 1962-68 (Table 7, Fig. 6); further, females constituted 23.5% of the catch in the 1960s, but only 16.2% of the catch in 1999. We demonstrated that cultured Pa- cific threadfin juveniles released in known nursery habi- tats survive and recruit successfully into the recreational fishery 1-2 years later. Our Pacific threadfin data indicate that recruitment of young fish to the population may be jeopardized because there are few mature females left in the population (recruitment overfishing), even with supplementation of hatchery-reared fish. The underlying problem of the threadfin fishery on Oa- hu and the other Hawaiian Islands is primarily an intense local harvest by subsistence and recreational fishermen, as well as habitat loss from coastal and upland develop- ment. Current state regulations, as well as unregulated removal of larger individuals from the population, contrib- ute to the male-biased sex ratios observed in our study. Stock recovery based on natural reproduction will be a long-term process. Implementation of an enhancement progi-am for Pacific threadfin focused on juveniles and perhaps larger females could speed the rate of recovery of the local population. Kanayama, R. 1973. Life history aspects of the moi Po/vrfac- tylus sexfilix in Hawaii, 50 p. State of Hawaii, Department of Lands and Natural Resources, Honolulu, Hawaii. 42 Fishery Bulletin 101(1) 0-25 1 A 1962-68 0.20 0.15 0.10 0.05 Z 0.00 0.25 0.20 0.15 0.10 0.05 I I Juveniles ^y////////i Males ^^^m Hermaphrodites ^^^ Females 0.00 10 30 40 Fork length (cm) Figure 6 Proportion of juveniles, males, hermaphrodites, and females for Pacific thread- fin from I A) 1962 to 1968 and (B) during the 1999 reward fishery program. Acknowledgments The authors acknowledge the contributions to this re- search made by Ken Leber, Peter Craig, Reiji Masuda, Robert Cantrell, Steve Arce, Scott Bloom, Tom Ogawa, Don Dela Pena, Rich Hall, Karl Keller and other members of the slock managenii'nl staffat The Oceanic Institute, and of the culture support provided by Tony Ostrowski and the staff of The Oceanic Institute finfish program. Jim Parrish, Reiji Masuda, Ken Leber, two anonymous reviewers and editors provided valuable .suggestions for the manuscript. This research was supported under NOAA gi-ant NA7(jFY0()59. Literature cited Anderson, R. O., and R .M. Neumann. 1996. Length, weight, and associated structural indices. In Fisheries techniques, 2"'' ed. (B. R. Murphy and D.W. Willis, eds. ), p. 447-482. Am. Fi.sh. Soc, Bethesda, MD. Blankenship, H. L., and K. M. Leber 1995. A responsible approach to marine stock enhancement. Am. Fish. Soc. Symp. 15:165-175. Beets, J., and A. Friedlander 1999. Evaluation of a con.servation strategy: a spawning aggregation closure for grouper in the Virgin Islands. En- viron. Biol. Fish. 55:91-98. Fnedlander and Ziemann: Impact of hatchery releases on recruitment of Polydactylus sexfilis 43 Bleeker, P. 1875. Recherches sur la fauna de Madagascar et de ses dependances d"apres les descouvertes de Francois P. L. Pollen et D. C. van Dam. 4" Parte. Poissons de Madagascar et de rie de la Reunion. Leiden, The Netherlands. Grimes, C. B. 1998. Marine stock enhancement: sound management or techno-arrogance? Fisheries 23(9):18-23. Howell, B. R., E. Moksness, and T. Svasand (eds.). 1999. Stock enhancement and sea ranching, 606 p. Fish- ing News Books. Oxford, England. Hosaka, E. Y. 1990. Shore fishing in Hawaii, 175 p. Petroglyph Press, Ltd., Hilo, HI, Imai. T, H. Takama, and I. Shibata. 1994. Estimates of the total amount of red sea bream caught by recreational party boats in Kanagawa Prefecture. Sai- bai Giken 23:77-83. |In Japanese.] Jefferts, K. B., P. K. Bergman and H. F Fiscus. 1963. A coded-wire identification system for macro-organ- isms. Nature (London) 198:460-462. Kaeriyama, M. 1996. Population dynamics and stock management of hatchery-reared salmons in Japan. Bull. Natl. Res. Inst. Aquacult. Suppl. 2:11-15. Kitada, S., Y. Taga, and H. Kishino. 1992. Effectiveness of a stock enhancement program evalu- ated by a two-stage sampling survey of commercial land- ings. Can. J. Fish. Aquat. Sci. 49:1573-1582. Leber, K. M. 1995. Significance of fish size-at-release on enhancement of striped mullet fisheries in Hawaii. J. World. Aquacult. Soc.,26(2):143-153. Leber, K. M. and S. M. Arce. 1996. Stock enhancement in a commercial mullet, Mugil cephalus L., fishery in Hawaii. Fish. Manage. Ecology 3: 261-278. Leber, K. M, S. M. Arce, D. A. Sterritt, and N. P. Brennan. 1996. Marine stock-enhancement potential in nursery habi- tats of striped mullet, Mugil cephalus. in Hawaii. Fish. Bull. 94:452-471. Leber, K. M., N. P. Brennan , and S. M. Arce. 1998. Recruitment patterns of cultured juvenile Pacific thread- fin, Polydactylus sexfilis (Polynemidae), released along sandy marine shores in Hawaii. Bull. Mar Sci. 62:389^08. Lowell, N. 1971. Some aspects of the life history and spawning of the moi [Polydactylus sexfilis). M.A. thesis, 45 p. Univ. Hawaii, Honolulu, HI. Masuda, R., and K. Tsukamoto. 1998. Stock enhancement in Japan: review and perspective. Bull. Mar Sci. 62:337-358. McEachron, L. W., R. L. Colura, B. W. Bumguardner, and R. Ward. 1998. Survival of stocked red drum in Texas. Bull. Mar Sci. 62:359-368. McEachron, L. W., and K. Daniels. 1995. Red drum in Texas: a success story in partnership and commitment. Fisheries 20:6-8. Myers, R. F 1991. Micronesian reef fishes: a practical guide to the iden- tification of the coral reef fishes of the tropical central and western Pacific, 298 p. Coral Graphics, Barrigada, Guam. Ostrowski, A., T. Iwai, S. Monahan, S. Unger, D. Dagdagan, P. Murawaka, A. Schivell, and C. Pigao. 1996. Nursery production technology for Pacific threadfin {Polydactylus sexfilis). Aquaeulture 139:19-29. Randall. R. E. 1996. Shore fishes of Hawaii, 216 p. Natural Worid Press, Vida, OR. Randall, J. E., G. R. Allen, and R. C. Steene. 1990. Fishes of the Great Barrier Reef and Coral Sea, 507 p. Univ. Hawaii Press, Honolulu, HI. Sadovy, Y. J. 1996. Reproduction of reef fishery species. In Reef fish- eries (N. V. C. Polunin and C. M. Roberts, eds.), p. 15-59. Chapman and Hall, London. Santerre, M. J., G. S. Akiyama, and R. C. May. 1979. Lunar spawning of the threadfin, Polydactylus sexfi- lis, in Hawaii. Fish. Bull. 76:900-904. Santerre, M. J., and R. C. May 1977. Some effects of temperature and salinity on labora- tory-reared eggs and larvae of Polydactylus sexfilis (Pisces: Polynemidae). Aquaeulture 10:341-351. Schramm, H. L., Jr, and R. G. Piper, eds. 1995. Uses and effects of cultured fishes in aquatic ecosys- tems, 608 p. Am. Fish. Soc. Symp. 15. Sokol, R. R., and F J. Rohlf 1981. Biometry, 859 p. WH. Freeman, San Francisco, CA. Szyper, J. P, M. J. Anderson, and N. H. Richman. 1991. Preliminary aquaeulture evaluation of moi {Polydac- tylus sexfilis ). The Progressive Fish-Culturist 53:2025. Tinker, S. W. 1982. Fishes of Hawaii, 532 p. Hawaiian Services, Inc., Honolulu, HI. Titcomb, M. 1972. Native use of fi.sh in Hawaii, 175 p. LTniv. Hawaii Press. Honolulu, HI. 44 Abstract — A general model for yield- per -recruit analysis of rotational (per- iodic) fisheries is developed and ap- plied to the sea scallop (Placopecten magellanicus) fishery of the northwest Atlantic. Rotational fishing slightly increases both yield- and biomass-per- recruit for sea scallops at F^y^. These quantities decline less quickly when fishing mortality is in-creased beyond ^MAX than when fishing is at a constant rate. The improvement in biomass- per-recruit appears to be nearly inde- pendent of the selectivity pattern but increased size-at-entry can reduce or eliminate the yield-per-recruit advan- tage of rotation. Area closures and rota- tional fishing can cause difficulties with the use of standard spatially averaged fishing mortality metrics and reference points. The concept of temporally aver- aged fishing mortality is introduced as one that is more appropriate for seden- tary resources when fishing mortality varies in time and space. Yield- and biomass-per-recmit analysis for rotational fisheries, with an application to the Atlantic sea scallop (Placopecten magellanicus) Deborah R. Hart Northeast Fisheries Science Center 166 Water St Woods Hole, MA 02543 E mail address Deborati HartsSnoaa gov There has been growing interest in using rotational fishing to manage ses- sile or sedentary stocks (e.g. Caddy and Seijo, 1998). Under such a strategy, fish- ing mortality in a given area is varied periodically. Typically, the area is closed for a period of time, then fished, and then closed again. The openings of the different areas are timed so that at least one area is open to fishing each year. This approach has been proposed or is being used for abalone, corals, sea cucumbers, geoduck clams, sea urchins, and several species of scallops (Sluc- zanowski, 1984; Garcia, 1984; Botsford et al., 1993; Caddy, 1993; Heizer, 1993; Campbell et al., 1998; Caddy and Seijo, 1998; Lai and Bradbury, 1998). Recently, area closures have been used to help manage the Atlantic sea scallop (Placopecten magellanicus) fish- ery off the northeastern United States. Three areas on Georges Bank were closed to scallop and groundfish fish- ing in December 1994 to help protect depleted groundfish resources. Subse- quently, there have been substantial in- creases in scallop abundance, biomass, and mean size in these areas; mean scallop biomass in the closed areas, as measured by the Northeast Fisheries Science Center (NEFSC) sea scallop survey, rose from 0.6 kg/tow in 1994 to 15.8 kg/tow in 2000.' During limited Manuscript accepted 20 September 2002. Fish. Bull. 101:44-.57(2003). ' NEFSC (Northeast Fisheries Science Cen- ter). 2001. Report of the 32nd north- east regional stock assessment workshop (32nd SAW). Stock Assessment Review Committee (SARC) consensus summary of assessments. NEFSC Ref Doc. 01-05. 289 p. (Available from NEFSC, 166 Wa- ter St., Woods Hole MA 02M:\.\ openings of these areas to fishing in 1999 and 2000, nearly 5000 metric tons (t) of scallop meats (about 20'* of the total landings during this period) were landed, while biomass levels remained high. In April 1998, two areas in the Mid-Atlantic Bight were closed to scal- lop fishing for three years in order to protect high concentrations of juvenile scallops. Scallop biomass has increased markedly since the closures in these areas as well, from 0.8 kg/tow in 1997 to 9.7 kg/tow in 2000^ and about 3500 t of scallop meats have been landed in these areas in the year since they were reopened in May 2001. These data suggest that temporary or rotational closures can help increase scallop bio- mass and yield. For these reasons, a rotational management system for the U.S. Atlantic sea scallop fishery is cur- rently under consideration. Many common fisheries models may not be appropriate for sessile stocks because these models assume spati- ally uniform fishing mortality (Caddy, 1975). Such a "dynamic pool" assump- tion is strongly violated when a sessile stock is fished rotationally so that a portion of the stock is not fished in a given year. For this reason, many previous analyses of rotational fisher- ies have used either spatially explicit simulations (e.g. Caddy and Seijo, 1998), per- recruit analyses of pulse fishing, where all vulnerable individu- als are removed from an area when the area is fished (e.g. Sluczanowski, 1984), or per-recruit analyses of a single co- hort (e.g. Gribble and Dredge, 1994). Spatially explicit models suffer from then- complexity, making it difficult to extract general principles from model Hart: Yield- and biomass-per-recruit analysis of rotational fisheries 45 simulations. Analysis of pulse fishing, although simple to apply and understand, is not applicable to those situations where only a portion of the available resource is removed when an area is opened periodically to fishing. Per-recruit analysis of a single cohort is not applicable to relatively nonselective multiple age-group fisheries. Botsford et al. (1993) developed a mixed-age rotational yield-per-recruit model for red sea urchins. They showed that rotational fishing for these urchins would increase egg production considerably, while slightly decreasing yield-per-recruit. Recently, Myers et al. (2000) presented a mixed-age per-recruit analysis of a possible rotational fishery strategy for sea scallops. The emphasis of this study was on the effect of putative high levels of indirect (noncatch) fishing mortality on yield-per-recruit, and on a proposed rotational plan that Myers et al. suggested would help ameliorate this effect. The purpose of the present article is to present a general theory for any type of periodic or rotational fishing strat- egy for a mixed-age sessile or sedentary stock. This work generalizes many of the above mentioned studies (in par- ticular, that of Botsford et al., 1993) and does not require an assumption of constant recruitment or specific spatial configuration (or both). This theory is applied to the Atlan- tic sea scallop fishery of Georges Bank. Measures of fishing mortality and overfishing defini- tions are usually based on models where fishing is as- sumed constant in space and time. In rotational fisheries, or in cases where part of a fishing ground has been closed indefinitely to fishing, these assumptions may be seriously violated, especially for stocks that are relatively immobile. Alternative measures of fishing effort and overfishing definitions are presented here that are more appropriate to fisheries of nonmobile stocks where rotational or indefi- nite closures are used. Methods The object of this analysis was to compute the expected yield-per-recruit and biomass-per-recruit of a cohort located in an area where fishing mortality may depend on the year and the variation in fishing mortality is periodic with time. Rotational fishing is usually thought of as a sequence of periodic closures and openings of different areas. The theory described here is more general, and can be applied to any situation where fishing mortality is varied periodically in a given area. Suppose the fully recruited fishing mortality in an area during year k is F^ and that fishing mortality rates vary pe- riodically with period p ( where p is in years ), so that F , = Ff. for all k. Let ^wi; be the mean of F^.F.^ F^^ For sim- plicity, it is assumed that there is one recruitment event and one new cohort per year. However, extension of the theory to multiple cohorts per year is straightforward. There are p possible patterns of fishing mortality expe- rienced by a cohort, depending on the point of the cycle when it enters the fishery. The cohort that enters in the first year will experience fully recruited fishing mortality rates: F F F F F F F (1) during successive years. The next cohort will experience the same fishing mortality rates, but in a different order: F„F„F„...,F^.F,K. (2) and so on. Two special cases are of particular interest: pulse rota- tion and symmetric rotation. Pulse rotation means that F^ = 0 for ^=l,2,...,p-l (the area is closed for p-1 years), then F >0 (the area is pulse fished for one year), and then Ff^=0 for k=p-¥l,p+2,..., 2p-l (the area is closed again), etc. Symmetric rotation, where p is even, means that F/, = 0 for 1 < ^ < p/2, and Ff. = 2F^v^. for p/2 < k < p, i.e. the area is closed for p/2 years and then fished at a constant rate for the next p/2 years. For each of the p patterns of fishing mortality, yield- and biomass-per-recruit can be calculated by using standard per-recruit techniques. Here, a method similar to the "ge- neric per-recruit" model described in Quinn and Deriso (1999) is used (see Appendix). The only unusual aspect is that the mortality terms Z and F^. in Equations 11-13 (see Appendix) depend explicitly on time, i.e. on the year of the rotational cycle. Each of the p cohorts will have differ- ent yield-per-recruit Y^,Y.^...,Y , and biomass-per-recruit ByB.^.-Mp values because the ages at which they experi- ence the fishing mortalities F^,F2,..-,F are different. Define y^VG ^"^^ ^avg ^^ be the means of the p patterns of cohort yield- and biomass-per recruit, respectively. Y^yg is the expected yield of a recruit chosen randomly with re- spect to cohort. In other words, F^vc '^ ^^^ long-term mean yield-per-recruit that can be expected from the rotational fishing strategy. Similarly, Bavg '^ the expected long-term mean biomass-per-recruit. Note that unlike conventional per-recruit theory, yield- and biomass-per-recruit vary with cohort, so that the mean yield- and biomass-per-re- cruit obtained at any point in time may be different from ^AVG and Bavg- Let y^',y2',...,y'P' be yield-per-recruit of the p cohorts, in decreasing order, so that y" is the highest yield-per- recruit out of all the p cohorts and Y'l'' the lowest, y" is an upper bound on the yield-per-recruit that might be obtained with a rotational strategy if, for example, the closure pattern were timed to optimize yield-per-recruit from a large year class. It is important when comparing rotational and con- stant fishing strategies to compare alternatives that have the same long-term survival rates, i.e the same natural mortalities and mean fishing mortality rates. If this is not done, then effects of rotation can be confounded with those due to variations in fishing mortalities. If there are initially N^ fully recruited individuals in an area that are fished at a constant rate F„, then there will be N=N,,exp{-pM-pFj (3) of these individuals remaining alive after p years. If instead, fishing mortality was varied on ap year rotation, so that in each year of the cycle, fishing mortality in an 46 Fishery Bulletin 101(1) Table 1 Estimated life history parameters for Georges Bank sea scallops. Von Bertalanffy growth parameters are from Serchuk et al. ( 1979). Relations of shell height (SH) to meat weight (MW) (see Eq. 7) were obtained by combining the data of Serchuk and Rak' with that of NEFSC (Footnote 2 in the general text). The natural mortality estimate is from Merrill and Posgay (1964). The selectivity pattern is based on the current gear configuration of scallop dredges with 89-mm rings (NEFSC, Footnotes 1 and 2 in the general text). M(/vr) a (In g) b ^mmtmm' ''full <™ni' h^ (mm) d KUyr) L^ ( mm ) (natural (SHMW (SHMW (MinSH (SH for full (cull (discard (growth) ( growth ) mortality) parameter) parameter) selected) selectivity) size) mortality Value 0.3374 152.46 0.1 -11.6038 3.1221 65 75 0.2 Serchuk, F. M, and R, S. Rak. 1983. Biological characteristics of offshore Gulf of Maine scallop populations: size distribution, relations of shell height to meat weight, and relative fecunditv patterns. Reference document 83-07, 42 p. [Available from Northeast Fisheries Science Center, 166 Water St.. Woods Hole, MA 02543.1 area is given by Fj, F.^ of individuals remaining alive after p years would be F , respectively, then the number N'p = No exp -pM-Y^F, (4) In order for the long-term survivorship of the two strate- gies to be equal (i.e., A^ =N^' ), the uniform fishing mortal- ity F|^ must equal the average fishing mortality AVG = it.. P L 1=1 (5) of the rotation plan. Therefore, F^vc. ^^ used to scale all the graphs and per-recruit comparisons. The model described above and in the Appendix was implemented as a Fortran-90 program where the integrals were numerically computed with a time step of 0.01 y. Parameters used in the model are given in Table 1 and represent estimates for growth and mortality of Atlantic sea scallops (Placopecten magellanicus), for which rota- tional management is currently under consideration. Results Yicld-per-recruit curves for no rotation (continuous uni- form fishing), three-year pulse rotation (i.e. the area is closed for two years and fished for one year), six-year pulse rotation, and nine-year pulse rotation are given in Figure 1. Note that the x axis in Figure 1 is the mean fishing mortality rate Fi^vi;- ^^^ they axis is mean (i.e. expected) yield-per-recruit V^vc- averaged over cohorts. Because the mean fishing mortality rate is the same for all points at the same .v coordinate, the three-year rotation has a fish- ing mortality rate during years when fishing occurs (F ) that is throe times as high, and the six-year rotation six times as high, as the constant F i no rotation ) strategy with the same F, AVG- Rotation affects the yield-per-recruit curve for sea scal- lop in three different ways (see Fig. 1). First, rotation modestly increases the maximum mean yield-per-recruit Yf^js^; the maximum mean yield-per-recruit for the nine- year rotation is about 9% greater than without rotation. Second, Fj^l« *i-^- the value of Fj^vc, where Yt^j^x '® °^" tained) increases somewhat under rotation, especially for longer rotation periods. Third, there is less yield penalty in rotational management for exceeding Fiy^a^x. For example, fishing at F = 1 results in a 38% loss of yield if there was no rotation, but only an 8% loss under a six-year pulse rotation. Although 6-yr pulse rotation results in only a 5% increase in yield-per-recruit over no rotation at their re- spective F^y^ values, the advantage of 6-yr pulse rotation atF= 1 is43'7c. Maximum yield-per-recruit for pulse fishing as a func- tion of the rotation period p is shown in Figure 2. The best yield-per-recruit is obtained for long periods of 9 to 10 years. However, this type of strategy would imply that a number of years would pass before any yield would be obtained from most recruits and this strategy would only slightly increase maximum yield-per-recruit over that of steady fishing. Depending on management goals, it might be reasonable to discount future yields, so that the pres- ent value of yield taken t years into the future would be discounted by exp(-&), where 5 is the annual discount rate (assumed 10%/yr). The rotation period that maximizes discounted yield-per-recruit is 6 years (Fig. 2). If prices as a function of meat weight are known, it would also be possible to do a similar analysis to maximize discounted value-per- recruit. Yield isopleths, commonly used to visualize yield-per- recruit analysis (Beverton and Holt, 1957), are given in Figure 3A (yield-per-recruit) and 3B (discounted yield- per-recruit). For rotational analyses, fishing mortality is placed on the .v-axis and rotational period on the y-axis. Note again that for longer rotation times, the decline in yield for fishing mortalities greater than F^,^^ is much less than without rotation. The value of F^^^j^^ and maxi- mum yield-per-recruit increases slightly with longer rota- tion periods. Biomass-per-recruit for no rotation, 3-, 6- and 9-yr pulse rotation strategies is given in Figure 4. Compared to con- stant fishing, rotational fishing gives increased biomass- per-recruit; this increase is most evident for the longer Hart: Yield- and biomass-per-recruit analysis of rotational fisheries 47 No rotation 3 yr rotation 6 yr rotation 9 yr rotation 04 F (/yr) Figure 1 Yield-per-recruit curves for Georges Bank sea scallops with no rotation and with 3- 6-. and 9-yr pulse rotations. rotations and higher fishing mortalities. At F[^i,\x- the increase in biomass-per-recruit is shght, un- less a very long (e.g. 9-yr) rotational period is employed. The performance of rotation can be assessed as a function of selectivity at size. Figure 5A gives maximal yield-per-recruit for a number of pulse rotation strategies and a variety of values of h^^^^, the smallest size selected by the gear; the size of full selectivity, /if^,,, was taken as h^^^^ + 23 mm (consistent with the assumed current gear selec- tivity pattern; see Table 1). Rotation can give sub- stantial yield-per-recruit advantages when the gear selects animals of well below optimal size, especially for longer periods. However, long-pe- riod rotation actually gives less yield-per-recruit than constant fishing for larger values of h^^^. Figure 5B gives a similar plot for biomass-per- recruit, where fully recruited fishing mortality is fixed at F = 0.3 in all cases. Unlike yield-per- recruit, rotational fishing increases biomass-per- recruit regardless of the selectivity pattern, espe- cially when the rotational period is long. Yield-per-recruit from difTerent cohorts under a rotational system can vary considerably. The cohort which recruits into the fishery at about the time of the closure produces the highest yield-per-recruit, whereas the cohort that reaches exploitable size at about the time that the area is opened has the lowest. Figure 6 gives the mean yield-per-recruit together with that of the cohorts with the highest and lowest yield-per-recruit under six- year pulse rotational management (i.e. i^^vc^ ^"'' ^""^ ^*'' • • • • o o o o o 2 4 6 8 10 Rotation period p(yr) Figure 2 Maximum yield-per-recruit (solid circles) and discounted (Wv) yield-per-recruit (open circles) for Georges Bank sea scallops with a pulse rotation of periods between 1 and 1 1 years. respectively). A 319^ increase in maximal yield compared to constant fishing (and 25'7c increase over the average yield- per-recruit under rotation) can be obtained from the cohort whose yield-per-recruit is the highest under rotation. Note that, unlike conventional yield-per-recruit cun'es for sea scallops, yield-per-recruit from this cohort is almost com- pletely insensitive to effort beyond a certain level. 48 Fishery Bulletin 101(1) :!'■ '•(i '■ (i ■■'■'■ (i 10 10 B 0 0 0 2 0 4 0 6 0 8 1,0 Figure 3 Yic'ld-per-rocruit (Ai and discounted ( 10'* ) yield-pcr-rotrult (B) isopletlis for Georges Bank sea scalloi)s with pulse rotation. Note that the v axes represent rotation period. Hart: Yield- and biomass-per-recruit analysis of rotational fisheries 49 250 n \\ \'\ \\ No rotation 200 - V 6 yr rotation 9 yr rotation 150 - 3 J 100 - 50 - X: >^^ ^::^;;;;;:;:rr^^ 0 - 1 1 I 1 0.0 02 0.4 0.6 08 F.J'V') Figure 4 Biomass-per-recruit for Georges Bank sea scallops with no rotation and with 3-, 6-, | and 9-yr pulse rotations. Table 2 Calculated values of F^^^, y^^X' ^maX' ^"""^ discounted iDsc) Yj^j^^^ (in grams, with a 10% discounting factor) for (A) pulse rotation with no incidental fishing mortahty, (B) pulse rotation with 15'7f incidental fishing mortality, and (Cl symmetric rotation with no incidental fishing mortality, p = period of rotation. P ^.iM.v ■'w.u- 5„^v DscY^,^^ P ^MAX ^M.\.\ ^M.\.\ ^^'^ ^ M.W A 1 0.217 17.25 84.0 10.66 B (cont.) 4 0.205 14.81 88.4 9.01 2 0.219 17.27 83.6 10.70 5 0.219 14.99 85.4 9.16 3 0.225 17.38 82.4 10.83 6 0.236 15.20 82.8 9.25 4 0.239 17.57 79.6 11.04 7 0.257 15.43 81.2 9.25 5 0.259 17.84 76.8 11.24 8 0.277 15.63 82.2 9.15 6 0.287 18.17 74.5 11.32 9 0.292 15.75 86.0 8.96 7 0.324 18.47 73.8 11.25 10 0.300 15.8 91.8 8.70 8 0.351 18.71 77.4 11.03 11 0.302 15.75 99.0 8.4 9 0.363 18.84 84.2 10.71 10 0.372 18.82 92.0 10.34 11 0.374 18.69 100.6 9.93 C 2 4 6 0.219 17.27 83.6 10.70 0.225 17.4 82.5 10.82 0.235 17.56 81.0 10.90 B 1 0.192 14.62 91.9 8.79 8 0.244 17.68 81.1 10.85 2 0.193 14.62 91.6 8.80 10 0.248 17.73 83.5 10.68 3 0.197 14.68 90.5 8.88 12 0.253 17.64 86.3 10.44 Yield-per-recruit curves for 6 yr symmetric rotation (i.e. closed for three years and then opened for three years), 10-yr symmetric rotation, 6-yr pulse rotation, and no rotation are given in Figure 7. Symmetric rotation gives yields-per-recruit that lie between that of pulse rotation and constant fishing. Maximum yield-per-recruit. together with the associated B^^x, and maximal discounted yield for pulse rotation, with and without 15'^'i incidental mortal- ity, and for symmetric rotation without incidental mortal- ity, are given for various rotation periods in Table 2. Results from yield-per-recruit runs with incidental fishing mortality (Table 2B) show similar patterns to 50 Fishery Bulletin 101(1) ; 20 No rotation 3 year rotation 6 year rotation 9 year rotation 160 140 120 100 - Figure 5 (A) Yield-per-recruit and (B) biomass-per- recruit for no rotation (solid line), 3-yr (dotted line), 6-yr (dashed line) and 9-yr (dot-dashed line) pulse rotations of sea scallops as a function of minimum selected shell height /i„,|„. Scallops are assumed to be fully recruited to the fishery at shell height /i,,,,^ + 23 mm. results with no incidental fishing mortality (Table 2A). Note that both yields and the value oi F^^.^y^ are reduced if incidental fi.shing mortality exists and that the pen- alty for overfishing without rotation is somewhat higher (about 679f loss in yield-per-recruit for fishing at F=l without rotation compared to 389f without incidental mortality). However, at ^\,,\x• the loss of yield due to inci- dental mortality is about the same for rotational fishing as for steady fishing. Discussion Rotational fishing can generate increased yield- and bio- mass-per-recruit for sea scallops compared to nonrota- tional fishing. The expected increase in maximum vield- per-recruit is modest (<109; ) for a fixed rotational pattern. The over SO'/f gain in yield-per-recruit obtained from cohorts that reached exploitable size near the time of the closure is partially cancelled by the loss of yield-per-recruit Hart: Yield and biomass-per-recruil analysis of rotational fishenes 51 25 -1 20 - 15 - -"''" 3 f" "~ £ 10- > / •"- ^ 5 ■ / 0 - f ^^^ Average cohort Worst yielding cohort 0 — ■ Best yielding cohort 0 0.2 0 4 0 6 0 8 ''wr. cyo Figure 6 Yield-per-recruit for a 6-yr pulse rotation of Georges Bank sea scallops from an average cohort, and those that benefit the most and the least from rotation. on those cohorts that reached exploitable size at about the time the area was reopened, thereby resulting in only a modest gain in yield-per-recruit. A more substantial gain in maximum yield-per-recruit (up to 30% greater than constant fishing) can be obtained if the closure is timed to optimally exploit an unusually large year class. These results are consistent with several studies that indicate that periodic fishing can often increase yields over con- stant fishing (Botsford, 1981; De Klerk and Gatto, 1981; McCallum, 1988; Clark, 1990; Myers et al., 2000). A second, and perhaps more important, advantage of rotational fishing is that it alleviates the impact of both growth and recruitment overfishing. Growth overfishing (i.e. fishing at a level higher than Fj^^j^) under rotational management induces a substantially smaller reduction in yield-per-recruit than would occur with constant fish- ing. Rotation also increases biomass-per-recruit for sea scallops, especially for levels of F above F^^y^, thereby reducing the impact of possible recruitment overfishing. It might be argued that overfishing should not be occurring in any case. However, even when management measures are taken to eliminate overfishing, it can still occur, for example, if 1) reference points are incorrect because of un- certainty in life history parameters; 2) fishing mortality, or the effect of management measures on fishing mortality, has been underestimated; or 3) there is localized overfish- ing because of spatial variation in fishing intensities or life history parameters (or variation in both), even though when averaged spatially, F,^y(. < F^^^ (Caddy, 1975; Hart, 2001). Rotational fishing can thus be thought of as part of a precautionary strategy. In so much as it may increase maximum yield, rotational management is superior to many other precautionary measures that reduce yield. The only costs of rotational management are the costs of administrating and enforcing such a system, and socioeco- nomic costs from temporary closures of traditional fishing grounds. The latter might be significant if closures force fishermen to make long distance steams to unfamiliar areas. Because the optimal F^vg under rotation is only slightly greater than the nonrotational F.^^^^^. the amount of effort and fleet capacity required to optimize yield-per- recruit under rotation is about the same as that needed under uniform fishing. Rotation also imposes practical constraints on the level of average fishing effort, thereby limiting the extent to which stocks can be overfished. Fishing mortality rates for U.S. sea scallop stocks were estimated as exceeding 1.0/yr during the late 1980s and early 1990s. ^ This would corre- spond under a 6-yr pulse rotation to an unaveraged fish- ing mortality of over F = 6 in the area open to fishing. Such a high fishing morality rate, corresponding to about a 98% exploitation rate for fully recruited scallops, is likely to be impractical for both physical and economic reasons. Thus, F.^y^, in a rotation plan would likely be considerably below the high levels seen in the late 80s and early 90s, even if there was no other restriction on fishing effort other than pulse rotation. Myers et al. (2000) claimed that "near-optimal yields are achieved across a wide range of fishing mortalities" in their rotational scheme. However, much of their analysis was confounded by their use of unaveraged open area fish- ing mortality (=pF^yq) on the x axis of their per- recruit curves. For example, in the case analyzed in Myers et al (2000), where one of p areas would be fished each year, the fishing mortality F applied in the area open to fishing in a 9-vr rotation (i.e. 1/9 of the area would be fished each 52 Fishery Bulletin 101(1) 15 - /"^"^^'^'SS^ 3 g 10 - / 5 - / 6 year pulse rotation 6 year symmetric rotation — - - 10 year symmetric rotation 0.0 0.2 0.4 0.6 0.8 10 f.ufiCyo Figure 7 Comparison of yield-per-recruit for Georges Bank sea scallops with 6-year and 10- year symmetric rotations with that of a 6-yr pulse rotation or no rotation. year and that F = F^yc, ^9' would represent one ninth of the actual effort as the same F applied to the whole area under non-rotational fishing. Use of unaveraged fishing mortality has the effect of stretching the .v axis by a factor of p, thereby making their graphs appear flatter than they actually are. F^vg ^^ representative of not only the true time-averaged fishing mortality but also in many cases would be proportional to spatially averaged fishing effort (as measured by, e.g., hours fished). Myers et al. (2000) also suggested that rotational fishing would help lessen the impact of indirect (incidental) fish- ing mortality on yield-per-recruit. The analysis given in the present study indicates that incidental mortality low- ers yield-per-recruit at /^max about the same amount re- gardless of whether or not rotational fishing is employed. At levels of fishing mortality well above F^i^y^, rotational fishing does appear to modestly decrease the loss of yield- per-recruit due to incidental mortality. This decrease is due to the fact that incidental mortality, by somewhat lowering F^^^, exacerbates the effects of overfishing, whereas rotation alleviates the loss of yield-per-recruit due to overfishing. The effectiveness of sea scallop rotational fishing can be understood by examining fishing mortality at size for vari- ous rotational strategies. F'igure 7 shows fishing mortality as a function of shell height for no rotation and for 3-, 6-, and 9-yr rotations for F^^q = 0.2 (Fig. 8A), and /-^vp = 0.6 (Fig. 8B). Rotational fishing (especially for longer periods) tends to reduce the fishing mortality on small scallops and shift this effort onto larger individuals, thereby in- creasing yield-per-recruit, especially when overfishing is occurring. The periodic peaks in fishing mortality seen in the rotational strategies occur at the sizes where a new cohort begins to be fished (i.e. when the scallops are some integer number of years past their age at 40 mm). In practice, these peaks are likely to be much less pro- nounced because of variations in individual gi'owth rates and settlement times. However, the qualitative pattern of increasing selectivity with size should not be affected by such variations. Sea scallops are an ideal candidate for rotational man- agement, combining fast growth and low natural mortality with a sedentary adult lifestyle. In addition, sea scallops are recruited into the fishery at a size that is well below optimal from a yield-per-recruit perspective. The increase in size-selectivity induced by rotation described above should therefore induce an increase in yield-per-recruit. However, in those fisheries where the size-at-entry to the fishery is much larger, rotation would not be expected to induce gains in maximal yield-per-recruit (see Fig. 5A). On the other hand, it appears that rotation increases biomass-per-recruit regardless of the size-selectivity of the fishery (Fig. 5B). Botsford et al. ( 1993) found that rota- tion increased biomass- but not yield-per-recruit for red sea urchins. These results are consistent with the above discussion because the minimum legal size for landing the urchins was already near-optimal. Although the exact levels of yield- and biomass-per- recruit obtained with or without rotation are sensitive to such factors as natural mortality and growth rates, the relative gains of rotation over constant fishing are much less sensitive to these factors. Rotation will improve yield-per-recruit under a broad range of parameter choices provided that li the ratio of growth to mortality K/M is Hart: Yield- and biomass-per-recruit analysis of rotational fisheries 53 0,30 025 020 0 15 - 0,10 - 0-05 - 0.00 1.2 1.0 0.8 0.6 0.4 0.2 - 0.0 No rotation Ttiree year rotation Six year rotation Nine year rotation 80 No rotation 3 yr rotation 6 yr rotation 9 year rotation 100 120 140 80 100 120 Stiell tieight (mm) 140 Figure 8 Mean fishing mortality at length for Georges Bank sea scallops with no rotation and with 3-. 6-. and 9-yr pulse rotations for (A) F = 0.2, and (B) f = 0.6. sufficiently high (greater than about 0.5 with the other parameters in the model fixed as given in Table 1), and 2) size-selectivity is suboptimal. Rotation improves biomass- per-recruit under even a wider range of parameters. Allee effects may occur in broadcast spawners such as urchins and scallops. Areas that are closed for several years may allow these animals to form dense aggregations (that would likely be heavily fished if not closed), thereby improving fertilization success (Botsford et al., 1993). Such an effect would mean that rotation could produce greater benefits in fecundity than would be suggested by biomass- or eggs-per-recruit curves. Metapopulation structure might also be considered when designing a rotational strategy. If recruitment is limited by the supply of settling larva, an area that is a source of larva might be fished less than that required to maximize yield-per-recruit in order to increase larval sup- ply (Tuck and Possingham, 1994). The calculations that indicate long optimal rotational periods assume low constant natural mortality, indepen- dent of age or density, based on the study of Merrill and Posgay (1964). There is some evidence that the natural mortality rate of sea scallops may increase with age or size for shell heights greater than about 110 mm (Mac- 54 Fishery Bulletin 101(1) Donald and Thompson, 1986). If this is the case, optimal rotational periods would be shorter than calculated here, although the jdeld-per-recruit formalism would remain valid. More serious problems would be caused if there is density-dependent mortality of adults or if high adult den- sities inhibited recruitment because rotational closures can induce higher densities than would constant fishing. If either of these processes occur, shorter rotation periods would be advisable to minimize this problem. For sea scal- lops, however, observations of areas that have been closed to fishing for a number of years give no indication that such density-dependent processes are occurring (Fig. 2b in Hart 2001, and Table B5-8 in NEFSC^). An extreme case of rotational fishing is true pulse fishing, where all exploitable individuals are removed at periodic intervals (see e.g. Sluczanowski, 1984). Thus, true pulse fishing corresponds to pulse rotation (as defined in the present study) with an infinite fishing mortality. Such pulse fishing is not optimal for sea scallops, as can be seen by the slight decline of yield-per-recruit at high fishing mortalities in Figure 1 because at very high fish- ing mortalities, the partial selectivity of the gear loses its effectiveness and all individuals that are even slightly selected to the gear (i.e. that are even slightly greater than /in,,,,) will be removed. To put it another way, at high fishing mortality rates, the additional (i.e. marginal) catch obtained from a further increase in F will disproportion- ately consist of small animals, thereby reducing yield- per-recruit. Pulse fishing would be optimal if knife-edge selectivity is assumed. For this reason, the assumption of knife-edge selectivity would lead to unrealistic results for cases such as sea scallops, where gear selectivity increases more gradually with size. Proper application of rotational theory therefore requires a careful examination of fishing selectivity with size. Pulse fishing can be related to the classic Faustmann theory of forest rotation (see e.g. Reed, 1986; Clark, 1990). In this theory, if a stand of trees in an area that has last been harvested t years previously has value Vit), then the optimal time to harvest the trees satisfies V'(t) = 8(V(t)-c) + 5 V{t)- exp(&)- 1 (6) where 5 c the discount rate; and the cost of harvesting. In the case of a fishery, Vit) would represent the expected value of the exploitable stock (e.g. those of shell height greater than /!„„„) at time t. (Note that in this context, unlike the original forest application, it is not necessary to assume that all exploitable individuals arc the same age.) In the case of scallops, assuming all scallops command the same price per unit weight, c = 0, and 5 = 0.1, this formula would give an optimal rotation period of about 6.1 years (the optimal period would be moderately longer for realis- tic positive values of c). This value corresponds well to the rotation time of 6 years that optimizes discounted yield- per-recruit (see Fig. 2). However, the yield-per-recruit for 6-yr pulse fishing, V(6), is less than 80% of the maximal yield-per-recruit obtained by fishing uniformly. Again, the reduced yield-per-recruit is due to the fact that pulse fish- ing induces knife-edge selectivity at h^^^. rather than the usual gradual increase in vulnerability to the gear Symmetric rotational strategies appear to give less benefit than does pulse rotation. However, optimal pulse rotation would require high, and possibly impractical, lev- els of effort in an area when it is opened (e.g. F of about 1.7 for a 6-yr pulse rotation). In addition, such a strategy would require that areas be closed most of the time, pos- sibly inducing social-economic disruptions by closing traditional fishing grounds for long periods. Compared to pulse rotation, symmetric rotation requires less concen- trated effort, allows areas to be open half the time, and is less sensitive to the assumption of constant natural mortality. One possible compromise between pulse and symmetric rotation is to close an area for half the time and then gradually increase effort during the opening. For example, an area might be closed for three years and then fished for the next three years at Ffj^^, 2Fjj^^, and ■^^MAX- respectively. (Questions have been raised regarding the appropriate- ness of the use of whole-stock fishing mortalities as tar- gets or reference points for fisheries of sedentary stocks that include rotational or long-term closures (or both) (NEFSC-). The solid line in Figure 9 gives the whole stock (biomass-weighted) fishing mortality (assuming constant recruitment everywhere) for a pulse rotational system consisting of six areas, one of which is fished each year in turn. This whole-stock fishing mortality was obtained by simply dividing the yield-per-recruit for a 6-yr pulse rotation by the corresponding biomass-per-recruit. The x axis is F^vG' which should be proportional to true effort. As can be seen, whole-stock fishing mortality is proportional to effort for low fishing mortalities, but then flattens to a maximum of just under 0.4. A similar situation can happen even if an area is fished uniformly, except that a portion of the area is set aside as an indefinite closure. The dashed line in Figure 9 gives an example for the case when lO"^'; of the area is permanently closed and is allowed to equilibrate to the biomass-per-re- cruit corresponding to zero fishing mortality Whole-stock fishing mortality shows a relationship to the actual fishing effort (the fishing effort in the open area only) in the open areas that is similar to that of rotational fishing. In both cases, closed area biomass dominates the whole-stock fish- ing mortality calculation at high fishing effort. The yield at high fishing effort is essentially derived from incoming recruitment, which is not sensitive to fishing effort for very high effort levels. Therefore, the whole-stock fishing mortality becomes nearly constant when effort is high. ^ NEFSC (Northeast Fisheries Science Center). 1999. Report of the 29th northeast reffional stock assessment workshop (29th SAW). Stock Assessment Review Committee (SARC) consen- sus summary of assessments. NEFSC Ref Doc. 99-14, 347 p. [Available from NEFSC, 166 Water St., Woods Hole, MA 02543.1 Hart; Yield- and biomass-per-recruit analysis of rotational fisheries 55 Figure 9 Whole-stock fishing mortality as a function of efTort (F^vg' ''"' ^^^ scallops with a 6-yr pulse rotation (solid line), and constant fishing with 10% of the area permanently closed (dashed line). The dotted line is the line y = x. The current situation for sea scallops in Georges Bank gives an even more extreme example of this phenomenon. About 80% of the biomass lies in the groundfish areas that have been closed to scallop fishing for most of the time since December 1994. Because these areas will be closed to scalloping in 2002, the whole-stock fishing mortality in this year cannot exceed the F^y^ reference point of 0.24. Therefore, according to the current overfishing definition (the whole-stock F is below F-^j^x'- ^^^ stock cannot be overfished. Nonetheless, the fishing mortality in the open areas may exceed F^i^x- resulting in growth overfishing in these areas. Thus, the stock in the open areas could be overfished from a yield-per-recruit perspective even if the whole-stock F is below F^y^. The opposite situation could also occur If scallops in the groundfish closed areas on Georges Bank were fished more than slightly above the Ff^^j^ = 0.24 reference point, the whole-stock fishing mortality would also be above this ref- erence point and overfishing would be considered to be oc- curring. However, an area that has been closed for a number of years should be fished harder, compared to an area that has never been closed, once the area is reopened in order to maximize yield-per-recruit. Thus, a strategy that would maximize yield-per-recruit might require a whole-stock F that would in some years be higher, and in some years lower, than the conventional overfishing reference point. A whole-stock fishing mortality rate may therefore not be the most appropriate metric for overfishing definitions when some areas are temporarily or permanently closed to fishing. Its value may not be representative of the yield- per-recruit that could be obtained at that level of fishing mortality. Furthermore, when most of the biomass is in closed areas, estimated whole-stock fishing mortality may be more sensitive to variations in recruitment and mea- surement error than to actual changes in effort. As an alternative to a whole-stock fishing mortality metric, the following considerations are suggested for a fishing effort measure that is compatible with yield-per- recruit calculations. (1) Stock from areas that are not fished in a given time period should not be included in the fishing mortality calculation for that time period. In a relatively sedentary stock, the amount of biomass in the closed areas is irrelevant in determining the yield-per- recruit that will be obtained from the stock in the open areas. (2) Time-averaging of fishing mortality in the open areas is required to take into account the previous fish- ing history of the area. An area that has been closed for a number of years needs to be fished harder once opened than an area that has been continuously fished in order to maximize yield-per-recruit. Based on these considerations, the time-averaged fish- ing mortality computed from the open areas only, F.wg' is an appropriate measure of fishing mortality in fisher- ies managed by using rotational or indefinite closures. It is natural to take the averaging period equal to the rotational period p. With this metric. Fy^,^-^ is only slightly sensitive to the rotational period p and completely insensi- tive to the level of closed area biomass. Indeed, even if no closures existed, but fishing effort varied with time, it may still be advisable to employ a time-averaged fishing mor- tality because the previous history of fishing mortalities strongly affects the level of effort required to maximize future yield-per-recruit. If an area has been fished harder than Fm.yx for a number of years, so that the population 56 Fishery Bulletin 101(1) size-structure in this area is smaller than the equilibrium size-structure obtained by fishing at F^^^, then fishing the next year at a level somewhat below i^M^x will improve long term yield-per-recruit. Similarly, if an area has been fished below F^^j^, so that its size structure is larger than what would occur when fishing at a constant rate of Fj^j^^^, then it may be optimal to temporarily fish at a level higher than F,^^. In summary, rotational fishing can improve yield- and biomass-per-recruit for long-lived sedentary stocks such as sea scallops. Rotational management can be part of a precautionary strategy because it can help alleviate the effects of growth and recruitment overfishing. Rotational management will however require a rethinking of conven- tional yield-per-recruit reference points. Acknowledgments I would like to thank T. Kenchington for discussions regarding the equivalence of long-term survivorship under rotational and constant fishing. This paper also benefitted from discussions with and comments from P. Rago, L. Jacobson, S. Murawski, F. Serchuk, A. Applegate, and the reviewers. Literature cited Beverton, R. J. H.. and S. J. Holt. 1957. On the dynamics of exploited fish populations, 533 p. Chapman and Hall, London, United Kingdom. Botsford, L. W. 1981. Optimal fishery policy for size specific, density-depen- dent population models. J. Math. Biol. 12:265-293. Botsford, L. W., J. F. Quinn, S. R. Wing, and J. G. Brittnacher 1993. Rotating spatial harvest of a benthic invertebrate, the red sea urchin, Strongylocentrotus franciscanus. In Pro- ceedings of the international symposium on management strategies for exploited fish populations. Alaska Sea Grant Report AK-SG-93-02, p. 409-428. Alaska Sea Grant Pro- gram. Anchorage, AK. Caddy, J. F. 1973. Underwater observations on tracks of dredges and trawls and some effects of dredging on a scallop ground. J. Fish. Res. Board Can. 30:173-180. 1975. Spatial models for an exploited shellfish population, and its application to the Georges Bank scallop fishery. J. Fish. Res. Board Can. .32:1305-1328. 1993. Background concepts for a rotating harvesting strat- egy with particular reference to the Mediterranean red coral, Corallium rubrum. Mar Fish. Rev. 55:10-18. Caddy,J. F,andJ. C. Soijo. 1998. Application of a spatial model to explore rotating harvest strategies for sedentary species. Can. Spec. Publ. Fish. Aquat. Sci. 125:3,59-365. Campbell, A., R. M. Harbo. and C. M. Hand. 1998. Harvesting and distribution of Pacific geoduck clams, Panopea abrupta. in British Columbia. Can. .Spec. Publ. Fish. Aquat. Sci. 125:349-358. Clark, C.W. 1990. Mathematical bioeconomics. The optimal manage- ment of renewable resources, 2"'' ed., 386 p. Wiley, New York, NY. De Klerk, P, and M. Gatto. 1981. Some remarks on periodic harvesting of a fish popu- lation. Math. Biosci. 56:47-69. Garcia, S. 1984. Modelisation et exploitation rationnelle des stocks de corail precieux: une premiere approche. FAO Fish. Rep. 306:109-121. Gribble, N., and M. Dredge. 1994. Mixed-species yield-per-recruit simulations of the ef- fect of seasonal closure on a Central Queensland coast prawn trawling ground. Can. J. Fish. Aquat. Sci. 51:998- 1011. Hart, D. R. 2001. Individual-based yield-per-recruit analysis, with an application to the Atlantic sea scallop, Placopecten magel- lanicus. Can. J. Fish. Aquat. Sci. 58:2351-2358. Heizer, S. 1993. "Knob cod"-management of the commercial sea cucum- ber fishery in British Columbia. J. Shellfish Res. 12:144- 145. Lai, H., and A. Bradbury. 1998. A modified catch-at-size analysis model for a red sea urchin iStrongylocentrotus franciscanus) population. Can. Spec. Publ. Fish. Aquat. Sci. 125: 85-96. MacDonald, B. A., and R. J. Thompson. 1986. Production, dynamics and energy partitioning in two populations of the giant scallop Placopecten magellanicus (Gmelinl. J. Exp. Mar Biol. Ecol. 101:285-299. McCallum, H. I. 1988. Pulse fishing may be superior to selective fishing. Math. Biosci. 89:177-181. Merrill, A. S., and J. A. Posgay. 1964. Estimating the natural mortality rate of sea scallop. Res. Bull. Int. Comm. N.W. Atlantic Fish. 1:88-106. Myers, R. A., S. D. Fuller, and D. G. Kehler 2000. A fisheries management strategy robust to ignorance: rotational harvest in the presence of indirect fishing mortality Can. J. Fish. Aquat. Sci. 57:2357-2362. Quinn, T J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford U Press. New York, NY, and Oxford, United Kingdom. Reed, W. J. 1986. Optimal harvesting models in forest management — a survey. Natural Resource Modeling 1:55-79. Serchuk, F M., P W. Wood, J. A. Posgay, and B. E. Brown. 1979. Assessment and status of sea scallop ^Placopecten magellanicus) populations of the northeast coast of the United States. Proc. Natl. Shellfish. Assoc. 69:161-191. Sluczanowski, P. R. 1984. A management oriented model of an abalone fishery whose substocks are subject to pulse fishing. Can. J. Fish. Aquat. Sci. 41:1008-1014. Tuck, G. N., and H. P. Possingham. 1994. Optimal harvesting strategies for a metapopulation. Bull, Math. Biol. .56: 107-127. Hart: Yield and biomass-per-recruit analysis of rotational fisheries 57 Appendix Basic yield-per-recruit model This appendix describes the basic yield-per-recruit model used for a cohort. In this model, recruits start at a specified shell height (or length) /((,. The shell height is converted into a starting age Oq by using a von Bertalanffy growth equation. The shell height at time t is also obtained by using the von Bertalanffy growth cui-ve. The shell height is converted into a meat weight by using a shell-height/ meat weight formula: w = exp(o -(- h ln(/?)), (7) where w and h are in units of grams and millimeters, respectively. Natural mortality occurs at a rate M, assumed for these simulations to be constant for all ages (M=0. 1 ). The fishing mortality rate F(h} on a scallop of shell height /; is given by Fih) = FffJ(h), where Fq is the fully recruited fishing mortality rate and J(h> is the selectivity of the gear. J(h) was taken to be 0 if /; is less than a minimum shell height ^min' 1 'f ' '^ greater than a fully recruited threshold size hr^ii, and interpolated linearly as Jih)-. h-K H, full Aim (8) ^f'min < '' < /'full- Individuals that are caught by the gear but are smaller than a maximum cull size hj, are dis- carded and are subject to a discard mortality d. In these simulations, d is taken to be 0.2 (DuPaulM; however, the results are not very sensitive to the exact value of this parameter All individuals caught at a size greater than hj are assumed to be landed and are included in the total yield. F^(h) denotes the rate at which scallops of shell height h are caught and retained (i.e. not discarded). The possibility has been raised that some scallops may be killed but not captured by the gear (Caddy, 1973; My- ers et al. 2000). Caddy (1973) estimated that 15-20% of the scallops remaining on the bottom in the path of a scallop dredge are killed but not captured by the dredge. Murawski and Serchuk^ estimated that less than 59; of the scallops remaining in the path of the dredge suffered incidental (noncatch) mortality. In order to use the above studies to estimate the relationship between incidental fishing mortality F, and the fully recruited capture fish- ing mortality rate F^^, it is necessary to know the efficiency '* DuPaul, W. D. 2000. Personal commun. Virginia Institute of Marine Science, P.O. Box 1346, Gloucester Point, VA 23062- 1346. ■• Murawski, S. A., and F. M. Serchuk. 1989. Environmental effects of offshore dredge fisheries for bivalves. ICES CM. 1989/K:27. e of the dredge on a fully recruited individual. Denote by ; the fraction of scallops that suffer mortality among those that were in the path of the dredge but that were not caught, so that / is estimated at 0.15-0.2 by Caddy (1973), and less than 0.05 by Murawski and Serchuk.' The ratio R of fully recruited scallops in the path of the dredge that are caught to those killed but not caught is R = el\i(l-e)\. (9) If fully recruited scallops suffer capture fishing mortality at rate F^, then the rate of incidental fishing mortality will be F, = FJR F,J (l-e)/e. (10) If e is taken as 50% (estimated as the average scallop dredge efficiency on Georges Bank^), then F, would be in the range 0.15 F^ to 0.2 F,, according to Caddy (1973) and less than 0.05 Fg according to Murawski and Serchuk. -^ To ascer- tain the effects of incidental fishing mortality on the 5aeld- per-recruit calculations, model runs were performed with no incidental mortality, and also when F| = 0.15 F^; incidental fishing morality was applied to all size groups. Let Z(h) be the total mortality rate at shell height /; (i.e. the sum of natural mortality, and discard, indirect, and landed fishing mortality). Then the fraction of recruits re- maining t years after the beginning of the simulation is S(n = exp Z{T}dT (11) Total yield- and biomass-per-recruit are calculated by the formulas: Y= [s(t)F^{h{t))w(h(t}}dt B= [S(t)w{hit))dt, (12) (13) where a^ = the ending age of the simulation, taken to be 30 -^a^. For convenience in these simulations, a^ is taken to be 2 years; this age is assumed to correspond to a shell height of precisely 40 mm. In the rotational simulations reported in this study, the fully recruited landed fishing mortalities Fjh> (h>hf^i^) are assumed to vary periodically and are given in year k by F^^^, where > is the year that the cohort reaches the starting age Oq. Rago, P. J. 2001. Personal commun. Northeast Fisheries Science Center, 166 Water St., Woods Hole. MA 02543. 58 Abstract— Southern bluefin tuna( SBT) ^Tluinnus maccoyit) growth rates are estimated from tag-return data associ- ated with two time periods, the 1960s and 1980s. The traditional von Ber- talanffy growth model (VBG) and a two-phase VBG model were fitted to the data by maximum likelihood. The traditional VBG model did not provide an adequate representation of growth in SBT, and the two-phase VBG yielded a significantly better fit. The results indicated that significant change oc- curs in the pattern of growth in rela- tion to a VBG curve during the juvenile stages of the SBT life cycle, which may be related to the transition from a tightly schooling fish that spends sub- stantial time in near and surface shore waters to one that is found primarily in more offshore and deeper waters. The results suggest that more complex growth models should be considered for other tunas and for other species that show a marked change in habitat use with age. The likelihood surface for the two-phase VBG model was found to be bimodal and some implications of this are investigated. Significant and substantial differ- ences were found in the growth for fish spawned in the 1960s and in the 1980s, such that after age four there is a difference of about one year in the expected age of a fish of similar length which persists over the size range for which meaningful recapture data are available. This difference may be a density-dependent response as a con- sequence of the marked reduction in the SBT population. Given the key role that estimates of growth have in most stock assessments, the results indicate that there is a need both for the regu- lar monitoring of growth rates and for provisions for changes in gi'owth over time (possibly related to changes in abundance) in the stock assessment models used for SBT and other species. Estimating long-term growth-rate changes of southern bluefin tuna (Thunnus maccoyii) from two periods of tag-return data William S. Hearn CSIRO Marine Research Private Bag No. 5 Wembley, Western Australia 6020 Australia E-mail address, bill.hearnigimanne.csiro au Thomas Polacheck CSIRO Marine Research GPO Box 1538 Hobart, Tasmania 7001 Australia Manuscript accepted 28 May 2002. Fish. Bull 101:58-74(2003). Estimating growth rates has been a major focus of fisheries research throughout the twentieth century, and a large body of literature exists on the topic (e.g. Lee, 1912; Ford, 1933; Wal- ford, 1946; Manzer and Taylor, 1947; Allen, 1966; Yukinawa, 1970; Pitcher and MacDonald, 1973; Kimura, 1980; Fournier et al., 1990). This literature reflects, at least in part, the funda- mental importance of information on growth rates in stock assessments and the subsequent provision of man- agement advice for commercially har- vested fish populations. For example, growth information is required for yield-per-recruit analyses and for the estimation of spawning stock biomass in the estimation of stock-recruitment relationships. In addition, for a number of species, estimates of growth rates have been the primary or only source of information that can be used to esti- mate the age of individual fish and the age distribution of commercial catches (particularly for tropical species and for tunas and billfish). Such informa- tion on age is a critical component required in the analyses and models used to assess and manage these fish stocks (Bayliff 1991; Clay, 1991; Caton. 1991; Wild, 1994; Wild and Hampton, 1994; Polacheck etal.'). Almost all the work on modeling growth has centered on modeling growth rate as a continuous, smooth, monotonically decreasing function of age, and the von Bertalanffy (1938) growth (VBG) equation, and its ex- tensions, have been the most common approach used. In addition, the growth process has frequently been modeled as static. Temporal variations in aver- age growth for fish of the same size, or age, (due, for example, to changes in the physical environment or popu- lation density) are often ignored or considered to be relatively minor (with some notable exceptions — e.g. Le Cren, 1958; Southward, 1967; de Veen, 1976; Toresen, 1990; Ross and Nelson, 1992; Kaeriyama, 1996; Sinclair and Swain, 1996). For the large pelagic tunas and billfishes, the von Bertalanffy growth equation and extensions has been the standard used for modeling growth (Bayliff, 1980). For a variety of tuna species, numerous growth studies have been conducted, and generally a rea- sonable range of parameter values has been estimated (e.g. see the sets of pa- rameter values estimates for the eight scrombrid species in Bayliff, 1980). ' Polacheck, T, A. Preece, A. Bctlchem, and N. Klaer 1998. Treatment of data and model uncertainties in the assessment of southern bluefin tuna stocks. In Fish- ery stock assessment models (F Funk et al", eds.), p. 613-637. Alaska Sea Grant College Program Report AK-SG-98-01. Univ. Alaska, PO. Box 755040, Fairbanks, AK 99775-5040. Hearn and Polachek: Long-term growth rate changes in Thunnus maccoyii 59 Bayliff ( 1988) also investigated regional growth differenc- es in Pacific skipjack and yellowfin tunas. Interpretation as to whether any differences found are merely an artifact of the data collection or procedures used or whether they reflect real temporal or spatial difference has generally not been possible because the basic data (e.g. tagging, hard parts, length-frequency data), data collection procedures, analytic approaches, and the areas and time periods from which the data were collected have varied greatly among studies. For southern bluefin tuna (SBT) (Thunnus maccoyii), extensive juvenile tagging programs were conducted in the 1960s and 1980s, and a large number of returns with measured lengths were recovered. From both periods, some returns were received after times at liberty in excess of 10 years. These two sets of tagging experiments provide the basis for the direct comparison of growth over a time span of 30 years. Also, because of the large number of tags returned in these studies, a more detailed examination of the adequacy of the von Bertalanffy growth equation as a model of the growth process is possible than with many data sets. These tagging data (primarily those from the 1960s) have been used as a basis for a number of analyses of growth rates (Murphy, 1977; Kirkwood. 1983; Hearn, 1986; Hampton, 1991; Lucas-). In the present paper, we present results of the analyses of the growth increment data from these two sets of tagging experiments. We ex- amine these data both in terms of 1) whether SBT growth differed between the tagging periods and 2) whether there was a change in the growth process between adult and juvenile SBT (i.e. whether a more complex model than the simple von Bertalanffy equation is required to provide an adequate description of SBT growth). The results presented here incorporate and build upon the already cited published analyses of these tag-return data, unpublished reports, and discussions of SBT growth in scientific meetings on SBT (e.g. Hearn and Hampton'; Hearn and Polacheck''; Anonymous^). - Lucas, C. 1974. Working paper on southern biuefin tuna pop- ulation dynamics ICCAT ( Intenational Commission for the Con- servation of Atlantic Tunas), SCRS/74/4. Collective Volume of Scientific Papers, vol. HI, p. 110-124. [Available from CSIRO Marine Laboratories, GPO Box 1538, Hobart, Tasmania 7001, Australia.) 3 Hearn, W. S. and J. Hampton. 1990. SBT growth change. Ninth trilateral meeting on SBT, Hobart, Australia, September 1990, SBFWS/90/8. 19 p. (Available from CSIRO and the Com- mission for Conservation of Southern Bluefin Tuna. P.O. Box 37, Deakin West, ACT 2600, Australia.) ■> Hearn, W. S., and T Polacheck. 1993. Estmiating SBT age- at-length relations for the 1960s and 1980/90s. Twelfth trilat- eral meeting on SBT, Hobart, Australia, October 1993. SBFWS/ 93/4, 21 p. [Available from CSIRO and the Commission for Conservation of Southern Bluefin Tuna, P.O. Box 37, Deakin West, ACT 2600, Australia.) ■' Anonymous. 1994. Report of the southern bluefin tuna trilat- eral workshop; Hobart, Australia, January/February 1994, 161 p. [Available from CSIRO and the Commission for the Conser- vation of Southern Bluefin Tuna, P.O. Box 37, Deakin West, ACT 2600, Australia. [ Background: the SBT stock and fishery SBT is a highly-migratory species that begins to spawn at about 10-12 years of age in waters south of Java during the southern summer, mainly from September to April (Farley and Davis, 1998). During the first year of life they tend to be transported south by the tropical Leeuwin Current to inshore waters between Perth and Esperance, Western Australia. From ages 1 to 4 years, they appear to mainly inhabit, at least in the summer months, the waters off the Great Australian Bight, southern New South Wales (NSW) and eastern Tasmania. Many move to oceanic waters during the winter months and apparently progressively so as they age. By five years of age almost all have migrated to oceanic waters between 30° and 50°S at all longitudes, but mostly in the Eastern Hemisphere. Substantial surface fisheries operated off the south coast of Western Australia from 1969 to the mid 1980s, off the south coast of NSW from 1963 to the early 1980s, and off South Australia from 1964 to the present. Since 1959 a major Japanese longline fishery has operated in oceanic waters between 30° and 50°S, mainly from the mid-Atlan- tic and westwards to a few degrees west of New Zealand. Materials and methods Tagging programs Description Large numbers of tagged fish were released by CSIRO staff in the period from 1959 to 1968 and again in the period from 1980 to 1984. The releases from these two periods are used in our present study. Most of the tagged fish were initially caught with pole-and-Iine gear with barbless hooks, although a relatively small number were caught with troll lines. After a fish was hooked, it was hauled aboard the vessel and placed on a measuring board (in the 1960s) or a vinyl cradle (in the 1980s), and its nose to caudal fork length was measured. The fish was then tagged by an operator who inserted a 12-cm plastic spaghetti dart tag into the fish about 4 cm to the rear of the second dorsal fin on either side of the fish and rer- eleased it into the water within about 30 seconds. After 1963 almost all fish were double tagged. The tag numbers and length of each fish were recorded, together with loca- tion and date of release. This information was later trans- ferred to a computer database. Tagging operations in both the 1960s and 1980s were concentrated in the nearshore, surface-water fisheries bordering the central and western southern coast of Aus- tralia and the southern coast of NSW. In the 1980s no tags were released from the NSW coast area because this com- ponent of the fishery had collapsed, and surface schools of juvenile SBT could no longer be found (Caton, 1991). The South Australian tagging took place in the Great Australian Bight or in the adjacent shelf waters generally between longitudes 128° and 136°E. Releases in Western Australia occurred in the Albany (between longitudes 112° and 119°E) and Esperance (between longitudes 119° and 125°E) areas. There were 33,309 juvenile SBT tagged by 60 Fishery Bulletin 101(1) CSIRO personnel during the 1960s (1959 to 1968) and 10,743 during the early 1980s (1980 to 1984). Of these fish, 1972 and 4280, respectively, were later recaptured. On recapture, fishermen recovered the tags and re- corded the fish's length (if measured), location, and date. The tags with the recorded information were returned to the scientific staff at CSIRO, who then provided a reward. Most of the recapture lengths were measured by fisher- men or factory staff but about 31% were measured by scientists. Those measured by scientists cannot be con- sidered a representative sample. In particular, all of the measurements for longer-term recaptured fish come from fishermen aboard Japanese longline vessels. In addition to length, longliners often reported the dressed weight and sometimes the whole weight, or both, of recaptured fish. In the 1960s Australian fishermen seldom reported any weight measurements, but in the 1980s they commonly reported the whole weight of recaptured fish. Data selection The tagging experiments were conducted mainly within a narrow range of months at each site; therefore returns within a few months would be most strongly influenced by the seasonal differences found in SBT growth (Hearn, 1986; Burgess et al., 1991; Leigh and Hearn 2000). A nine-month period at liberty coincides with a low frequency in the times at liberty for the experi- ments; therefore we excluded data from analyses with less than 270 days at liberty. We also excluded data for which fish were tagged by fishermen, or when the recovery length, year, or month were reported by the tag finder to be unknown or uncertain. Previously reported weight-length relationships (Wara- shina and Hisada, 1970; Hampton, 1986; Robins^) were used to identify and screen out dubious recapture data. The details of the screening procedures are documented by Hearn" and Anonymous.-^ Longline recaptures were excluded if the expected weight of a recaptured fish for its reported length was either less than 2/3 of the reported weight or greater than 1.5 times the reported weight. Some of the major inconsistencies were thought to be due to measuring the length of a fish without its tail or with- out its head (Lucas^). For surface fish in the 1980s, a high proportion of the weight-length data for recaptured fish from four vessels was inconsistent with the weight-length relationships noted above. All tag-return data from these four vessels were excluded. Another 2.5% of the 1980 data were excluded because of highly unlikely values for the ra- tio of the reported weight to length of the recaptured fish. '" Robin.s, J. P. 1963. Synopsis of biolofjical data on .southern bluefin tuna, Thiinnus ihvnnus maccoyn (Ca.stlcnaui 1872. FAO Fisheries Report 6(2), p. 562-587. [Available from CSIRO Marine Laboratories, GPO Box 1538, Hobart, Tasmania 7(X)1, Australia.) " Hearn, W. S. 1982. Fish tagging: data processing, editing and storage. In CSIRO data base for .southern bluefin tuna {Than- nus maccoyii (Castlenau)) (J. Majkowski, ed.l. p. 8-9. CSIR(J Marine Laboratories, Rep. 142. lAvailablc from CSIRO Marine Laboratories, GPO Box 1538, Hobart, Tasmania 7001, Australia.] For the screening methods used, no assumption was made about the underlying growth curve, and these meth- ods were designed so that they would not induce a bias into the results. The selection process yielded data sets that were sufficiently large for valid analyses, being 730 and 1450 for the 1960s and 1980s data sets, respectively. Note that for other tuna species the selection process used in our study (particularly the deletion of recaptured fish with short times-at-liberty) may cause problems because of smaller data sets (e.g. skipjack and yellowfin tunas in Bayliff, 1988). Experimental assumptions The use of the tag-return in- crement data for estimating growth rates requires the fol- lowing assumptions about the tagging protocols and data collection procedures: 1 Tagging does not retard growth. 2 The tagged fish are uniquely and correctly recorded at release and recapture. 3 The lengths of fish are measured without bias at re- lease and recapture. 4 A wide range of fish sizes are represented, in recap- tures at least. 5 There are no significant size-selection processes for fish within similar age ranges. With respect to tagging effects, Hampton (1986) and Hearn (1986) have shown that there can be a significant weight loss of 7-12% for tagged fish in the first month after release. However, tagged fish recover this weight loss within a year at liberty, and there is no apparent difference between tagged and untagged fish after this time (Hearn, 1986). (There is little information available on weight loss of tagged fish at liberty between one month and one year) In terms of length, Hearn and Hampton' could not detect a reduction of growth from growth increment residuals in the tag-return data even within the first 30 days after release. Limited data from the effect of handling and tag- ging fish in commercial farm pens indicated no retarda- tion in growth in length after 150 days. These farm fish did show a loss in weight when first caged, but the weight was regained over a period of a few months (Anonymous''); therefore we do not think that tagging had any substan- tial effect on the growth rate of tagged fished in our study. With respect to the other assumptions, all fish were tagged with uniquely numbered tags. During tagging operations, tags were arranged in blocks of sequential numbers to avoid confusion and the misrecording of tag numbers. Return of the physical tag was required for fishermen to obtain the reward, and the double tagging of almost all fish since 1963 has allowed cross verification of tagging numbers, which allows little scope for error in the record- ing of tag numbers. Approximately 23*7^ of the length mea- surements for the selected recaptured fish were measured by scientists. Mainly due to the deletion of short-term recaptured fish (i.e. < 270 days), this is less than that for all data (31%). For the fishermen-measured lengths, there was no reason to suspect any consistent bias, and comparison of the residuals for fishermen- and scientist- Hearn and Polachek; Long-term growth rate changes in Thunnus maccoyii 61 measured returns in the fitted models below did not indi- cate any systematic pattern. The recaptured fish used in our study ranged in size from 60 to 175 cm, although the number of fish in the larger size ranges was relatively small — less than 2% were larger than 140 cm. (The con- sequences of the small number of fish in the large-size category are discussed below. ) Within both the surface and longline fishery, a range of sizes and age classes is har- vested within a single operation. No indication exists that within the size range encompassed by a cohort at a given age, there existed significant gear or fishery size selectiv- ity. Overall, the above basic assumptions seem reasonable in modeling growth from these SBT tagging data. (1991). In this model, fish grow according to one model (or parameter set) up to a certain length and according to another thereafter. In our analyses, we assumed that fish have VBG throughout their lives but grow according to one set of VBG parameters (L_j and k-^) up to length L* and according to a second set (L_.^ and k.-,) at larger sizes, the two-phase VBG model. Thus, the predicted length as a function of time for this model is L , l-e-*>"-'°' for tt* Analytical methods Models Two basic models were used to analyze growth information from the tag-return data. The first was the simple VBG model: where /, =L__(l-e-*"-'"'), (1) = the length that fish grow to asymptotically; = the length of a fish at age (or time) t; = the exponential rate at which the growth rate slows; and = the hypothetical age (or time) when a fish is of length zero. where t* = the predicted time for a fish to reach L*. Note that t* can be solved for in terms of four of the parameters of the model ltQ,k^, L^j, and L*): where k. (o = h + -r^°s Mogfl ^ ^~: ^«i (4) and 1,= the length of a fish at the time of tagging, t^ When applied to tag-return data, this equation can be used to predict the growth increment as a function of the length at release and the time at liberty: As with this simple VBG model. Equation 3 can be solved to predict the growth increment as a function of the release length and the time of liberty 5t = t.2-t{. SI = {L_ -l)(l-e -kSl) (2) where SI = the growth increment; St = the time at liberty; and / = the length of release. Note, in this study we simplified the growth model by not accounting for seasonal growth. However, data on recap- tured fish with short times at liberty were specifically deleted to ensure that our results were robust after this simplification. Preliminary analyses of the tag-return data suggested that a simple and time invariant von Bertalanffy growth model may not provide an adequate description of the growth rate for SBT. These preliminary analyses suggested S = (L^,i-/,,)(l-e"*''') i{t2(l-e"''^'*) if/,>^*. (5) It should be noted that in some of the analyses considered below, the estimate of L^j did not converge (i.e. the esti- mate for L. , was essentially infinite). In such cases, the estimated growth rate is linear, with growth rate /?,, and for the first phase we replaced the von Bertalanffy growth function with a simple linear one: « = /?,*, 1 Growth rates in the 1960s and the 1980s were not equal; 2 There were systematic deviations from a VBG curve, possibly corresponding to different growth processes or models for adults and juveniles. Consequently, in the present study, we considered a more complex model than the simple VBG and conducted sepa- rate, as well as combined, analyses of the tag data from the two periods. The more complex model selected was the two-phase growth model developed by Bayliff et al. and t* = t^-(L*-l,^)/Ry (6) Model-fitting procedure A large body of literature exists on statistical approaches for estimating growth from tag- return data (e.g. Fabens, 1965; Sainsbury, 1980; Kirkwood and Somers, 1984; Francis, 1988; James, 1991; Hampton, 1991; Wang et al., 1995). The most appropriate approach depends on the error structure assumed for the model. We followed the maximum-likelihood approach and general error structure described by Hampton (1991). The mea- sured growth increment offish "/" is 62 Fishery Bulletin 101(1) SI, = E\a] + e, + e„ (7) where f, is due to measurement error in the observed growth increment (i.e. the combined effect of any errors in measuring the lengths at the time of release and recap- ture) and e, is due to process or model error The latter may be a function of /j, St, SI, and the model parameters. For the measurement error component, we allowed for different variances, depending upon whether the recap- tured fish was measured by an independent and scientifi- cally trained individual or by a fisherman. Scientifically trained individuals (i.e. scientists) included fishery observ- ers, port samplers, and CSIRO staff We assumed that the measurement error was normally distributed, with mean zero and variance o^-, where .v is one of /"or s for recaptured fish measured by fishermen or scientists. The choice of the functional form for the process error in growth models is a complex issue. One approach has been to consider that process error stems from variability among individuals in the expected value of the growth parameters (e.g. Sainsbury, 1980; Hampton, 1991; Wang et al., 1995). This approach in the case of the two-phase VBG model would result in many potential structures for the process error component because there could be indi- vidual variability in the expected value of any single or possible combination of parameters (of which there are 25 combinations). There is little theoretical basis for deciding which of these 25 combinations to use. As an alternative, we selected a more empirical approach. A function that increases with longer times at liberty seemed appropriate, and was also consistent with preliminary analyses. We ex- plored both linear and quadratic functional relationships between the times at liberty and the process error com- ponent. The quadratic term was found to be insignificant, and therefore we chose to report only results for a simple linear functional relationship, namely 0;;^St. Hence the cor- responding variance of the expected gi'owth increment of fish ( is V( ^^ ) = a^- + a,~ St^. It should be noted that without independent data on measurement error any constant component in process error would be totally confounded with the measurement error term in the model. Therefore, a^ should be considered as a combined measurement and process error term. Both a^ and (T„, were estimated empiri- cally by maximum likelihood tag increment data. Assuming a Gaussian error distribution, the likelihood function is =ni--,.^.p(.it«!j (8) The estimates of the parameters are found by minimizing ■ln(L) = ^^ ,„,p.„.,„.it«a! (9) D.E. Shaw, CSIRO Div Maths, and Stats.), which uses the Nelder and Mead (1965) method. Model selection The estimation of the full two-phase VBG models across both tagging periods contains 16 parameters (five model parameters plus three variance parameters for each time period). We examined a variety of alternative hypotheses to test whether the number of parameters could be reduced by eliminating some or equating them. For the model parameters, we considered whether the L„ or k terms were equal either between time periods or between the first and second phases within a time period. We also considered the simple VBG model, for which L* doesn't exist. For the L '■ parameter, we considered whether the esti- mates were different between the two time periods. We also examined models in which the value of L* was de- termined by assuming that the expected growth rate for a fish of length L* was equal for both growth phases (i.e. by assuming that the changes in growth rates as a function of length is a continuous function). Under this assumption (10) The minimum value was obtained for all models by using the minimizing subroutine MINIMD (programmed by This model is referred to as the continuous rate two- phase model in the "Results" section. However, this model is not smooth because it has a discontinuity in the deriva- tive of the growth rate atL*. For the variance parameters, we considered whether any of them could be eliminated and also whether a^ = Of. We used the log-ratio test and AIC criterion (Akaike, 1974) to identify the most parsimo- nious model. Results Best fits to the 1960s and 1980s data Table 1 contains the maximum likelihood solutions for various assumptions when fitting growth models to either the 1960s or 1980s tag-return data separately. Using the AIC criteria, we found the best-fit model for the 1960s tag- return data was one with linear growth in the first phase and with the change between the two phases at approxi- mately 74 cm (row 1, Table lA). The fit to this model com- pared with all other parameter combinations yielded both the lowest AIC and negative log-likelihood values. The fit, however, was only marginally better then the fit (row 3, Table lA) to the two-phase VBG curve with common k parameters (e.g. where the difference in the negative log- likelihood values is 1.21). Except for the first phase growth parameters, the estimates for the other parameters are nearly identical between these two models. This similar- ity reflects the fact that growth is nearly linear over the initial part of a VBG curve. Thus, by having a relatively high L. I (271 cm), essentially similar gj-owth rates can be achieved up through the 74 cm size range when ^j = ^-2, as compared with linear growth in the first phase. It should Hearn and Polachek: Long-term growth rate changes in Thunnus maccoyii 63 Table 1 (A) Estimation of SBT growth parameters, and tests, from 1960s tag-return data with time at liberty of at least 270 days; (Bl Estimation of SBT growth parameters, and tests, from 1980s tag-return data with time at liberty of at least 270 days, "na" = not applicable because this is the normal von Bertalanffy curve, i.e. with only one phase. Common parameters Number of parameters ^.., *, L^2 k2 L* a^ ^/ cr„, -Log likelihood AIC A none 6 22.23' 0.0000 210.90 0.1063 74.24 0.000 3.122 2.992 2055.53 4123.07- ^«l=^»2 6 212.73 0.1451 212.73 0.1044 75.75 0.000 3.134 2.999 2057.39 4126.77 6 271.35 0.1060 211.35 0.1060 74.71 0.000 3,130 2.997 2056.74 4125.49 5 172.67 0.1723 172.67 0.1723 na 2.201 3.782 2.855 2099.04 4208.08 continuous rate 6 114.14 0.4289 205.45 0.1128 81.55 0.000 3.203 3.021 2065.89 4143.78 «fm=0 7 760.47 0.03425 191.33 0.1330 70.00 3.478 5.258 0.000 2088.19 4190.39 0=0f 6 22.20 0.0000 209.65 0.1085 74.04 2.301 2.301 3.180 2068.33 4148.66 a=af=0 6 454.10 0.05660 214.26 0.1033 74.70 0.000 0.000 3.752 2071.57 4155.15 B none 8 226.70 0.1649 182.52 0.1841 84.90 2.305 4.501 3.018 4509.99 9035.97 L^,-L^, 7 183.09 0.2276 183.09 0.1832 85.65 2.266 4.497 3.031 4510.44 9034.88 «1=«2 i.l=i.2 h -h 7 210.24 0.1841 182.61 0.1841 84.99 2.276 4.492 3.0.30 4510.02 9034.03* 5 156.45 0.2884 156.45 0.2884 na 1.626 4.209 3.405 4530.88 9071.76 continuous rate 7 141.07 0.3590 182.25 0.1842 97.70 2.148 4.473 3.085 4514.54 9043.09 f^„,=o 6 206.71 0.1883 180.82 0.1883 84.85 4.149 5.920 0.000 4526.06 9064.11 ci=af 6 210.74 0.1858 182.23 0.1858 85.36 3.948 3.948 3.183 4529.25 9070.49 a=a^O 5 209.46 0.1875 181.49 0.1875 85.30 0.000 0.000 4.737 4545.25 9100.49 ' Here A, is zero,i.e. the growth rate is constant in the first phase; therefore we give the estimate of the growth - The least AIC value for estimates from the 1960s data. Na= not applicable because this is the normal von phase. ' The least AIC value for estimates from the 1980s data. rate instead ofL.j. Bertalanffy curve, i.e. w th only one also be noted that the two-phase VBG model with common L^ (row 2, Table lA). was very similar to the common k parameterization, reflecting the high correlation between L„ and k in the VBG models. For the 1960s, the continu- ous two-phase VBG model was rejected, P< 0.005 (row 5, Table lA). For the 1980s data, the best-fit model based on the AIC values was for the two-phase VBG model with a common value for the k parameter in both phases (row .3, Table IB). The estimate of the size at which the change between the two phases occurs was 85 cm (compared to the estimate of 74 cm for the 1960s data). As with the results of the 1960s data, the common-/^ model, common- L model, and the full two-phase model yielded very similar values for both the likelihood and parameter estimates in the second phase, but not for those in the first phase. This similarity reflects the high correlation between k and L in the VBG model, so that over the limited size range below L* nearly identical growth rates can be achieved in the common-^ model by decreasing the value of L^j. For the 1980s (as with the 1960s), the continuous two-phase VBG model (7 parameters), was rejected, P< 0.005 (row 5, Table IB). For both the 1960s and 1980s data, the two-phase model provided a substantially and significantly better fit to the tag return data than a simple VBG model. This can be seen in Table 1 (A and B) by comparing the negative log-likelihood and AIC values for the simple VBG model (row 4) with any that include a two-phase component, particularly the continuous rate two-phase VBG model. We also fitted a smooth Richards' (1959) growth model (a generalization of the VBG model) to the data, which fitted better than the simple VBG model, but worse than the two-phase VBG models. Note, however, that the log-ratio test and AIC criterion may not be fully applicable for testing the differences between the simple and two-phase VBG models because the simple VBG model can arise in more than one way as a submodel of the two-phase model (e.g. with common L . and k parameters or from L* equaling zero or infinity) (Davies, 1977, 1987). Nevertheless, the large magnitude of the differences in the log-likelihood values indicates a significance difference. For the 1960s data, it should be noted that the scientist measurement error (crj was estimated to be essentially zero when it was included as an explicit term in several of the models. In these cases, we refitted the models exclud- ing this parameter. Common sense dictates that measure- ment errors would not be zero. The most informative data 64 Fishery Bulletin 101(1) Table 2 Comparison of SBT growth parameter estimates for the 1960s and 1980s, between absolute maximum likelihood and local maxima | likelihood. Common Number of -Log parameters parameters ^-1 ^ L„., k. L* -30 ( 1 t r 1 1 1 1 E D 2 4 6 8 10 12 14 c Time at liberty (yr) O) c ^ o 30 - 3 D "D U> Q <> ^ O OJ % o tr 20 10 ,\:->t^^^i.' .> o 0 o o o 00 0 0 ^wsp^j^^^yfj o o 0 o o -10 oo ^ o o o -20 o -30 1 1 1 1 1 1 I 60 80 100 120 140 160 180 Expected recapture lengtti (cm) Figure 1 (continued) Hearn and Polachek: Long-term growth rate changes in Thunnus maccoyii 67 2100 A 2095 - /' 2090 - / 2085 1 / 2080 \ / 2075 \ / 2070 \ 1 2065 \ /"\ _,.,--''' 2060 - ' 2055 - ■o o o J= OJ S" _l 1 2050 5 4550 1 1 1 1 1 1 1 1 1 0 60 70 B 80 90 100 110 120 130 140 4545 - 4540 - 4535 - 4530 - ~-N, 4525 - 4520 - / —■'' 4515 4510 ^ ^^s -•'''' 4505 - 4500 5 1 1 1 1 1 1 1 1 1 0 60 70 80 90 100 110 120 130 140 L- (cm) Figure 2 Negative log-likelihood value as a function ofL'; (Al for the 1960s; (Bl for the 1980s. 68 Fishery Bulletin 101(1) 350 300 250 200 150 100 50 rr 0 60 350 300 250 200 150 100 50 B 60 70 70 80 80 90 100 90 100 /.• (cm) 110 110 120 130 120 130 Figure 3 F"requoncy distribution for the estimates of L* from lOOO bootstrap simulation results: (A) for the 1960s; iBiforthe 1980s. Hearn and Polachek; Long-term growth rate changes in Thunnus maccoyii 69 Table 3 Southern bluefin tuna growth parameters and tests, from jointly analyzing the 1960s and 1980s data. Common Number of -Log parameters parameters ^-, ^ l^.. k.^ L* 0, ""t fJ,„ likeHhood AIC none (60) 13 22.23' 0.0000 210.90 0.1063 74.24 0.000 3.122 2.992 (80) 210.24 0.1841 182.61 0.1841 84.99 2.273 4.496 3.030 6565.55 13157.10 CT,„60=CT„,80 12 22.23 0.0000 210.98 0.1062 74.25 0.000 3.115 2.998 210.31 0.1840 182.67 0.1840 85.00 2.309 4.514 2.998 6565.56 13155.11 L60,,,,=Z.80.,, 12 213.29 0.1462 210.49 0.1069 75.03 0.000 3.133 3.001 213.29 0.1804 184.20 0.1804 85.00 2.323 4.518 3.001 6567.43 13158.85 L60„i=L60_j, =Z,80„, U 211.41 0.1480 211.41 0.1060 75.08 0.000 3.128 3.004 211.41 0.1827 183.24 0.1827 85.00 2.315 4.515 3.004 6567.44 131.56.87 /t60,=A'80| 12 185.24 0.1804 210.46 0.1069 75.11 0.000 3.119 3.008 213.26 0.1804 184.21 0.1804 85.01 2.325 4.510 3.008 6568.07 13160,15 L60.,,=L80„2 11 22.20 0.0000 195.25 0.1232 73.74 0.000 3.120 3.010 233.55 0.1584 195.25 0.1584 85.22 2.302 4.516 3.010 6568.89 13159.78 /f60,=*802 11 23.85 0.0000 186.76 0.1374 70.48 0.000 3.097 3.044 258.76 0.1374 207.04 0.1374 86.23 2.262 4.495 3.044 6575.52 13173.03 L*60=Z,*80 12 110.51 0.4707 196.71 0.1269 85.45 0.000 3.196 3.014 211.13 0.1827 182.97 0.1827 85.45 2.306 4.505 3.014 6574,09 13172.19 CT,60=a^80 12 22.22 0.0000 210.59 0.1072 74.10 1.602 3.023 3.068 209.66 0.1842 182.49 0.1842 84.94 1.602 4.472 3.068 6570.46 13164.92 0^60=0)80 11 22.28 0.0000 210.43 0.1064 74.23 0.000 4.174 2.975 210.40 0.1844 182.62 0.1844 85.02 2.394 4.174 2.975 6577.96 13177.91 ' Here k^ is zero. i.e. the growth rate is constant in the first phase; therefore we give the estimate of the growth rate insteac ofZ..,. SBT growth rates in the 1960s are estimated to be less than those in the 1980s up to 144 cm (Fig, 4), Comparison of the 1960s and 1980s expected growth curves over time for a 55-cm fish are presented in Figure 5. In making this comparison, we assumed that a 55-cm fish is approximately one year of age (Anonymous"') and that size at age one did not change between the 1960s and 1980s, as supported by length-frequency data from these two time periods (Leigh and Hearn, 2000; Anonymous''). Thus, Figure 5 can also be considered as an estimate of the expected length-at-age curve. Figure 5 indicates that the overall expected growth was significantly faster in the 1980s than in the 1960s, so that a fish of 55 cm or age 1 would take approximately four years in the 1960s to achieve the same length that would have been achieved in three years in the 1980s. A feature of the best-fitted estimated growth param- eters is that the expected growth curves intersect at -170 cm, so that after age 13 a fish from the 1960s is estimated to be larger than a fish from the 1980s. This crossover is driven primarily by the difference in the estimates of L .;. The standard log-likelihood test indicates a low probabil- ity, P=0.01, that L,^., for the 1960s and 1980s are the same. However, the analyses of the bootstrap estimates of L ,., indicate that the estimates are bimodal, reflecting the bi- modal distribution of L*. Random sampling from the boot- strap distributions for L.2 showed that in G.V/c of cases the 1960s L„,2 estimate was less than the 1980s estimate. For a two-sided test at the 5'7c significant level, at least 2.5'7f (and at most 97.5%) of the bootstrap samples would have been expected to have the 1960s L ., less than that of the 1980s to justify the hypothesis that the two L^., are equal. Thus, based on the bootstrap results, the hypothesis of equality cannot be rejected. Most of the 6.1% of cases are associated with the 1960s L,,., less than 180 cm, which are in turn associated with the upper mode of L* in Figure 3A, i,e, near L* = 91 cm. It is worth noting that only three recapture lengths were greater than 170 cm. There are, therefore, very minimal data for estimating growth rates beyond 170 cm and for precisely estimating L^.,. Discussion The results in this study indicate that the traditional VBG model does not provide an adequate representation of growth in SBT. There appears to be a significant change in the pattern of growth in relation to a VBG curve during the juvenile stages of the SBT life cycle. This, in turn, may be related to the transition from a tightly schooling fish that spends substantial time in near and surface shore waters 70 Fishery Bulletin 101(1) 40 r 35 - 30 -"■•-... IT 25 >s "'*.^ E o 0) 2 20 - 1 'irn'" 1 you b ^ 1980's 5 o 5 15 - ^^ """"■■■■-.....^ 10 ^^-^::.i,^^ 5 n 1 1 1 1 1 1 1 1 U 40 60 80 100 120 140 160 180 200 Length (cm) Figure 4 Comparison of the 1960s and 1980s best-fit estimates of southern bluefin tuna growth rates as a function of length. to one that is found primarily in more offshore and deeper waters. In this regard, recent information from archival tags indicates that SBT between 80 and 90 cm (about two to three years old) commonly migrate during winter months to offshore oceanic waters in the Indian Ocean and the Tasman Sea and begin to feed at substantial depth (Gunn and Block, 2001). In contrast, catches and samples off Albany, Western Australia, show that many SBT less than 70 cm stay in nearshore Australian waters during winter (Hynd, 1965; Murphy"^; and release data analyzed in this study). Thus, the growth changes estimated to be near L* = 80 cm may correspond to a marked change in the SBT behavior during these winter months. The von Bertalanffy growth equation and its modifica- tions have been the standard for modeling tuna growth. The life history dynamics for most tuna species (e.g. north Pacific bluefin, albacore, bigeve, and yellowfin tuna) have a bimodal component analogous to that of SBT. Thus, juveniles are frequently found in densely packed surface schools, whereas at larger sizes individuals are rarely found near the surface and appear not to occur in densely " Murphy, G.I. 1979. Southern bluefin tuna. Au.st. CSIRO Div. Fish. Oceanogr. Fishery Situation Report 1, 10 p. (Available from CSIRO Marine Research, CPO Hox 1538 Hobart 7000, Australia. I packed schools (although there is little direct information on schooling for these larger fish). Moreover, mature tuna expend considerable energy in the spawning process, and in some cases swim thousands of kilometers and incur considerable weight losses during spawning (Warashina and Hisada, 1970). Bayliff et al. (1991) also found that growth models with a rate discontinuity at a certain size provided a better fit to Pacific northern bluefin tuna tag- return data than a simple continuous growth model. The extent to which this may be a general phenomenon in tuna or other fish species with marked changes in habitat use with age is not clear However, the results from our study and those of Bayliff et al. ( 1991 ) suggest that a growth rate with a discontinuity at a certain size may be more common than existing modeling of growth may indicate. Complex growth models, which deviated from a simple continuous growth curve, have generally not been considered, and the available data, in many cases, may not have sufficient power to be able to statistically identify more complex growth processes if they exist. Although the complex two-stage growth model used in our study clearly provided a substantial and significant improvement in fit to the growth-increment data, the mod- el itself presents problems in terms of the biological inter- pretation of the parameter estimates for L*. The bimodal nature ol' the likelihood function means that the size and Hearn and Polachek: Long-term growth rate changes in Thunnus maccoyli 71 Figure 5 Comparison of the 1960s and 1980s best-fit estimates for the expected length of southern bluefin tuna as a function of age, assuming that the expected length of an age-1 fish is 55 cm. age where the change in growth occurs is not well defined. This, in turn, confounds the evaluation of the plausibility of different specific possible biological hypotheses underly- ing the change. Moreover, although the change in growth patterns may be quite rapid, a large discontinuity in the growth rates at a specific length seems unrealistic. The continuous two-stage VBG model did not fit the 1960s and 1980s data as well as the discontinuous two-stage VBG models. However, the two-phase VBG models fitted the data better than the simple von Bertalanffy growth curve (Table 1, A and B) and its generalization — the simple Richards' ( 1959) curve (senior author, unpubl. results). From both the statistical estimation and biological per- spective, we think there is scope for the development of more appropriate complex growth models. In this context, there is also need for the development of estimation proce- dures for these complex models that can take into account alternative error structures that allow for individual variability in the growth rate parameters (e.g. Sainsbury, 1980: James, 1991; Wang et al., 1995 ). In the joint analysis of the 1960s and 1980s data, (7„, was the only parameter found not to be significantly different between the two data sets. However, caution is warranted in any comparison and interpretation of growth curves determining parameter values because of the well-known negative correlation between k and L. of the VBG growth model and the bimodal nature of the likelihood surface, as already noted. In particular, the differences in the esti- mates of L^2 should not be taken as strong evidence that the asymptotic growth of SBT decreased or that there was a crossover in the growth rates. These complex growth changes are difficult to explain from a biological perspec- tive and, as noted above, the bootstrap results indicate that the hypothesis that the L^g parameters are equal cannot be rejected. Moreover, we would note that there is a paucity of tag return data for larger fish. A total of only seven tags were recovered from fish with lengths exceed- ing 165 cm and only three for fish with lengths in excess of 170 cm. Fitting VBG models does not provide reliable es- timates of growth when extrapolated beyond the range of the data because of the large negative correlation between k and L ^. We, therefore, do not think that the current data are sufficient to determine whether, in fact, L^^ differed between the 1960s and 1980s. One of the primary applications of the estimated SBT growth curves is to provide estimates of the age distribu- tion of commercial catch in stock assessments based on catch-at-age analyses (e.g. Anonymous'*). The predicted growth curves (assuming that an age-1 fish is 55 cm) indicate that the estimated ages of 165-cm fish have di- verged by about a year for the curve based on a common L ,,, compared with those for which L„2 is allowed to differ 72 Fishery Bulletin 101(1) between the 1960s and 1980s. For smaller sizes, the diver- gence is substantially less (e.g. for fish 140 cm or less the divergence is less than three months). In terms of using the growth rate data to estimate ages from lengths, these results indicate that for the older reproducing fish the re- sults will be highly sensitive to assumptions about L^2- The results from these tagging studies clearly show that growth rates for SBT hatched in the 1980s had increased in relation to those cohorts hatched in the 1960s. The in- crease in growth rates is substantial, so that a fish, on av- erage, would have been expected to take four years to grow from 55 cm to 111 cm in the 1960s, but only three years to do so in the 1980s. In other words, after age 4 there is a difference of about one year in the expected age of a fish of similar length, and this difference persisted over the size range for which meaningful recapture data were available. The change in growth and its magnitude are consistent with the analyses in Leigh and Hearn (2000) of the modes in length-frequency distributions of juvenile fish captured in the Australian surface fishery. The underlying causes of the change in SBT growth rates are unknown. They could be associated with changes in environmental conditions, population size, or a combination of the two. The change in SBT growth rates between the 1960s and 1980s is associ- ated with very substantial declines in both the adult and juvenile components of the SBT stock (Polacheck et al.'; Anonymous^). There is an increasing number of examples in which growth rates have been reported to be inversely corre- lated to fish population numbers because of intraspecific competition. For example, Le Cren (1958) documented an increase in the growth rate of perch after a planned reduc- tion of a lake population. In a converse case, Kaeriyama (1996) reported a decline in the growth rate of Japanese chum salmon following a many-fold increase in its popula- tion size because of a most successful hatchery enhance- ment scheme. Other accounts are published in Southward (1967), de Veen (1976), Toresen (1990), Ross and Nelson (1992), and Sinclair and Swain (1996). However, the re- ports are mainly on species for which direct aging data are reliable and regularly collected over a lengthy period, or the fish are hatchery reared. The hypothesis that the increase in SBT growth rates was the result of the marked reduction in SBT papulation size would seem plausible, given the similar associations that have been found in a number of fisheries phenom- ena. As discussed in Leigh and Hearn (2000), changes in juvenile SBT growth rates based on analyses of length- frequency data are also consistent with the change having a density-dependent component. In this regard, it is worth noting that preliminary analyses of tag return data from the 1990s indicate that growth rates in the 1990s were similar to those in the 1980s (Polacheck and Preece"^). Thus, these preliminary results are also consistent with the change in growth being a density-dependent response as both juvenile and adult SBT abundances remained at low levels during this period (e.g. Anonymous^; Polacheck and Preece'°). Large uncertainty exists about possible recovery of the SBT stock in the near future (e.g. Anony- mous^), but continued monitoring of SBT gi'owth may provide one indicator of stock recovery. To simplify our investigation we did not consider sea- sonal growth. We avoided possible bias, due to seasonal growth, by analyzing data only from fish with times at liberty more than or equal to 270 days. This restriction provided an efficient mechanism to focus on the long-term growth process and was effective because the resultant sets were large. Large numbers of recaptured fish with reliable information and times at liberty more than 9 months seem rare for other tunas, in which case the added complication of accounting for possible seasonable growth would be required to ensure the robustness of the results. The analyses in this paper represent the first docu- mented examples of substantial temporal changes in growth rates that persisted for an extended portion of the life span in a large pelagic tuna resource. For tuna stocks in general, estimates of growth rates play a major role in stock assessments and in the subsequent management advice derived from these assessments. Acknowledgments We thank the many crew and scientific staff who par- ticipated in the 1959-84 SBT tagging operations. We are especially grateful for Australian and Japanese fishermen who returned tags with information on recapture length. The 1983-84 tagging program was financially supported by an Australian Government grant. Literature cited Akaike, H. 1974. A new look at the statistical model identification. Institute of Electrical and Electronic Engineers Transac- tions on Automatic Control, AC-19, p. 716-723. IEEE Control Systems Society, New York, NY. Allen, K. R. 1966. A method of fitting growth curves of the von Berta- lanfl'y type to observed data. J. Fish. Res. Board Can. 23: 163-179. Bayliff.W. H 1980. Synopsis of biological data on eight species of scorn- brids. Inter-Am. Trop. Tuna Comm., Spec. Rep. 2 (W. H. Bayliff, ed. ), 530 p. lATTC, San Diego. CA. 1988. Growth of skipjack, Katsuwonus pelamis. and yellow- fin, Thiinniis alhaccires. tunas in the eastern Pacific Ocean, as estimated from tagging data. Inter-Am. Trop. Tuna Comm. Bull. 19(41:311-385. ■' Anonymous. 1998. Report of the 1998 Scientific Committee meeting 3-6 August 1998, Tokyo, Japan. lAvailable from the Commission for the ('onservalion of Southern Hluefin Tuna, PO Box 37, Deakin West, ACT 2600, Australia.] '" Polacheck, T., and A. Preece. 1998. Preliminary comparisons of the growth rates of southern bluefin tuna in the 1990s with tho.sc in the 1960s and 1980s. Tenth SBT recruitment moni- toring workshop, 14-17"' September 1998, Hobart. Australia. RMWS/9H/5, 11 p. lAvailable from CSIRO and the Commis- sion for the Conservation of Southern Bluefin Tuna, P.O. Box 37, Deakin West, ACT 2600, Australia.] Hearn and Polachek: Long-term growth rate changes In Thunnus maccoyii 73 1991. Status of northern bluefin tuna in the Pacific Ocean. Inter-Am. Trop. Tuna Comm., Spec. Rep. 7:29-88. Baylifi; W. H., I. Ishizuka, and R. B Deriso. 1991. Growth, movement, and attrition of northern bluefin tuna. Thunnus thynnus, in the Pacific Ocean, a.s determined by tagging. Inter-Am. Trop. Tuna Comm. Bull. 20(1): 1-94. Burgess, D., A. Caton. J. Gunn, W. Hearn, T Murray, and C. Proctor. 1991. Aging and growth of juveniles and adults. In Review of aspects of southern bluefin tuna: biology, population and fisheries (A. E. Caton, ed.), p. 210-224. Inter-Am. Trop. Tuna Comm., Spec. Rep. 7. Caton, A. E. 1991. Review of aspects of southern bluefin tuna: biology, population and fisheries. Inter-Am. Trop. Tuna Comm., Spec. Rep. 7:181-357. Clay D. 1991. Atlantic bluefin tuna (Thunnus thynnus thynnus (L.)l: a review. Inter-Am. Trop. Tuna Comm., Spec. Rep. 7:89-179. Davies, R. B. 1977. Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrics 64:247- 254. 1987. Hypothesis testing when a parameter is present only under the alternative. Biometrics 74: 33-43. de Veen, J. F. 1976. On changes in some biological parameters in the North Sea sole (Solea solea L.). J. Cons. Int. Explor. Mer 37:60-90. Fabcns, A. J. 1965. Properties and fitting of the von Bertalanffy growth curve. Growth 29:265-289. Farley, J. H., and T L. O. Davis. 1998. Reproductive dynamics of southern bluefin tuna, Thunnus maccoyii. Fish. Bull. 96: 223-236. Ford, E. 1933. An account of the herring investigations conducted at Plymouth during the years from 1924 to 1933. J. Mar Biol. Assoc. U.K. 19:305-384. Fournier, D. A., J. R. Sibert, J. Majkowski, and J. Hampton. 1990. MULTIFAN a likelihood method for estimating growth parameters and age composition from multiple length frequency data sets illustrated using data from southern bluefin tuna (Thunnus maccoyii). Can. J. Fish. Aquat.Sci. 47:301-317. Franci-s, R. I. C. C. 1988. Maximum likeHhood estimation of growth and growth variability from tagging data. NZ J. Mar Fresh- water Res. 22:42-51. Gunn.J. S. andB.A. Block. 2001. Advances in acoustic, archival and satellite tagging of tunas. In Tuna — physiological ecology and evolution (B. A. Block and E. D. Stevens, eds. ), p. 167-224. Academic Press, New York, NY. Hampton, J. 1986. Effect of tagging on the condition of southern bluefin tuna, Thunnus maccoyii, (Castlenaul. Aust. J. Mar Fresh- water Res. 37:699-705. 1991. Estimation of southern bluefin tuna Thunnus mac- coyii growth parameters from tagging data, using von Ber- talanffy models incorporating individual variation. Fish. Bull. 89:577-590. Hearn, W. S. 1986. Mathematical methods for evaluating marine fish- eries. Ph.D. diss., 195 p. Univ. New South Wales, Sydney, New South Wales. Hynd, J. S. 1965. Southern bluefun tuna populations in south-west Australia. Au.st. J. Mar Freshwater Res. 16: 25-32. James, I. R. 1991. Estimation of von Bertalanffy growth curve param- eters from recapture data. Biometrics 47: 1519-1530. Kaeriyama, M. 1996. Population dynamics and stock management of hatchery-reared salmons in Japan. Bull. Natl. Res. Inst. Aquacult., Suppl. 2:11-15. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Kirkwood, G. P 1983. Estimation of von Bertalanffy growth curve param- eters using both length increment and age-length data. Can. J. Fish. Aquat. Sci. 40:1405-1411. Kirkwood, G. P., and I. F. Somers. 1984. Growth of two species of tiger prawn, Penaeus esculentus and P. semisulcatus, in the western Gulf of Carpentaria. Aust. J. Mar Freshwater Res. 35:703-712. Le Cren, E. D. 1958. Observations on the growth of perch (perca fluviatilis L.) over twenty two years with special reference to the effects of temperature and changes in population density. J.Anim.Ecol. 27:287-334. Lee, R. M. 1912. An investigation into the methods of growth determi- nation in fishes. Publ. Circ. Cons. Explor. Mer 63:34. Leigh, G. M., and W. S. Hearn. 2000. Changes in growth of juvenile southern bluefin tuna (Thunnus maccoyii): an analysis of length-frequency data from the Australian fishery. Mar Freshwater Res. 51:143- 154. Manzer, J. I., and F H. C. Taylor 1947. The rate of growth of lemon sole in the Strait of Georgia. Fish. Res. Board Can. Prog. Rep. Pac. Coast Stns. 7224-27. Murphy, G. I. 1977. New understanding of southern bluefin tuna. Aust. Fish. 36(11:2-6. Nelder, J. A., and R. Mead. 1965. A simplex method for functional minimization. Com- put. J. 7:308-313. Pitcher, T J., and P D. M. MacDonald. 1973. Two models for seasonal growth in fishes. J. Appl. Ecol. 10:559-606. Richards. F J. 1959. A flexible growth function for empirical use. J. Exp. Bot. 10:290-300. Ross, M. R., and G. A. Nelson. 1992. Influences of stock abundance and bottom-water temperature on growth dynamics of haddock and yellow- tail flounder on Georges Bank. Trans. Am. Fish. Soc. 121: 578-587. Sainsbury, K. J. 1980. Effect of individual variability on the von Bertalanffy growth equation. Can. J. Fish. Aquat. Sci. 37:241-247. Sinclair, A. F, and D. P. Swain. 1996. Comment: Spatial implications of a temperature-based growth model for Atlantic cod Gadus morhua off the eastern coast of Canada. Can, J. Fish. Aquat. Sci. 53:2909-2911. Southward, G. M. 1967. Growth of Pacific halibut. Rep. Int. Pac. Halibut Comm. 43, 40 p. 74 Fishery Bulletin 101(1) Toresen, R. 1990. Long-term changes in growth of Norwegian spring- spawning herring. J. Cons. Int. Explor. Mer 47:48-56. von Bertalanfly, L. 1938. A quantitative theory of organic growth. (Inquiries on growth laws. II). Hum. Biol., 10:181-213. Walford. L.A. 1946. A new graphic method of describing the growth of animals. Biol. Bull. (Woods Hole, Mass.) 90:141-147. Wang. Y. G., M. R. Thomas, and I. F. Somers. 1995. A maximum likelihood approach for estimating growth from tag-recapture data. Can. J. Fish. Aquat. Sci. 52:252-259. Warashina, I., and K. Hisada. 1970. Spawning activity and discoloration of meat and loss of weight in the southern bluefin tuna. Bull. Far Seas Fish. Res. Lab. (Shimizu) 3:147-166. |In Japanese with English abstract.) Wild, A. 1994. A review of the biology and fisheries for yellowfin tuna, Thunnus albacares, in the eastern Pacific Ocean. FAO Fish. Tech. Pap. 336/2:52-107. Wild, A., and J. Hampton. 1994. A review of the biology and fisheries for skipjack tuna, Katsuwonus pelamis, in the Pacific Ocean. FAO Fish. Tech. Pap. 336/2:1-51. Yukinawa, M. 1970. Age and growth of southern bluefin tuna, Thunnus maccoyii (Castlenau) by use of scale. Bull. Far Seas Fish. Res. Lab. (Shimizu) 3:229-257. 75 Abstract— The life history of the At- lantic sharpnose shark {Rhizoprion- odon terraenouae) was described from 1093 specimens collected from Virginia to northern Florida between April 1997 and March 1999. Longitudinally sectioned vertebral centra were used to age each specimen, and the period- icity of circuli deposition was verified through marginal increment analysis and focus-to-increment frequency dis- tributions. Rhizoprionodon terraenovae reached a maximum size of 828 mm precaudal length (PCD and a maxi- mum age of 11-1- years. Mean back-cal- culated lengths-at-age ranged from 445 mm PCL at age one to 785 mm PCL at age ten for females, and 448 mm PCL at age one to 747 mm PCL at age nine for males. Observed length- at-age data (estimated to 0.1 year) yielded the following von Bertalanffy parameters estimates: L„= 749 mm PCL (SE=4.60), K = 0.49 (SE=0.020), and tg=-0.94 (SE=0.046) for females; and Z,„ = 745 mm PCL (SE = 5.93), ft:=0.50 (SE=0.024l, and ?o = ^0.91 (SE = 0.052) for males. Sexual maturity was reached at age three and 611 mm PCL for females, and age three and 615 mm PCL for males. Rhizoprionodon terrae- novae reproduced annually and had a gestation period of approximately 11 months. Litter size ranged from one to eight (mean=3.85l embyros, and in- creased with female PCL. Life history of the Atlantic sharpnose shark {Rhizoprionodon terraenovae) (Richardson, 1836) off the southeastern United States Joshua K. Loefer South Carolina Department of Natural Resources Manne Resources Research Institute 217 Fort Johnson Road P,0, Box 12559 Charleston, South Carolina 29412-2559 E mail address loefer|(S'mrddnr state scus George R. Sedberry South Carolina Department of Natural Resources Marine Resources Research Institute 217 Fort Johnson Road PO Box 12559 Charleston, South Carolina 29412-2559 Manuscript accepted 22 July 2002. Fish. Bull. 101:75-88 (2003). The Atlantic sharpnose shark (Rhizo- prionodon terraenovae) is a small carcharhinid that inhabits the coastal waters of the western North Atlantic from the Bay of Fundy to the Yucatan (Castro, 1983). It is the most common small coastal species off the southeast- ern U.S. coast and the Gulf of Mexico ( Branstetter, 1990). This species is frequently encountered by a variety of commercial fishing gear, including bottom longline, gill net, bandit reel (used by the snapper-grouper fishery), and shrimp trawl. Rhizoprionodon ter- raenovae is also a common catch in the recreational hook-and-line fishery. The age and growth of this spe- cies has been described in the Gulf of Mexico by Parsons (1981, 1983a, 1985) and Branstetter (1981, 1986, 1987a). Although those studies provided sig- nificant information on the age and growth of R. terraenovae, data were collected from 1979 to 1984, a time in which fishing pressure on the R. terrae- novae population was probably not as high as at present (Cortes, 1995). The previous studies dealt with fishes only from the northern Gulf of Mexico, and therefore may not represent the entire stock, although the stock structure for R. terraenovae in the northwestern Atlantic remains unclear (Heist et al., 1996). No published age and growth studies exist for specimens collected from the southeastern U.S. Atlantic coast. The reproductive biology of this species has been studied in both the Gulf of Mexico and off the southeast- ern U.S. coast (Parsons, 1983b; Cas- tro, 1988, 1993; Castro and Wourms, 1993), but the lack of concurrent age and growth data off the southeastern United States limits the utility of these data for fishery management. Considering the importance of accu- rate and timely age, growth, and repro- ductive information to fishery manage- ment, this study had two objectives: to describe age, growth, and reproduction in the southeastern U.S. population of R. terraenovae; and to compare these data to those of previous studies on the same species in the Gulf of Mexico. Materials and methods Rhizoprionodon terraenovae (/!=1093) were collected throughout the year in coastal waters from April 1997 through March 1999. Collection sites ranged from Chesapeake Bay, Virginia, to Port Canaveral, Florida (Fig. 1). The majority of specimens were collected off the coast of South Carolina. A vari- ety of sampling gears were employed for sample collection: bottom longline (47'7f of specimens), otter trawl (22%), port-sampling of commercial fishing 76 Fishery Bulletin 101(1) -38° •34" 26' 84° 80° 100 76° 200 300 400 km Figure 1 Sample collection sites and distribution by area (roughly equivalent to state borders) for R. terraenovae collected during this study, 1997-99. (D) represents locations where one or more R. terraenovae were captured. (A) = 13 males (694-793 mm PCL); (B) = 52 females (215-786 mm PCD, 51 males (200-765mm PCL); (C) = 497 females (197-813 mm PCL), 441 males (225-828 mm PCL); (D) = 8 females (302-763 mm PCL), 7 males (320-658 mm PCL I; (E) = 6 females (335-738 mm PCL), 16 males (271-720 mm PCL). vessels (16%), rod and reel (12'7( ), gill net (39^ ), and other miscellaneous gear types (2%). Following capture, the sex of each specimen was deter- mined and the specimen was weighed (to the nearest 0.1 kg), evaluated for sexual maturity, and its body length was measured. Four body length measurements (to the nearest mm) were taken from each individual: precaudal length (PCL, measured from the tip of the snout to the anterior termination of the precaudal pit), fork length (FL), natu- ral total length (NTL, measured with tail in a "natural" swimming position (Parsons, 1985]), and total length (TL, measured with dorsal portion of tail bent parallel to the body axis). Unless otherwise noted, precaudal lengths are used throughout this study. Regression relationshiijs of TL, NTL, and FL on PCL were derived to facilitate com- parison with other studies. The claspers of males were measured from the clasper tip to the anterior termination of the vent. The siphon sac was measured from the base of the clasper fin (where the sac originates) to the anterior termination of the sac. The condition of the seminal vesicles was also recorded. Male maturity was indicated by calcification of the claspers and the presence of a fully formed siphon sac (Clark and von Schmidt, 1965; Parsons, 1983b). Gonadosomatic indices (GSIs, Parsons, 1983b) were calculated for male sharks with the formula GSl = gonad weight (g)/hody tveight (g) x 100. Loefer and Sedberry: Life history of Rhizophonodon terraenovae off the southeastern United States 77 The ovaries and uteri of females were examined macro- scopically for indicators of maturity, such as yolking eggs, embryos, or placental scars. Vitellogenic oocytes were eas- ily identified by their bright yellow coloration in contrast to the pale white coloration of nonvitellogenic oocytes. If vitellogenic oocytes were present, the diameter of all vitel- logenic oocytes in the ovary was measured (to the nearest 0.1 mm) with dial calipers. If maturing oocytes were not present, the most differentiated nonvitellogenic oocytes (which were noticeably larger that the rest of the oocytes in the ovary) were measured. Any embryos were removed from the uteri, counted, their sex determined, and mea- sured (TL). Female maturity was determined by the presence of embryos, umbilical scars in the uterus from previous pregnancy, or the presence of large vitellogenic oocytes (greater than 15 mm diameter) nearing ovulation (Parsons, 1983b). A segment of the vertebral column extending from the cervical region (dorsal to the branchial chamber) to the origin of the first dorsal fin was removed from each specimen and frozen. Vertebrae from the cervical portion of the spinal column were used for aging because of the shallow concavity of the intermedalia and the size simi- larity between adjacent centra in this region. The shal- low concavity of the vertebrae facilitated processing and measurement during aging (Branstetter and McEachran, 1986). Age determination was attempted on 890 of the 1093 specimens collected during the study. Vertebrae selected for aging were separated from the frozen seg- ment, defrosted, and soaked in 5% sodium hypochlorite for 5-30 min (depending on size) and were removed from the solution as soon as all excess connective tissue had been dissolved. A longitudinal section approximately 500 pm thick was cut from the center of each vertebrae with a Mark- V watering saw and allowed to air-dry for at least 24 h. Dried sections were then attached to glass slides with Accu-mount 60 mounting medium and hand polished with wet 600-grit sandpaper to a thickness of approximately 350 pm. Several staining or ring elucidation techniques (e.g. Parsons, 1983a; Branstetter, 1986; Brown and Gru- ber, 1988; Hoenig and Brown, 1988) failed to significantly increase increment visibility; therefore all aging was per- formed with unstained vertebral sections. Vertebral sections were read on a dissecting microscope with transmitted light and a polarizing filter at 20x mag- nification. Increment radii and marginal increments were measured through the center of the corpus calcareum (Fig. 2) with OPTIMAS image analysis software (Media Cyber- netics, 1999). Precaudal length was regressed on centrum radius (CR) for males and females to test for an isometric relationship. The increments observed in vertebral sections were narrow circuli similar to those described by Simpfen- dorfer (1993), as opposed to the gi-owth bands described by Branstetter (1987a). All increment counts were made without knowledge of the size, sex, or collection date of the specimen. The primary reader (senior author) counted increments on all samples twice; each reading was sepa- rated by at least two months. Increment counts that were not in agreement were counted a third time. If the third Figure 2 Diagrammatic representation of a vertebral sec- tion; bm = birth mark, c = circuli, cc = corpus calcareum, cr = line of centrum radii and annuli measurements, f = focus, i = intermedalia. count did not agree with one of the first two counts, the specimen was excluded from the analysis. The secondary reader (coauthor) counted increments from all specimens not eliminated by the primary reader's analysis. Between- reader disagi'eements were re-examined by both observ- ers simultaneously. All specimens for which a consensus could not be reached were discarded. The index of average percentage error (lAPE; Beamish and Fournier, 1981) was used to estimate precision between the final readings of the primary reader and the initial readings of the second- ary reader The annual periodicity of increment formation was verified through marginal increment analysis and focus- to-increment frequency distributions. Absolute marginal increment distances were converted to "relative" marginal increments by dividing the distance between the last in- crement and the edge of the centrum by the width of the last fully formed growth band (Skomal, 1990; Natanson, et al., 1995). This conversion compensated for differences in growth rates between age classes. Back-calculated lengths at previous ages were esti- mated from vertebral measurements by using a modified Fraser-Lee equation proposed by Campana ( 1990): L„ = L,, + |(C„ - C,.) (L,. - L„)/{C,, - Cf,)], where L^ = length at age; L_ = length at capture; C^ = centrum radius from focus to increment a; and C = centrum radius at capture. 78 Fishery Bulletin 101(1) Lq and Cq are biologically derived intercepts that repre- sent the fish length and centrum radius, respectively, at which the proportionality between fish length and centrum growth are initiated. For the purposes of this study, mean body length and centrum radius at birth were used as the biologically derived constants (Sminkey and Musick, 1995). The observed age-class data were used to estimate "actual ages" to 0.1 year These were calculated by the number of circuli present plus growth since the formation of the last circulus. All specimens were given a 1 June birth date, which approximates the middle of the pupping season. This process corrected for growth since the last in- crement, preventing the potential overestimation of size- at-age that might result from analyzing the data by year class alone. All three types of length-at-age data (observed age class, observed actual age, and back-calculated age) were fitted to the von Bertalanffy growth equation (VBGE; von Bertalanffy, 1938): L, = LJ1 -Kit- 'o'), where L, = length at age t; L^ = asymptotic length; K = growth coefficient; and Tq = theoretical age at zero length. Each of the three types were analyzed for sexes combined, as well as for each sex separately. The parameters for the VBGE were estimated through a stepwise Gauss-Newton iterative fitting process computed by JMP statistical analysis software (Anonymous, 1998). Results The sharpnose shark was abundant throughout the year in coastal waters within the sampling area. The ratio of males to females in the overall sample was not signifi- cantly different from a 1:1 ratio (chi-square test, « = 1091. a=0.65, v=l,;^2=1.39,P=0.24). Linear regression of TL, NTL, and FL on PCL resulted in the following equations: TL = 29.804 + 1.279PCL NTL = 31.678 + 1.254PCL FL= 11.249 -H 1.075PCL (;i = 1009, ;-=0.99,P<0.0001); (n=493,r2=0.99,P<0.0001); (n = 1083, r2=0.99, P<0.0001). Reproduction and maturity Size-at-maturity estimates were based on observations of 526 males and 564 females. The smallest fully mature male was 600 mm PCL, and the largest immature male was 615 mm PCL. All males greater than 615 mm PCL and 36'" '" -ales from 600 to 615 mm PCL were fully mature. The onset and completion of maturity in male R. tcrraenorae were demonstrated by the onset of develop- ment in the claspers and siphon sac (Fig. 3). Males began to mature at 500 mm PCL. The maturation of claspers and siphon sac reached completion approximately one year later, at 600 to 615 mm PCL. The smallest maturing female was 509 mm PCL and con- tained one maturing oocyte five mm in diameter The second smallest maturing female was 529 mm PCL. The smallest gravid female was 591 mm PCL. The largest immature fe- male, based on lack of embryos or uterine scarring, was 611 mm PCL. Females from 591 to 611 mm PCL were either gravid (63%) or contained large (>10 mm diameter) matur- ing oocytes and were close to their first ovulation (377^ ). All females greater than 611 mm PCL were mature. Mean GSI and mean ovarian egg diameter (MOD) both demonstrated prominent peaks during the calendar year Male GSI values were highest in April and high values were also present in March and May (Fig. 4). However, the seminal vesicles remained turgid and full of semen for some time following the seasonal testicular degeneration which began in May. Female MOD values were highest in May and June. An increase in standard error along with a drop in mean value for the month of June (Fig. 4) demon- strated that ovulation began at that time. The extremely low MOD in July indicated the completion of ovulation. Litter sizes ranged from one to eight, and generally in- creased with female PCL (Fig. 5). Mean litter size was 3.85 embryos, and significantly more embryos were found in the left uterus (mean=2.19) than in the right (mean=1.65; chi-square test, «=558, a=0.05, v=4. ^2=62.62, P<0.0001). Nonlinear regression of litter size on female PCL resulted in the following equation («=278, /■'-=0.51, P<0.0001): Litter size = -11.07 -i- 0.021 PCL + 1.37 X lO-^lPCL- 710.9)2 Rhizoprionodon terraenovae were born at approximate- ly 212 mm PCL. The smallest free-swimming neonate was 190 mm PCL, and the largest full-term embryo was 242 mm PCL. Most pupping occurred from mid-May to early June. However, a small number of neonates appeared as early as mid-April. Consequently, mean embryo total length was at a minimum in July and at a maximum in June (Fig. 6). The sexes of uterine embryos were not sig- nificantly different from the expected 1:1 ratio (chi-square test, ?!=844, a=0.05, v=l, ;f-=0.076, P=0.78). Age and growth Separate linear regressions of PCL on centrum radius (CR) for males and females were not significantly different (ANCOVA, P=0.065l and were therefore combined (Fig. 7) to yield the following formula: PCL = 61.80 -I- 124.48Ci? (r2=0.963, n=812, P<0.0001). The regression line slightly overestimated centrum radius for the largest individuals (>700 mm PCL) of both sexes. Data transformation, as well as nonlinear regression, failed to increase the r^ value, and only the largest speci- mens were affected. Nonlinear regression of total body weight on length was significantly different between males and females (AN- COVA after log-transformation, P<0.001), and resulted in the following equations: Loefer and Sedberry: Life history of Rhtzopnonodon terraenovae off tfie southieastern United States 79 O 300 250 - 200 - 150 ■ 100 - 50 - itu - 120 ■ • ••• • °«9 100 80 ■ 60 ■ 40 ■ 20 • n ■ n = 441 2«^°^ n = 290 200 300 400 500 600 700 800 900 Precaudal lengtti (mm) Figure 3 Relation of clasper and siphon sac length to precaudal length for male R. terraenovae; (•) represents individuals with uncalcified claspers; (O) represents individuals with fully calcified claspers. Females: W, = e(-is.62)pcL(3M) (/•^'=0.99, P<0.0001, /i=458); Males; w = e(-i8.i8.pcL'2 96i (?-^'=0.99, P<0.0001. n=454). where W, = total body weight. Aging was attempted on 890 specimens, 812 of which were aged without elimination. Agreement between the first and second counts conducted by the primary reader was 66%, with 91% within one increment, and 99% within two. Those sections that showed disagreement between the first and second reading (n=303) were counted a third time, and 96% agreed with one of the first two readings. The remaining 4% (12 specimens) were excluded from the analysis. Agreement between readers was 72%, with 95% within one increment and 99%. within two. Vertebrae for which counts did not agree between readers (246 out of 878) were re-examined by both readers simultaneously. A concurrent age could not be reached on 66 vertebrae, which were eliminated from the study. The lAPE between the final readings of the primary reader and the initial readings of the secondary reader was 7.4%. Size-frequency distributions of the discarded individuals (data not shown) closely matched those of the raw data set and did not indi- cate the elimination of a large number of individuals from any age class during the aging process. Mean relative marginal increments for age classes 1+ through 7+ combined demonstrated a minimum in July (Fig. 8). The O-i- age class was excluded from this analysis to ensure that growth from the birth mark did not affect the results. Frequency distributions of focus-to-increment measurements for ages 0+ through 7+ demonstrated single modes for all annuli in each age class for both males and females (Fig. 9). Most R. terraenovae were found to have an increment in the intermedalia and an associated change in the angle of the corpus calcareum. which is similar to the birth mark 80 Fishery Bulletin 101(1) described by other authors (e.g. Casey et al., 1985; Branstetter, 1987b; Simpfendorfer, 1993). There were 239 young of the year R. terraenovae collected during this study, 88 of which contained no discernible birth mark. All young of the year lacking a birth mark were captured in June and July (Fig. 10), whereas all young of the year captured from August through April had a birth mark. Both marked and unmarked centra were noted in July and showed a readily apparent trend; individuals with a birth mark were sig- nificantly larger than those without a birth- mark (/-test, df=96, ^=-7.138, P<0.0001). Back-calculated lengths-at-age were sim- ilar to observed lengths-at-age in all cases, although observed values were slightly higher for all age classes (Table 1). There was no evidence of Lee s phenomenon in the older age classes. Back-calculated size at the birth mark overestimated size at birth as determined by observations of neonates and full-term embryos. The VBGE estimates calculated by age class, actual age, and back-calculated age demonstrated little variation either within or among data types (Table 2). The VBGE parameters from all data types corresponded well with known life his- tory parameters for size at birth and maximum size. Unless otherwise noted, all comparisons throughout the remainder of this study were based on VBGE es- timates derived from the "actual age" data type. Discussion Reproduction Length-at-maturity estimates for male R. terraeno- vae were similar among the three published studies. Parsons ( 1983b) estimated male maturity at -610 to 653 mm PCL (lengths from other studies were con- verted to PCL by using the formulae derived from the current study) and Branstetter (1987a) estimated the same at 600 mm PCL. We determined that males reach full maturity at -600 to 615 mm PCL. The three studies failed to agree on length at maturity for female R. terraenovae. Branstetter (1987a) and Parsons (1983b) approximated the size of females at maturity at 660 mm PCL and from 650 to 690 mm PCL, respectively. We found, however, that females mature at a smaller size, from 590 to 610 mm PCL. The reproductive seasonality of R. terraenovae in our study appeared to lack synchrony; males reached their reproductive peak in April and females in May and June. Mature males dissected in late May and June had vis- ibly atrophied testes compared to those collected in April and early May. However, their seminal vesicles were still highly swollen and contained large amounts of semen. This condition indicated that male R. terraenovae were 20 - r 1.2 18 - 1 16- £ 14 - ^ b^ o ■04 1 X 3 a ■0.2 » 2 - n - Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Figure 4 Mean gonadosomatic index and mean ovarian egg diameter by month for female R. terraenovae . Open circles indicate females (n=275). closed circles indicate males l/i=214). Error bars represent mean ± one standard error 8l 5 T 7 - i « 6- Q. =) 18 /^ ■'■ 36/^* 57 ^ C 10 'V^ ^^ J 20 40 ^L^^ S 3- 1 - 2 •^6^ 'h -^ "o -^s- ■h ^s 'h Female precaudal length (mm) Figure 5 Mean litter size on female size class. Solid line represents best-fit quadratic equation. Numbers indicate sample size for each data point. Flrror bars represent mean ± one standard error. still capable of mating during May and June, when female MOD values were highest. Therefore, the mating season of fl. terraenovae off the southeastern U.S. coast appeared to last from mid May to early July. Simpfendorfer (1992) noted a similar misalignment of peaks in reproductive seasonality between the sexes in R. taylori. The largest litters noted in our study contained eight pups (/?=4l. This increases the maximum litter size re- ported for R terraenovae in the northwestern Atlantic Loefer and Sedberry: Life history of Rhizoprionodon terraenovae off the southeastern United States 81 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Figure 6 Mean total length by month for embryos collected from April 1997 through March 1999. Numbers indicate sample size for each data point. Error bars represent mean + one standard error 1200 1 1000- PCL = 61 80+ 124 48CR ^^ {mm) CO o o " iifirf^ ^ 600 ^^0^' ra a> -400 - -ii^^\ Q. _^JgPP^ X POO - ^^^ 00 10 2 0 30 40 50 60 70 80 90 Centrum radius (mm) Figure 7 Linear regression of precaudal length I mm) on centrum radius (mm i for R Icr- | racnovae; (X) represents males, (Ol represents females. (Parsons, 1983b; Castro and Wourms, 1993). Early reports of up to 12 pups in sharpnose sharks collected from Cuban waters (Bigelow and Schroeder, 1948) were likely the re- sult of misidentification (Castro and Wourms, 1993). Age and growth The PCL-CR regression line slightly overestimated cen- trum radius for large individuals. This trend has also been noted in large female Carcharhiniis obscurus (Natanson et al, 1995) and appears to result from a change in the slope of the linear relationship as growth becomes asymptotic near the maximum length of the species. This phenomenon was deemed to have a minimal effect on the linear regres- sion formula used in this study. Although the linear rela- tionship appears to undergo an immediate change in slope at about 700 mm PCL. there are not enough data following this change (that is. the animal does not increase substan- 82 Fishery Bulletin 101(1) Jan Feb War Apr May Jun Jul Aug Sep Oct Nov Dec Figure 8 Mean relative marginal increment (mm) by month for age classes 1+ through 7+. Numbers indicate sample size for each data point. Error bars represent mean + one standard error. Table 1 Mean, minimum, and maximum length s-at-age (mm) and statistics for observed actual and back-calculated ages (O-lO-i- years). O-i- 1 + 2+ 3-^ 4-1- 5-1- 6-1- 7-1- 8+ 9-1- lO-i- Females Back-calculated mean 249 452 556 619 665 698 722 740 754 775 777.9 minimum 189 307 422 521 563 600 627 673 711 754 maximum 301 573 646 742 764 795 795 800 785 804 SD 19 32 35 36 35 34 29 28 22 19 n 379 305 273 232 186 137 90 42 13 5 1 Observed mean 320 513 629 676 700 717 741 755 762 788 787.0 minimum 197 391 469 606 615 345 663 688 726 764 maximum 465 624 707 780 765 805 812 810 796 813 SD 63 51 49 33 30 66 26 31 23 20 n 123 32 42 46 50 47 48 29 8 4 1 Males Back-calculated mean 247 452 564 634 675 695 708 717 728 715 minimum 191 310 372 519 582 625 651 690 706 maximum 317 553 681 760 778 809 753 764 743 SD 21 36 45 40 38 32 24 21 16 n 337 260 225 191 159 102 49 15 4 1 Observed mean 323 509 600 676 716 722 722 732 743 720 minimum 200 340 387 578 623 653 661 699 729 riKixinnini 466 602 730 777 796 828 763 773 757 SD 63 59 69 46 39 34 24 21 14 n 116 35 34 32 57 53 35 10 3 1 Loefer and Sedberry: Life history of Rhtzopnonodon lerraenovae off tfie southeastern United States 83 Females Age 0+ n =74 Age 1 + n =30 0 80 Males 0 70 0 60 OSOi A AgeO+ n =76 0 40 H / \ 0 30 / \ 0 20 / \ 0 10 / \ Age 1 + n =34 Age 4+ n =47 Age 4 + n =41 Focus to increment distance (mm) Figure 9 Focus to increment distance (mm) frequency distributions for males and females age 1+ to 7+. The first distribution represents the birth mark in all cases, subsequent distributions represent (from left to right! measurements to the first, second, third, fourth, fiflh, sixth, and seventh increments, respectively. 84 Fishery Bulletin 101(1) 40- 35- O 30- E E 25. w ra 20. E ^ 15. O e 0 8 0 o 8 o 8 o 0 0 if » = o 1 i 1 Q. 05. n n Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 10 Centrum radius of age 0+ R. terraenovae, by month. (♦) represents individuals without a birth mark, (O) represents individuals with a birth mark. Table 2 Von Bertalanffy growth parameters oi Rhizoprionodon terrraenovae from the southeastern coast of the United States. Von Berta- lanfTy growth parameters from previous studies in the northern Gulf of Mexico are included for comparison. Von Bertanalanffy growth parameters L^ SE SE SE Sex (mm PCD K ^0 ofZ.„ ofK oU, n Data type Study Females 752 0.52 -1.07 5.33 0.025 0.052 433 age class current Males 746 0.53 -1.07 6.83 0.030 0.059 379 Sexes combined 750 0.52 -1.07 4.23 0.019 0.039 812 Females 749 0.49 -0.94 4.60 0.020 0.046 433 estimated actual age current Males 745 0.50 -0.91 5.93 0.024 0.052 379 Sexes combined 748 0.50 -0.92 3.65 0.015 0.034 812 Females 738 0.46 -0.90 2.64 0.006 0.015 1856 back-calculation current Males 726 0.53 -0.79 3.14 0.009 0.016 1447 Sexes combined 732 0.49 -0.85 2.02 0.006 0.011 3303 Sexes combined 820 0.36 -0.99 — — — 20 estimated actual age Branstetter 11987a) Males 709 0.39 to 0.53 -2.01 — — — 15 age class Parsons(1985) tially in length following the .shift) to reliably fit a second regression line. The back-calculation equation used in our study does not employ the linear regression in its calcula- tions and was minimally affected by the negative bias that this phenomenon had on the slope of the regression. Marginal increment analysis in the present study in- dicated that growth increments form in summer. This finding is contrary to that of earlier studies on R. ter- raenovae, which indicated winter deposition (Parsons, 1985; Branstetter and McEachran, 1986; Branstetter 1987a). However, other species in this genus have been shown to deposit increments during the summer months. Simpfendorfer (1993) demonstrated summer (February) increment deposition in R. taylori in Australian waters. He cited stress during the breeding season as a possible cause because hepatosomatic index and condition factor in both sexes were low during the mating season, an indica- tion of probable stress. Furthermore, growth increments in elasmobranchs may reflect periods of slow calcareous accretion that have been compressed by increased growth (Gelsleichter, 1998). This pattern of deposition may result in increments from periods ol' slow growtii not becoming Loefer and Sedberry: Life history of Rhizopnonodon terraenovae off tfie soutfieastern United States 85 900 - 800 - 700 - E 600- g' 500- ™ 400 ■ 3 to m 300 ■ dI 200 ■ 100 - ^gsi"^^^ "^ - Branstetter (1987b) Parsons (1985) Present study 01 23456789 10 11 12 Age (years) Figure 11 A comparison of growth curves for R. terraenovae from the present study with those from previous works by other authors. Growth curve from the present study is based on estimated actual ages for both sexes (parameter values are presented in Table 2). visible for some time after their actual formation until enough new tissue has grown distally to provide the compression and contrast necessary for reliable identifica- tion. In other words, the increments observed in our study first became visible in July, but may have actually formed one to several months earlier. It should be noted that the methods of vertebrae processing and examination followed during our study were more similar to those of Simpfen- dorfer (1993) than to those of Parsons (1985) or Branstet- ter (1987a). These methods may have contributed to the close similarity found in both the physical appearance (i.e. that of "check marks" as opposed to pairs of growth bands) and temporal deposition of increments between our study and that of Simpfendorfer ( 1993). We found young of the year R. ter-racnovae with and without a birth mark. This is unusual in that most studies that have documented the presence of a birth mark have found one present in all specimens examined (e.g. Casey et al., 1985; Branstetter, 1987b; Simpfendorfer, 1993). Simp- fendorfer ( 1993) suggested that the "birth" mark in R. tay- lori was probably laid down sometime after birth because he observed the same overestimation of size at birth by back-calculations noted previously in our study. No tempo- ral estimation of the lag between birth and the formation of a birth mark has been published. The young-of-the-year R. terraenovae examined during our study demonstrated a distinct temporal transition from the lack of a birth mark to the presence of a birth mark iFig. 10). The data sug- gest that the birth mark is not actually laid down at birth in June, but approximately one month later in July. This time lag may explain the overestimation of size at birth by back-calculation. It is possible that the mechanism for the formation of the birth mark lies in the switch from embryonic to normal somatic growth, which may not occur immediately following parturition. The von Bertalanffy growth parameters derived for our study demonstrated differences from those derived for previous studies (Fig. 11). Parsons (1985) estimated an L^ of 709 mm PCL, and Branstetter (1987a) 820 mm PCl" L„ for our study was 745 mm PCL for males, and 749 mm PCL for females. The ^q value produced by Parsons was low at -2.01 yr, whereas the values produced by Branstetter (-0.99 yr) and our study (-0.90 yr for males and -0.94 yr for females) agreed well with the known gestation period of approximately 11 months. Parsons (1985) estimated K by several methods, resulting in values ranging from 0.39 to 0.53. The higher values agreed well with the estimates of our study (0.49 for females and 0.50 for males). Brans- tetters ( 1987a) estimate of K was 0.36, lower than that of the current study. Yearly growth rate estimates by Parsons (1983b) and Branstetter (1987a) revealed an increase of 133 to 211 mm PCL during the first year of life. 94 mm during the second year, 55 mm during the third year, and 16 to 32 mm growth after maturity. We found similar, though slightly higher growth rates: 198 to 202 mm PCL during the first year, 100 to 108 mm during the second, 63 to 69 mm dur- ing the third, and from 0 to 46 mm thereafter. Parsons (1985) determined age at maturity by three methods: extrapolation of growth rates to size at maturity, the VBGE, and Holdens method (Holden, 1974). The esti- mates produced by these methods ranged from 2.0 to 3.5 for males, and 2.4 to 3.9 for females. Branstetter (1987a) compared his von Bertalanffy-derived estimates to those of Parsons (1985), and found his results in general agree- ment with Parsons" higher estimates. Branstetter (1987a) 86 Fishery Bulletin 101(1) thus concluded that males mature in three years and fe- males in four. In our study, males reached full maturity at 2.4 to 2.6 years of age, making them functionally mature at the third breeding season following birth. Females were found to mature at 2.2 to 2.5 years, which would also re- sult in full maturity just prior to the third postnatal breed- ing season. Although it was noted in both previously cited studies that males matured six months to one year earlier than females, no such discrepancy in age at maturity be- tween the sexes was apparent in our study. Differences between studies The differences between this and previous studies on R. terraenovae are likely a combination of many contributing factors. These studies were conducted in different regions at separate times and may reflect clinal or temporal differ- ences (or both) between Gulf of Mexico and northwestern Atlantic R. tej-raenovae populations. However, there are other contributing factors that must be considered as well, most notably differences in data collection and analysis techniques. Parsons' ( 1985) growth curves were based on males and were grouped into age classes (not assigned actual ages). His von Bertalanffy parameters were then derived by us- ing the Ford and Walford plot method (Parsons, 1985), re- quiring the use of mean lengths of each age class. This age class grouping does not take into account growth since the deposition of the last increment, and may therefore bias the Ford and Walford plot by pulling the data to a faster asymptote (Branstetter and McEachran, 1986; Branstet- ter, 1987a;). This bias produced a low L, (706 mm PCD and /f,(-2.01 years) in Parsons' estimates (Branstetter and McEachran, 1986; Branstetter, 1987a). This phenomenon was not evident in VEGE estimates based on age classes in our study, which were very similar to estimates based on actual ages (Table 2), and was probably due to the fact that iterative fitting of age data to the VBGE by computer software (an option unavailable to Parsons at the time of his study) is less sensitive to unaddressed growth than the graphically based Ford and Walford plot method. Although the aging technique used by Branstetter (1987a) was similar to that of our study (counts on lon- gitudinal sections of cervical centra). Parsons' (1985) aging technique took ring counts from the face of centra that had been removed from a more posterior region of the vertebral column than the region chosen in our study. It has been stated by several authors (Branstetter and McEachran, 1986; Martin and Cailliet, 1988; Kusher et al.. 1992) that increment counts made from sections of verte- bral centra are generally preferable to those taken from the face of unsectioned centra. Sectioned centra allow for better documentation of the increment structure near the edge because the increments become narrower and more difficult to delineate with increasing age (Branstetter and McEachran, 1986; Martin and Cailliet, 1988; Kusher et al., 1992). This distinction is critical when the potential consequences of age underestimation (including overesti- mation of K', growth rate, and maximum sustainable yield) are considered. Based on comparison of our work to that of previous studies (Branstetter, 1981, 1987a; Parsons, 1983a, 1983b, 1985), there may be differences between the Gulf of Mexico and southeastern U.S. Atlantic populations of Atlantic sharpnose sharks. The question then becomes whether these differences are clinal or temporal in nature. Clinal variation, for instance, may explain the differences noted in size and age at maturity in female/?, terrae/iovae. Simpfendorfer ( 1993 ) noted differences in size at maturity between populations of R. taylori in Australia, as did Par- sons (1993) and Carlson et al. (1999) between populations of Sphyrna tiburo and Carcharhinus acronotus. respec- tively, off the Gulf coast of Florida. However, the extended time frame between the current and previous studies ( 15 to 20 years), also opens the possibility that the differences are related to a temporal change in population structure of the species across the entire Gulf and Western Atlantic region. In the earlier studies, data were collected during a time when fishing pressure (both directed and indirected) on R. terraenovae was lower than at present, and fisheries were shown to have dramatic effects on shark populations in less time (Anonymous'). The differences noted between the studies may thus be a manifestation of temporal changes in population structure of the species as a whole over the last two decades. A more current study on Gulf of Mexico R. terraenovae is needed to properly address these potential population differences. Conclusion Small shark species such as R. terraenovae tend to show rapid growth in the first few years of life and a dramati- cally slower growth rate once maturity is reached. This aspect of their growth complicates age estimation by ver- tebral increments because the most recent marks in older specimens are so closely spaced that accurate counting and measurement become problematic. The overlapping of increments in these older specimens or the lack of iden- tifiable increment formation altogether due to asymptotic growth may lead to an underestimation of ages in large adults. Althhough the maximum age demonstrated in our study was ll-i- years, the actual life span ofR. terraenovae may be longer The life history parameter estimates that have been pre- sented in our study are based on one of the largest short- term samples collected for any study of elasmobranch life history to date. The most significant aspect of this study is the documentation of differences in size and age at matu- rity between female R. terraenovae in the Gulf of Mexico and females off the southeastern U.S. coast. A difference in age of maturity of one year in an animal with a relatively short life span, such as R. terraenovae, can have a dramatic effect on the outcome of population models (see Cortes, 1995). Although the documentation of age at maturity dif- ferences by different researchers may be highly susceptible Anonymous. 1993. Fishery management plan for sharks of the Atlantic Ocean, 167 p. U.S. Dep. Commerce., NOAA, NMFS, Silver Spring, MD 20910. Loefer and Sedberry: Life history of Rhizopnonodon terraenovae off tfie soutfieastern United States 87 to analytical bias during the aging process, the documenta- tion of differences in size at maturity is unmistakable. Acknowledgments This project was funded by the University of Charleston (South Carolina), The Marine Resources Monitoring, Assessment and Prediction program (MARMAP), the Cooperative Atlantic States Shark Pupping and Nursery Survey (COASTSPAN), and the W. F. Pate Memorial fund. The authors would like to thank the following: MARMAP personnel; the Southeast Area Monitoring and Assess- ment Program (SEAMAP) personnel; Glenn Ulrich and the crew of the RV Anita: Dean Grubbs, Jack Musick, and the crew of the RV Bay Eagle: Bill Roumillat and the SCDNR Inshore Fisheries Project; Reese Hair and the crew of the FV Malachi HI: Steve Johnson and the crew of the FV Miss Gina: and Gary Mckillop and the crew of the FV Boiizai, for assistance in sample collection. Pat- rick Harris. Charles Wenner, Steven Branstetter, Antony Harold, Glenn Parsons. Enric Cortes, Jim Gelsleichter. Colin Simpfendorfer, and Gregor Cailliet offered advice and constructive criticism. Literature cited Anonymous. 1998. JMP: Statistics and graphics guide, .593 p. SAS Institute, Inc., Cary, NC. Beamish. R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-98.3. Bigelow, H. B., and W. C. Schroeder 1948. Sharks. In Fishes of the western North Atlantic, part one, vol. 1 (J. Tee-Van, C. M. Breder, S. F. Hildebrand, A. E. Parr and W. C. Schroeder, eds. ). p. 59-546. Mem. Sears Found. Mar Res., Yale Univ. Branstetter, S. 1981. Biological notes on the sharks of the North Central Gulf of Mexico. Contrib. Mar Sci. 24: 13-34. 1986. Biological parameters of the sharks of the Northwest- ern Gulf of Mexico in relation to their potential as a com- mercial fishery resource. Ph.D. diss., 138 p. Texas A&M University. TX. 1987a. Age and growth validation of newborn sharks held in laboratory aquaria, with comments on the life history of the Atlantic sharpnose shark. Rhizopnonodon terraenovae. Copeia 1987(21:291-300. 1987b. Age and growth estimates for blacktip. Carcliarhi- niis limbatus. and spinner, C. brevipinna. sharks from the northwestern Gulf of Mexico. Copeia 1987(4):964-974. 1990. Early life-history implications of selected carcha- rhinid and lamnoid sharks of the northwest Atlantic. //; Elasmobranchs as living resources: advances in the biol- ogy, ecology, systematics, and the status of the fisheries (H. L. Pratt Jr, S. H. Gruber, and T. Taniuchi. eds.), p. 17-28. U.S. Dep. Commer. NOAA Tech. Rep. NMFS 90. Branstetter, S., and J. D. McEachran. 1986. Age and gi-owth of four carcharhinid sharks common to the Gulf of Mexico: a summary paper. In Indo-Pacific fish biology: proceedings of the second international confer- ence on Indo-Pacific fishes (T. Uyeno, R. Arai, T. Taniuchi, and K. Matsuura, (ds.), p. 361-371. Ichthyological Soc. of Japan, Tokyo. Brown, C. A., and S. H. Gruber 1988. Age assessment of the lemon shark, Negapriun bre- virostns, using tetracycline validated vertebral centra. Copeia 1988(31:747-753. Campana, S. E. 1990. How reliable are growth back-calculations based on otoliths? Can. J. Fish. Aquat. Sci. 47:2219-2227. Carlson, J. K., E. Cortes, and A. G. Johnson. 1999. Age and growth of the blacknose shark. Charchannus acronotus in the Eastern Gulf of Mexico. Copeia 1999(3 1: 684-691. Casey J G., H. L. Pratt Jr, and C. E. Stillwell. 1985. Age and gi-owth of the sandbar shark {Carcharhiniis plumbeus) from the western North Atlantic. Can. J. Fish. Aquat. Sci. 42:963-975. Castro, J. I. 1983. The sharks of North American waters, 180 p. Texas A&M Univ Press, College Station, TX. 1988. Investigations in the reproductive biology of sharks. Ph.D. diss., 110 p. Clemson Umv, SC. 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. Fishes 38:37-48. Castro, J. I., and J. P. Wourms. 1993. Reproduction, placentation, and embryonic develop- ment of the Atlantic sharpnose shark. Rhizopnonodon terraenovae. J. Morph. 218:257-280. Clark, E., and K. von Schmidt. 1965. Sharks of the central gulf coast of Florida. Bull. Mar Sci. 15:13-83. Cortes, E. 1995. Demographic analysis of the Atlantic sharpnose shark, Rhizoprionodon terraenovae, in the Gulf of Mexico. Fish. Bull. 93:57-66. Gelsleichter, J. 1998. Vertebral cartilage of the clearnose skate. Raja egtan- teria: development, structure, ageing, and hormonal regu- lation of growth. Ph.D diss., 215 p. College of William and Mary, Gloucester Point, VA. Heist, E. J., J. A. Musick, and J. E. Graves. 1996. Mitochondrial DNA diversity and divergence among sharpnose sharks, Rhizopnonodon terraenovae. from the Gulf of Mexico and Mid-Atlantic Bight. Fish. Bull. 94:664- 668. Hoenig, J. M., and C. A. Brown. 1988. A simple technique for staining growth bands in elas- mobranch vertebrae. Bull. Mar Sci. 42:334-337. Holden, M. J 1974. Problems in the rational exploitation of elasmobranch populations and some suggested solutions. /:; Sea fisher- ies research (F. R. Harden Jones, ed.), p. 117-137. Elek (Scientific books), London. Kusher, D. I., S. E. Smith, and G. M. Cailliet. 1992. Validated age and growth of the leopard shark, Tiia- kis semifasciata, with comments on reproduction. Envi- ron. Biol. Fishes 35:187-203. Martin, L. K., and G. M. Cailliet. 1988. Age and growth determination of the bat ray, Myli- obatis californica, in central California. Copeia 1988(3): 762-773. Media Cybernetics 1999. Optimas image analysis software, version 6.51. Me- dia Cybernetics, L. P., Silver Spring, MD. 88 Fishery Bulletin 101(1) Natanson, L. J., J. G. Casey, and N. E. Kohler. 1995. Age and growth estimates for the dusky shark, Car- charltinus ohscurus, in the western North Atlantic Ocean. Fish. Bull. 93:116-126. Parsons, G. R. 1981. Growth and reproduction of the Atlantic sharpnose shark, Rhizoprionodon terraenovae (Richardson). M.S. thesis, 71 p. Univ. of South Alabama, Mobile, AL. 1983a. An examination of the vertebral rings of the Atlantic sharpnose shark. Rhizoprionodon terraenovae. Northeast Gulf Sci. 6:63-66. 1983b. The reproductive biology of the Atlantic sharpnose shark, Rhizoprionodon terraenovae (Richardson). Fish. Bull. 81:61-73. 1985. Growth and age estimation of the Atlantic sharpnose shark, Rhizoprionodon terraenovae: a comparison of techniques. Copeia 1985(1 ):80-85. 1993. Geographic variation in reproduction between two pop- ulations of the bonnethead shark, Sphyrna tiburo. Envi- ron. Biol. Fishes 38:25-35. Simpfendorfer, C. A. 1992. Reproductive strategy of the Australian sharpnose shark, Rhizoprionodon taylori (Elasmobranchii: Carcharh- inidae), from Cleveland Bay, Northern Queensland. Aust. J. Mar. Freshwater Res. 43:67-75. 1993. Age and growth of the Australian sharpnose shark, Rhizoprionodon taylori, from North Queensland, Australia. Environ. Biol. Fishes 36:233-241. Skomal, G. B. 1990. Age and growth of the blue shark, Prionace glauca, in the North Atlantic. M.S. thesis, 82 p. Univ. of Rhode Island, Kingston, RI. Sminkey, T. R., and J. A. Musick. 1995. Age and growth of the sandbar shark, Carcharhinus plumbeus, before and after population depletion. Copeia 1995(4):871-883. von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquiries on growth laws II). Hum. Biol. 10:181-213. 89 Abstract— We present a method to integrate environmental time series into stock assessment models and to test the significance of correlations between population processes and the environmental time series. Param- eters that relate the environmental time series to population processes are included in the stock assessment model, and likelihood ratio tests are used to determine if the parameters improve the fit to the data significantly. Two approaches are considered to integrate the environmental relation- ship. In the environmental model, the population dynamics process (e.g. recruitment) is proportional to the environmental variable, whereas in the environmental model with pro- cess error it is proportional to the environmental variable, but the model allows an additional temporal varia- tion (process error) constrained by a log-normal distribution. The methods are tested by using simulation analy- sis and compared to the traditional method of correlating model estimates with environmental variables out- side the estimation procedure. In the traditional method, the estimates of recruitment were provided by a model that allowed the recruitment only to have a temporal variation constrained by a log-normal distribution. We illus- trate the methods by applying them to test the statistical significance of the correlation between sea-surface tem- perature (SST) and recruitment to the snapper (Pagrus auratus) stock in the Hauraki Gulf-Bay of Plenty, New Zea- land. Simulation analyses indicated that the integrated approach with additional process error is superior to the traditional method of correlating model estimates with environmental variables outside the estimation pro- cedure. The results suggest that, for the snapper stock, recruitment is posi- tively correlated with SST at the time of spawning. A general framework for integrating environmental time series into stock assessment models: model description, simulation testing, and example Mark N. Maunder George M. Watters Inter-Amencan Tropical Tuna Commission Scnpps Institution of Oceanography 8604 La Jolla Shores Drive La Jolla, California 92037-1508 E-mail address (for M N Maunder) mmaunderfgiiattcorg Manuscript accepted 20 September 2002. Fish. Bull. 101:89-99 (2003). Identifying a clear relationship between an environmental variable and pro- cesses in the dynamics of the population (recruitment, natural mortality, growth) or the fishery (catchability) would al- low improved estimation and prediction of model parameters and derived quan- tities. It is well known that the environ- ment plays a large role in the population dynamics and catchability offish stocks. Many researchers (Green, 1967; Joseph and Miller, 1989; Hinton and Nakano, 1996; Lehodey et al., 1997; Shepherd et al., 1984) have identified correlations between population processes and envi- ronmental factors, and others (Hunter, 1983; Bertignac et al., 1998; Lehodey et al., 1998) have suggested hypotheses for the underlying causes of these cor- relations. Incorporation of environmen- tal time series into stock assessment models may provide additional informa- tion to help estimate model parameters, particularly when fishing observations (catch, effort, length-frequencies) are missing. For the management of fish stocks, it can be advantageous to be able to predict future catch rates and popula- tion sizes. Because there is often a delay due to the propagation of the recruit- ment signal in the population structure or because statistical and numerical models can provide predictions for some environmental variables (e.g. tempera- ture) (or for both reasons), the relation- ship can be used to predict future catch rates or population sizes. Statistical catch-at-age analysis (e.g. Fournier and Archibald, 1982; De- riso et al, 1985; Methot, 1990) is more appropriate than cohort analysis (vir- tual population analysis) to include relationships between an environ- mental variable and processes in the dynamics of the population. In cohort analysis, if there are missing data, they are simply extrapolated without any statistical methods, which may cause bias in the parameter estimates. Also, the potential correlation with an envi- ronmental series is calculated outside of the estimation procedure, producing several disadvantages, including the loss of information and the difficulty of propagating uncertainty (Maunder, 1998a, 2001a, 2001b). However, in sta- tistical catch-at-age analysis, there are robust statistical methods (maximum likelihood, with all the parameters esti- mated together by obtaining the best fit between predicted and observed data) that allow inclusion of multiple data sets and the integration of the environ- mental series into the stock assessment model. These methods automatically al- low for missing data and provide confi- dence inteivals, and the hypotheses can be easily incorporated and tested. The methods used to integrate the environmental series into the stock as- sessment model can be applied to differ- ent processes in the population, but are illustrated here with the case of recruit- ment. Recruitment is the fundamental process in the population dynamic that is responsible for the fluctuations of the stock size. Many studies (e.g. Francis, 1993) show that environmental vari- ables affect the recruitment. In statisti- cal catch-at-age analysis, recruitment combines an average value with an an- nual deviate, constrained by using a 90 Fishery Bulletin 101(1) distributional assumption (e.g. Maunder and Starr, 2001). This constraint allows the estimation when there is no in- formation (i.e. missing data). Traditional methods that re- late recruitment to environmental factors use correlation analysis of an environmental time series with estimates of recruitment from a stock assessment model. For example, cohort analysis is first used to generate a time series of re- cruitment. Then the time series of recruitment is regressed against sea-surface temperature (SST). This two-step pro- cedure has a number of disadvantages (Maunder, 1998a, 2001a, 2001b), including the loss of information and the difficulty of propagating uncertainty. We introduce a method suggested by Maunder (1998a; see Maunder and Starr, 2001) that incorporates environ- mental time series into stock assessment models and tests the significance of the correlation between the population processes and the environmental time series. We test the model with simulated data and compare the results to correlating model estimates with environmental vari- ables outside the estimation procedure. We illustrate this method with an application that investigates the correla- tion between SST and recruitment within the context of a statistical catch-at-age analyses used to assess snapper (Pagrus auratus) in the Hauraki Gulf-Bay of Plenty, New Zealand (Maunder and Starr, 2001). Materials and methods Integrating environmental indices into stock assessment models Parameters that relate the environmental time series to population processes were included in the statistical catch- at-age stock assessment model. We added additional struc- ture to the stock assessment model for each parameter of the stock assessment model (X) that was hypothesized to 1) have temporal variation, 2) be correlated with an envi- ronmental time series, and 3) have sufficient information in the data to be estimated for multiple time periods. This structure included a mean value for the stock assessment model parameter (^i), temporal deviations in the stock assessment model parameter (f,), a parameter that relates the environmental series to the stock assessment model parameter (/3), and a scaling parameter (a) that ensures that ji is the mean value for the stock assessment model parameter over the time period used in the model. X, = // exp (a + lili + e,). (1) where t = time, and /,= the value of the environmental time series at time t. The parameter a ensures that ^ is equal to the mean over the whole time period (Gilbert'; see Maunder and Starr, 2001). Therefore, a removes the log normal bias and bias caused by an unnormalized environmental time series and is defined as a = In (2) where n is the number of time periods. The additional structure requires that a set of param- eters (f,) that are constrained by a distributional assump- tion and two free parameters (|j, /3) be estimated. The distributional assumption (referred to as a "prior" in the following description and represented by the negative logarithm of the prior probability, see Eq. 3) is a prior on the degree of temporal variation in the stock assessment model parameter The default assumption is a normal distribution (assuming that the stock assessment model parameter is lognormally distributed) with mean zero and given standard deviation. Information about this distribu- tion can be obtained from estimates for similar species (e.g. Myers et al., 1995). The prior -In Prior (f |cr) = V (£^ 2ct- (3) ' Gilbert, D. J. 1999. Personal cdiiirmin. National Institute of Water and Atmospheric Research Limited. P.O. Box 14-901, Wellington, New Zealand. keeps the temporal deviations close to zero if there is no information in the data to the contrary. It is important to note that the prior is also needed to avoid making /3 a redundant parameter The parameters ^ and j5 and the set of parameters £, are estimated simultaneously with the other parameters of the stock assessment model, and the negative logarithm of the prior is added to the negative log-likelihood function of the stock assessment model. The parameter estimates are really the mode of the posterior distribution, but we treat them in a likelihood context. The influence of the environ- mental time series can be removed from the analysis by fixing /3 at zero. Therefore, likelihood ratio tests can be used to determine if the /3 parameter significantly improves the fit to the data. If the addition of /3 reduces the total negative log likelihood by more than about 1.92 units (/-, ^^^y qc^ ), then the additional parameter significantly improves the fit to the data at the 0.05 level, and there is a statistically sig- nificant correlation between the population process and the environmental time series. Similar tests can be performed to test the significance of the set of temporal deviation parameters, f,, by taking into account the number of ad- ditional parameters. Hilborn and Mangel ( 1997 ) provided a simple description of the likelihood ratio test. (The Akaike information criterion or the Bayes information criterion could also be used. ) Therefore, by fixing, or not, /J or f at zero we can define three types of statistical models; 1 Traditional model /J is (ixi'd at zero, the parameter set f, is estimated, and a significant relationship is determined by testing if the correlation coefficient between f, and the environ- mental time series is significantly different from zero. Maunder and Watters: Integrating environmental time series into stock assessment models 91 2 Environmental model /3 is estimated, each value in the parameter set f, is fixed at zero, and a significant relationship is deter- mined by testing if /i = 0, using a likelihood ratio test. 3 Environmental model with process error Both p and e, are estimated and a significant relation- ship is determined by testing if /3 = 0, with a likelihood ratio test. Simulation testing Simulation analysis was carried out to test the perfor- mance of the integi'ated approach and to compare this approach to the traditional model. A simple age-structured model (Appendix I) was set up to simulate a population for 20 years, starting from an unexploited population and gen- erating catch, effort, and catch-at-age data. The simulated recruitment was generated as having a component based on an environmental time series and a random compo- nent. Each component was given the same variance (0.6- ). The environmental time series was randomly generated with /3 = 1 for each simulation. The standard deviation of the observation error in the CPUE index, o^^p^,^, was set at 0.6, and the sample size of the catch-at-age data was set to 50. The same age-structured model was then fitted to the data to estimate the model parameters. The three models defined in the previous section (traditional model, environmental model, and environmental model with pro- cess error) were tested with the simulated data. In addi- tion to the parameters outlined in the description of the three models, average recruitment, the catchability coef- ficient, and the standard deviation of the fit to the CPUE data were also estimated. We also used a model that had constant recruitment to provide likelihood values to use in testing the significance of the environmental model. The simulation analysis was repeated 500 times for four scenarios: 1) using catch-at-age data for all years, 2) using catch-at-age data for the first 10 years, 3) using catch-at- age data for the last 10 years, 4) using catch-at-age data for all years, but using j3 = 0 when generating the simulat- ed data. Scenario 4 was used to investigate the probability of type-I error of the models when used in combination with the statistical tests. For each set of simulated data and each model, we determine how often a significant rela- tionship between the logarithm of annual recruitment and the environmental time series is detected, the estimate of the slope of the relationship between the logarithm of an- nual recruitment and the environmental time series, the estimates of average recruitment, and the depletion level (ratio of current to unexploited biomass). We also calcu- late minimum-width 95% confidence intervals for average recruitment, using the likelihood profile method for the simulated data sets with catch-at-age data for all years. Application: relating recruitment in the Hauraki Gulf-Bay of Plenty snapper stock to SST Recruitment to the Hauraki Gulf snapper iPagrua auratus I stock is correlated with temperature (Paul, 1976). The abundance of l-i- snapper in the Hauraki Gulf estimated by trawl surveys has been shown to have a positive corre- lation with SST (Francis, 1993) and air temperature (Gil- bert, 1994) around or just after the time of spawning in the previous year This relationship has also been shown with catch-at-age analysis to continue to hold as snapper enter the fishery at ages 4 and older (Maunder and Starr, 1998). We applied the integrated approach described in this study in combination with the age-structured statistical catch-at-age model described in Maunder and Starr (2001 ) to the Hauraki Gulf-Bay of Plenty snapper stock. The model was fitted to catch-at-age data and biomass estimates. The biomass estimates were available for 1985 and 1994 and were obtained from analysis of tagging data. The majority of the catch-at-age data were available from 1990 to 1997, but there were some catch-at-age data of dubious quality, small sample size, and high variability for 1970 to 1973. The annual recruitment at age 1 was estimated for the time pe- riod of the model ( 1970-98) and also for 18 age classes (ages 2 to 19) that comprised the initial conditions in 1970. Results Simulation analysis For all four sets of simulated data the environmental model had the highest probability of detecting a relation- ship between recruitment and the environmental time series (Table 1). This model had a very high probability of detecting a relationship even when there was no relation- ship in the simulated data (Table ID). This indicates that the likelihood ratio test is not appropriate for the environ- mental model (see Appendix III). For all data sets, except that with only catch-at-age data in the first 10 years, the traditional model had a higher probability of detecting a relationship between recruitment and the environmental time series than did the environmental model with process error The environmental model with process error had a lower probability of detecting a true relationship than the traditional model, but also had a slightly lower probability of type-I error (the probability of incorrectly accepting a nonexistent relationship) than the traditional model. The probability of detecting a relationship was reduced as the number of catch-at-age data sets was reduced. The environmental model with process error did not show any bias in the estimate of the slope of the relation- ship between the logarithm of annual recruitment and the environmental time series, /3 (Table 1 ). For this model, the variation in the estimates of /3 increased when fewer years with catch-at-age data were available. The environmental model showed a small negative bias and slightly more er- ror in the estimates of p. The traditional model showed a large negative bias in the estimate of the slope of the relationship between the logarithm of annual recruitment and the environmental time series, and this bias increased as the amount of catch-at-age data was reduced. The tra- ditional method also had larger error, which increased as less catch-at-age data were available. The errors in the estimates of average recruitment and ^ajr^B,, increased slightly with less catch-at-age data 92 Fishery Bulletin 101(1) Table 1 (A) Results from the simulation analysis in which all the catch-at-age data were used. 9c = percentage of data sets that produced a | significant relationship between the environmental time series and recruitment; fi = average (average absolute relative error) of the estimates of the slope of the relationship between the environmental time series and recruitment; R„ = average (average absolute relative error) of the estimates of the average recruitment; B^^^^/Bg = average error (average absolute relative error) in the estimate of the ratio of current to unexploited biomass. p was set to 1 when generating the simulated data. (B) Results from the simulation analysis using only the first 10 years of catch-at-age data. /3 was set to 1 when generating the simulated data (C) Results from the simulation analysis using only the last 10 years of catch-at-age data, fi was set to 1 when generating the simulated data. (D) Results from the simulation analysis using all the catch-at-age data, but setting /3 = 0 when generating the simu lated data. (Abso- lute, rather than relative, error was used for j3. ) EMwPE = environmental model with process error % P Ro BcJBg A Traditional 92 0.86(0.25) 996(0.05) -0.01 (0.15) Environmental 99 0.95(0.25) 1,024(0.10) 0.03(0.24) EMwPE 83 1.01(0.23) 988 (0.05) -0.04(0.14) B Traditional 57 0.45(0.55) 1.011(0.07) -0.03(0.17) Environmental 95 0.94(0.33) 1.078(0.17) 0.01(0.27) EMwPE 60 1.05 (0.30) 1,004 (0.07) -0.03 (0.15) C Traditional 81 0.68 (0.36) 1.022(0.09) 0.04(0.21) Environmental 97 0.95(0.32) 1,035(0.10) 0.04(0.25) EMwPE 74 1.00(0.26) 1.007(0.08) -0.02(0.19) D Traditional 3 -0.01 (0.20) 1.005(0.04) 0.01(0.10) Environmental 61 -0.02 (0.22) 1.038(0.09) 0.07(0.21) EMwPE 2 -0.01(0.21) 1,003(0.04) 0.01 (0.10) Table 2 Results related to the confidence intervals for average recruitment from age data. EMwPE = Environmental model with process error the simulation analys is obtained by using all the catch-at- Lower bound average Upper bound average Lower bound SD Upper bound SD True is within True is below True is above Traditional Environmental EMwPE 888 963 887 1,120 1,117 1,118 59 94 56 132 176 120 0.91 0.44 0.92 0.03 0.32 0.01 0.08 0.24 0.07 (Table 1). The errors in these estimates were shghtly greater for the environmental model (more bias and larg- er absolute error) than for the environmental model with process error and traditional model. The confidence intervals on R^ were, on average, greater for the traditional and environmental model with process error than for to the environmental model (Table 2), which greatly underestimated the width of the confidence inter- vals. However, the confidence intervals for the traditional and environmental model with process error showed the true value falling below the confidence interval less often than it fell above it. As expected, an environmental relationship was more difficult to correctly detect with the traditional model in situations with missing data (e.g. when catch-at-age data were missing in the last few years of the series), and, as stated above, using the environmental model is inappro- priate because it has a high probability of detecting a sig- nificant relationship when none exists. The environmental model also has a tendency to under estimate the width of the confidence interval for /?„ and the true value frequently falls outside of thi.s confidence interval. Therefore, the envi- ronmental model with process error is the model of choice. Application: relating recruitment in the Hauraki Gulf-Bay of Plenty snapper stock to SST The environmental model with process error has the lowest negative log-likelihood, but this model has many more parameters than the environmental model (Table 3). Maunder and Walters: Integrating environmental time series Into stock assessment models 93 Table 3 Results from applying the three methods (traditional, environmental, and environmental with process error) to the snapper appli- cation. Constant = recruitment is constant each year and equal to the average recruitment. EMwPE = environmental model with process error, n/a = not applicable. Average recruitment Constant Traditional Environmental EMwPE 13,315 (11,381-15,381) 11,406 (8,500-14,603) 13,530 (11,527-15,569) 12,029 (9,147-15,328) Number of p -ln( Likelihood) parameters n/a 482.5 3 0.20 466.8 50 0.48 473.5 4 0.55 464.4 51 Using the likelihood ratio test at the 0.05 level and com- pensating for the difference in the number of parameters being estimated in each model, we consider the environ- mental model to be the model of choice. If the likelihood ratio test were used, the environmental model with pro- cess error would be chosen over the traditional model, indicating a statistically significant correlation between SST and recruitment. Due to the weaknesses of the envi- ronmental model discussed above, we concentrated on the results of the traditional model and the environmental model with process error. The time series of estimated recruitments from the traditional model showed very little annual variation in recruitment for the first half of the time series and for the last few years of the time series (Fig. lA). This indicates that there is very little information in the data (catch-at- age) about annual recruitment for these time periods and that the prior on the recruitment residuals constrains the estimated recruitment to be close to the average recruit- ment. This result is consistent with the catch-at-age data, which, ignoring the inconsistent data from the 1970s, started in 1990. The greatest age in the catch-at-age data that had individual information was age 19; therefore the 1971 cohort is the first for which there is information. However, at the current exploitation rates, very few snap- per live to be more than 10 years of age, so that there is very little information about cohort size for any of the co- horts produced during the 1970s. The environmental model with process error indicated high variation in recruitment for the whole time period (Fig. IB). This is due to the formulation of the recruitment submodel, for which the annual anomalies are anomalies from the temperature-recruitment relationship; if there is no information in the data about recruitment for a par- ticular year, the recruitment will follow the temperature- recruitment relationship. The correlation of the estimated recruitment from the traditional model with SST had a low r-square (0.26), but it was statistically significant at the 0.05 level when a two-tailed test was used. In addition, the slope of the rela- tionship between recruitment and SST was much less for the traditional model than for to the environmental model with process error (Table 3). The estimates of recruit- ment from the traditional model included a large number of estimates that were close to the mean because there 1950 1960 1970 1980 1990 2000 B 5 ,0 ,5 ,0 ,5 ,0 1950 1960 1970 1980 Year 1990 2000 Figure 1 Annual estimates of relative recruitment strength at age 1 for the Hauraki Gulf-Bay of Plenty snap- per stock from the traditional (A) and environ- mental model with process error (Bl models was no information in the data about these recruitments. Therefore, it was inappropriate to use these recruitments to correlate with SST and, if used, they would result in a poor fit. However, a significant correlation, as obtained in this application, suggests that the correlation is probably stronger than apparent from the analysis, which should give confidence that a relationship exists and provide an incentive to apply the integrated models. The environmental model with process error did not show a statistically significant improvement over the environmental model because, ignoring the 1970s data, the catch-at-age data were available only for the last part of the time period. The recruitment anomalies were esti- mated for the whole time period, as well as for the initial conditions. Many of these recruitment anomalies had very 94 Fishery Bulletin 101(1) little information associated with them and therefore did not add anything to the estimation procedure. However, they do add additional parameters, which reduce the pos- sibility of accepting the model when using the likelihood ratio test. If the recruitment anomalies were estimated for only a limited number of years, it is likely that the envi- ronmental model with process error would be a statistical- ly significant improvement over the environmental model. Statistical tests could be carried out to determine which annual recruitment anomalies should be estimated, but this would be very time consuming. Reducing the number of annual recruitment anomalies may also cause an un- derestimation of the confidence intervals. For the snapper example, removing the anomalies for the initial conditions may be a good compromise. Discussion We have developed a general framework for integrating environmental time series into stock assessment models that appears to perform better than traditional methods. The method is flexible and it can be used to model many dif- ferent functional relationships between population or fish- ing processes and environmental time series and to include multiple environmental time series for any population model parameter (see Appendix II). Furthermore, it can be used with any statistical stock assessment model. The method can be used to test whether an environmental time series describes temporal variation in model parameters. The traditional model, which estimates annual recruit- ment within the stock assessment model and subsequently correlates the recruitment with the environmental series outside the stock assessment model, performs poorly. It has a reasonable probability of detecting a relationship be- tween recruitment and the environmental series, but this probability decreases rapidly as the number of years with missing catch-at-age data sets increases. The probability of incorrectly detecting a relationship when one is not pres- ent is low. This method has reasonable confidence-interval coverage for average recruitment and little bias or variance in the estimates of model parameters. The factor causing the poor performance of the traditional model is the large bias in the estimate of the slope of the relationship between recruitment and the environmental time series, which in- creases as the number of years with missing catch-at-age data increases. The bias occurs because the traditional model has a penalty on the absolute size of the annual recruitment deviations. This penalty constrains an annual recruitment anomaly to be close to the mean recruitment when there is little or no information about the recruit- ment in that year. Therefore, when the logarithm of the annual recruitment is correlated with the environmental time series, the estimated slope of the relationship is biased downward. Even in situations for which there is sufficient information for every recruitment anomaly, there will be a small tradeoflin the size of the anomaly, which reduces the contribution of the penalty to the objective function and the likelihood from the catch-at-age data. Unfortunately, if the penalty on the annual recruitment anomalies is removed. the estimation process can become unstable, particularly in data-poor situations for which the bias is greater. The amount of time that is required by the estimation algorithm also increases if the penalty is removed. When the penalty on the size of the recruitment anomalies is removed, the bias in estimates of the slope of the relationship between recruitment and the environmental time series is reduced when using all the catch-at-age data, but the variance in the estimates is greatly increased. In addition, when removing the penalty there was a large positive bias when using only the last 10 years of catch-at-age data and a large negative bias when using only the first 10 years of catch-at-age data. It is not known what results would be obtained if cohort analysis, which does not use a constraint on the annual recruitment anomalies, is used instead of the statistical catch-at-age analysis. It should be remembered that cohort analysis cannot be used or assumptions that are unlikely to be satisfied will have to be made when catch-at-age data are missing for some years. The environmental model, which has a deterministic relationship between recruitment and the environmental time series that is integrated into the stock assessment model, also performs poorly. This method has poor confi- dence interval coverage for average recruitment because the size of the confidence intervals are gi'eatly underes- timated. The method has larger bias and variance in the estimates of model parameters compared to the other two methods. There is a small negative bias in the estimate of the slope of the relationship between recruitment and the environmental time series. The environmental model has a very high probability of detecting a relationship between re- cruitment and the environmental series, and this probabil- ity only decreases slightly as the number of missing years of catch-at-age data sets increases. However, this model has a very large probability of incorrectly detecting a relationship when one is not present. Therefore, when using the environ- mental model, the likelihood ratio test should not be used to determine if there is a significant relationship between recruitment and an environmental time series. The value used to compare to the x- statistic in the likelihood ratio test for the environmental model is highly correlated with the catch-at-age sample size; therefore simulation analysis is needed to find the appropriate /- statistic for the given sample size (see Appendix III). This is also important for calculating confidence intervals that are also based on the X~ statistic and is the reason for the poor coverage for i?^. The environmental model with process error, which has a relationship between recruitment and the environmen- tal time series that is integrated into the stock assessment model with additional process error, performs well. This model has a reasonable probability of detecting a relation- ship between recruitment and the environmental series, but this probability is lower than those of the other two models, and decreases as the amount of data is reduced. It has a low probability of incorrectly detecting a relation- ship when one is not iircsent. These probabilities could be im[)r()ved by using simulation analysis to find the appro- priate ;f- statistic (see Appendix III). This method has rea- sonable confidence interval coverage for average recruit- ment and has little bias or variance in the estimates of Maunder and Walters: Integrating environmental time series into stock assessment models 95 model parameters. There is very little bias in the estimate of the slope of the relationship between recruitment and the environmental time series. For the environmental model with process error, when there is little or no information in the data to estimate the recruitment for that year, the penalty on the annual re- cruitment anomalies causes recruitment to be estimated close to the recruitment predicted by the relationship between recruitment and the environmental time series. Therefore, if there is a relationship between recruitment and the environmental time series, this model should pro- vide better estimates because additional information is included in the estimation procedure. This model has the favorable property that if there is no relationship between recruitment and the environmental time series, the model estimates ji to be small, eliminating any influence of the relationship between recruitment and the environmental time series, and still estimates the annual recruitment anomalies to represent the variation in annual recruit- ment. The likelihood ratio test can be used to detect a rela- tionship between recruitment and the environmental time series, and if a relationship does not exist, the results with P fixed at zero can be used. However, including [i in the es- timation procedure, even when there was no relationship between recruitment and the environmental time series, did not increase the error in the parameter estimates in relation to the model with j3 fixed at zero (see the results for the traditional model. Table ID). The method we describe can be used to integrate en- vironmental time series for parameters of the stock as- sessment model other than recruitment. The influence of the environment on catchability of the fish would be an obvious choice because there are numerous publications on the topic. For example. Green (1967) suggested that ther- mocline data would improve estimation of tuna abundance from catch and effort data, by allowing for the differentia- tion between changes in tuna abundance and catchability due to vertical distribution of tunas influenced by tempera- ture. We have used a method similar to the method that is presented in the present study to incorporate SST into the purse-seine catchability parameters for yellowfin and big- eye tuna (Maunder and Watters, 2001; Watters and Maun- der, 2001). Maunder (2001a) presented a general method to integrate the standardization of CPUE data into stock assessment models, including the integration of environ- mental variables. Growth rates have been observed to have temporal variation, and this variation has been correlated with environmental factors. Several authors have pre- sented growth curves that include temperature data (e.g. Mallet et al., 1999). Movement is another process that may be influenced by the environment. Lehodey et al. (1997) showed that spatial shifts in the western Pacific skipjack tuna population are linked to the movement of a large pool of warm water and that the movements of this large pool are related to El Nino-Southern Oscillation events. Once a correlation between the environmental time se- ries and the population process has been determined, this relationship can be used to improve the predictive abil- ity of the model. For example, if a relationship between SST at the time of spawning and recruitment has been determined, and the age at recruitment to the fishery is 3 years, recruitment to the fishery can be estimated 3 years in advance. One should be cautious about assuming that these relationships are valid and will continue to hold into the future, however Hilborn and Walters ( 1992) cautioned about using environmental data because there are many environmental indices that one can try, and if the data set has a few large and a few small observations, it is likely that one of the environmental data sets will correlate with the data. Myers (1998) reviewed a number of published cor- relations between recruitment and environmental factors and found that few of the correlations held when retested at later dates. Maunder and Starr ( 1998) also advised cau- tion because they found that a strong cohort may not enter the fishery when expected because of variations in growth rates. We have found that, when applying this method to the bigeye tuna data, there is an inconsistency in the pre- 1997 data and the data for 1997 and 1998 caused by much stronger than expected year classes entering the fishery in 1997 and 1998. There is also difficulty in deciding on the management strategy if environmental regime shifts are influencing the productivity of the stock (Maunder, 1998b). An advantage of the integrated approach, particularly the environmental model with process error, is that it more fully describes the uncertainty in the relationship be- tween the population process and the environmental time series, and therefore this uncertainty can be included in any management advice based on the relationship. Conclusions Integrating environmental relationships in a statistical stock assessment model is an improvement over the tra- ditional statistical model when there are large gaps in the data. However, it is important to include process error to avoid the high probability of detecting spurious correla- tions seen in the environmental model when using the like- lihood ratio test. Therefore, the environmental model with process error is the model of choice because 1 ) there is no bias in the estimates, 2) when there is no relationship with the environmental series, it is equivalent to the traditional model, 3) when such a relationship exists, the recruitment estimates are improved, particularly if there are important gaps in the data, 4) it may be used for prediction, and 5) uncertainty about the relationship can be modeled. Acknowledgments We thank Dave Fournier for advice on using AD Model Builder and related software, and Bill Bayliff, Rick Deriso, Shelton Harley, and Ransom Myers for commenting on the manuscript. Literature cited Bertignac, M., P. Lehodey, and J. Hampton. 1998. A spatial population dynamics simulation model of 96 Fishery Bulletin 101(1) tropical tunas using a habitat index based on environmen- tal parameters. Fish. Oceanogr. 7:326-334. Beverton. R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish. Invest. Ser. II, Mar. Fish. G.B. Minist. Agric. Fish. Food 19, 533 p. Deriso, R. B., T. J. Quinn II. and P. R. Neal. 1985. Catch-age analysis with auxiliary information. Can. J. Fish. Aquat. Sci. 42:815-824. Forsbergh, E. D. 1989. The influence of some environmental variables on the apparent abundance of skipjack tuna, Katsuwonus pela- mia. in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm. Bull. 19:433-569. Foumier, D., and C. P. Archibald. 1982. A general theory for analyzing catch at age data. Can. J. Fish. Aquat. Sci. 39: 1 195-1 207. Francis, M. P. 1993. Does water temperature determine year class strength in New Zealand snapper (Pa.grusa!/ra;;iS,Sparidae)? Fish. Oceanogr. 2:65-72. Gilbert, D. J. 1994. A total catch history model for SNA 1. New Zealand Fisheries Research Document 94/24. 16 p. NIWA (Na- tional Institute of Water and Atmospheric Research), Well- ington, New Zealand. Granger, C. W. J. 1993. Forecasting in economics. /;; Time series prediction: forecasting the future and understanding the past (A. E. Weigend and N. A. Gershenfeld, eds.), p. 529-538. SFI Studies in the Sciences of Complexity, Proc. Vol. XV. Addi- son-Wesley, Boston, MA. Green, R. E. 1967. Relationship of the thermocline to success of purse seining for tuna. Trans. Am. Fish. Soc. 96:126-130. Hilbom, R., and M. Mangel. 1997. The ecological detective: confronting models with data, 315 p. Princeton Univ. Press, Princeton, NJ. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dy- namics and uncertainty, 570 p. Chapman and Hall, New York, NY. Hinton, M. G., and H. Nakano. 1996. Standardizing catch and effort statistics using physi- ological, ecological, or behavioral constraints and environ- mental data, with an application to blue marlin iMakaira nigricans) catch and effort data from Japanese longline fisheries in the Pacific. Inter-Am. Trop. Tuna Comm. Bull. 21:171-200. Hunter, J. R. 1983. On the determinants of stock abundance. In From year to year (W. S. Wooster, ed.), p. 11-16. Washington Sea Grant Program, Univ. Wash., Seattle, WA. Joseph, J., and F. R. Miller 1989. EI Nirio and the surface fishery for tunas in the east- ern Pacific. Japan. Soc. Fish. Ocean., Bull. 53:77-80. Lehodey, P.. J. Andre, M. Bertignac, J. Hampton, A. Stones, C. Menkes, L. Memory, and N. Grima. 1998. Predicting skipjack tuna forage distributions in the equatorial Pacific using a coupled dynamical bio-geochemi- cal model. Fish. Oceanogr 7:317-325. Lehodey, P., M. Bertignac, J. Hampton. A, Lewis, and J. Picaut. 1997. El Nino Southern Oscillation and tuna in the western Pacific. Nature 389:715-718. Mallet, J. P, S. Charles, H. Persat, and P Auger 1999. Growth modellingin accordance with daily water temp- erature in European grayling iThymallus thymallus L.). Can. J. Fish. Aquat. Sci. 56:994-1000. Maunder, M. N. 1998a. Integration of tagging and population dynamics models in fisheries stock assessment. Ph.D. diss., 306 p. Univ. Washington, Seattle, WA. 1998b. Problems with using an environmental based re- cruitment index: examples from a New Zealand snapper iPagrus auratus) assessment. In Fishery stock assess- ment models (F. Funk, T J. Quinn II, J. Heifetz, J. N. lanelli, J. E. Powers, J. J. Schweigert, P. J. Sullivan, and C. I. Zhang, eds.), p. 679-692. Alaska Sea Grant College Program Report No. AK-SG-98-01, Univ Alaska, Fairbanks, AK. 2001a. A general framework for integrating the standard- ization of catch-per-unit-effort into stock assessment models. Can. J. Fish. Aquat. Sci. 58:795-803. 2001b. Integrated tagging and catch-at-age analysis ( ITCAAN). In Spatial processes and management offish populations (G. H. Kruse, N. Bez, A. Booth, M. W Dom, S. Hills, R. N. Lipcius, D. Pelletier, C. Roy S. J. Smith, and D. Witherell, eds. ), p. 123- 146. Alaska Sea Grant College Program Report AK-SG-01- 02. Univ. Alaska, Fairbanks, AK. Maunder, M. N., and P. J. Starr. 1998. Validating the Hauraki Gulf snapper pre-recruit trawl surveys and temperature recruitment relationship using catch at age analysis with auxiliary information. New Zealand Fisheries Research Document 98/15, 23 p. NIWA (National Institute of Water and Atmospheric Research), Wellington, New Zealand. Maunder, M. N., and P. J. Starr. 2001. Bayesian assessment of the SNAl snapper (Pagnis aui'atus I stock on the northeast coast of New Zealand. N.Z. J. Mar. and Freshwater Res. 35:87-110. Maunder, M. N., and G. M. Watters. 2001. Status of yellowfin tuna in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm. Stock Assess. Rep. 1:5-86. Methot, R. D. 1990. Synthesis model: an adaptable framework for analy- sis of diverse stock assessment data. Inter North Pacif Fish. Comm. Bull. 50:2.59-277. Myers, R. A. 1998. When do environment-recruitment correlations work? Reviews in fish biology and fisheries 8:285-305. Myers, R. A., J. Bridson, and N. J. Barrowman. 1995. Summary of worldwide spawner and recruitment data. Can. Tech. Rep. Fish. Aquat. Sci. 2020, iv -i- 327 p. Paul, L. J. 1976. A study on age, gi'owth, and population structure of the snapper, Chrysophrus auratuf (Forsteri. in the Hau- raki Gulf New Zealand. New Zealand Ministry of Fisher- ies and Agriculture, Fish. Bull. 13, 62 p. Ricker, W. E. 1954. Stock and recuitment. J. Fish. Res. Board Can. 11: 559-623. Shepherd, J. G., J. G. Pope, and R. D. Cousens. 1984. Variations in fish stocks and hypotheses concerning their links with climate. Rapp. P.-V. Reun., Cons. Int. Explor Mer 185:255-267. Watters, G. M., and M. N. Maunder 2001. Status of bigeye tuna in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm. Stock Assess. Rep. 1:109- 210. Maunder and Walters: Integrating environmental time series into stock assessment models 97 Appendix I: description of simulator and estimator The following is a description of the model equations used for the data simulator and for the estimator. The model is run from an unexploited state at the start of the fishery for 20 years. The model includes 10 age classes, where the 10th age class is a plus group. Dynamics Af,,i = floexp(j8/,+ff +a) a = In ^exp(ff +/}/,,) (I.l) (1.2) A^,.. = (^,-i,o-i(l- ",-A-i')'^""" ' f0Ta3.84. Now, con- sider a simple example where Pj = W— and .V, = nexpijil^ + t)). with the penalty - In Prior (f | ct) = V 2a' and CT is a constant. Consider two models: 1) f ^ = 0 and 2) estimate f. For model 1, as A' increases x~ increases in proportion to N, as explained above, because the penalty term is constant. However, for model 2, as N gets large, the Maunder and Walters: Integrating environmental time series into stocl< assessment models 99 relative size of the penalty compared to -InLp gets smaller and therefore the estimates of e^ change so that p, gets closer to p,. Therefore, for model 2, x^ does not increase proportionally with A^. An appropriate test for the environmental model would be to produce sets of random environmental indices that have the same variance and auto-correlation as the actual environmental index to determine the appropriate value of X- that would give the desired type-I error This test would overcome the sample size effect. The method could also be used to refine the test for the environmental model with process error. 100 Abstract— \\V fxamiiu'd the spatial and temporal distribution, abundance, and growth of young-of-the-year (YOY) Atlantic croaker (Micropogonias undu- latus) in Delaware Bay, one of the northernmost estuaries in which they consistently occur along the east coast of the United States. Sampling in Del- aware Bay and in tidal creeks in salt marshes adjacent to the bay with otter trawls, plankton nets and weirs, between April and November 1996-99, collected approximately 85,000 YOY. Ingress of each year class into the bay and tidal creeks consistently occurred in the fall, and the first few YOY ap- peared in August. Larvae as small as 2-3 mm TL were collected in Septem- ber and October 1996. Epibenthic indi- viduals <25 mm TL were present each fall and again during spring of each year, but not in 1996 when low water temperatures in January and Febru- ary apparently caused widespread mortality, resulting in their absence the following spring and summer. In 1998 and 1999, a second size class of smaller YOY entered the bay and tidal creeks in June. When YOY survived the winter, there was no evidence of growth until after April. Then the YOY grew rapidly through the summer in all hab- itats (0.8-1.4 mm/d from May through August). In the bay, they were most abundant from June to August over mud sediments in oligohaline waters. They were present in both subtidal and intertidal creeks in the marshes where they were most abundant from April to June in the mesohaline portion of the lower bay. The larger YOY began egressing out of the marshes in late summer, and the entire year class left the tidal creeks at lengths of 100-200 mm TL by October or November when the next year class was ingressing. These patterns of seasonal distribu- tion and abundance in Delaware Bay and the adjacent marshes are similar to those observed in more southern estuaries along the east coast; however, growth is faster — in keeping with that in other northern estuaries. Seasonal distribution, abundance, and growth of young-of-the-year Atlantic croaker (Micropogonias undufatus) in Delaware Bay and adjacent marshes* Michael J. Miller David M. Nemerson Kenneth W. Able Marine Field Station Institute of Marine and Coastal Sciences Rutgers University 800 c/o 132 Great Bay Boulevard Tuckerton, New Jersey 08087-2004 Email address (for K. W. Able, contact author) ablea'imcsrutgers edu Manuscript accepted 20 August 2002. Fish. Bull. 101(1):100-115(2003). Atlantic croaker (Micr-opogonias undu- latus) is a commercial and sport fishery species that inhabits demersal habitats in estuarine, coastal, and continental shelf systems along the Atlantic coast of North America and in the Gulf of Mexico (Joseph, 1972). They spawn primarily over the continental shelf during a protracted spawning season that, based on the presence of larvae along the Atlantic coast, may extend from early July through March (Lewis and Judy, 1983; Cowan and Birdsong, 1985; Warlen and Burke, 1990; Hettler et al., 1997; Nixon and Jones, 1997; Able and Fahay, 1998). The exact loca- tion of spawning in the Middle Atlantic Bight (MAB) may be related to the areal extent of favorable warm bottom waters for spawning (Norcross and Austin, 1988) and may sometimes occur within or close to the mouth of Chesa- peake Bay (Barbieri et al., 1994b; Reiss and McConaugha, 1999). The larvae and postlarvae have been observed to be more abundant in deeper layers of water that may facilitate transport into and retention within estuarine nursery areas (Weinstein et al., 1980; Norcross, 1991). The young-of-the-year (YOY) usually begin to enter estuaries and tidal creeks along the Atlantic coast in September, or occasionally August, and they are often common components of the fish fauna in tidal creeks and estuaries until fall of the next year from New Jer.sey southward along the Atlantic coast and in the (lull'of Mexico (Chao and Musick, 1977; Knudsen and Herke, 1978; Weinstein, 1979; Currin et al., 1984; Ross, 1988; Able and Fahay, 1998). In some years there is a second pulse of small YOY that arrives in estuaries along the Atlantic coast in the spring or summer; this pulse may be the offspring from later spawning (Chao and Musick, 1977; Ross, 1988). In general, the YOY use estuarine habitats with salinities ranging from almost pure freshwater to seawater (Miglareseetal., 1982). Although YOY Atlantic croaker are present in some Atlantic coast estua- rine habitats during the winter (Ha- ven, 1957; Bearden, 1964; Dahlberg, 1972; Chao and Musick, 1977; Shenker and Dean, 1979; Bozeman and Dean, 1980; Able and Fahay, 1998), they ap- pear to experience winter mortality in the MAB in years with unusually cold winters (Massman and Pachcco, 1960; Joseph, 1972; Chao and Musick, 1977; Wojcik, 1978). Recent laboratory studies have found that YOY Atlantic croaker do not survive in sustained water temperatures of 3°C or lower I Lankford and Targett, 2001 ); therefore extended periods of low winter water temperatures may have drastic effects on their overwinter survival in some estuaries. Previous studies have indicated that YOY Atlantic croaker reach about 107-187 mm TL after their first year Contribution 2002-22 from the Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ 08901. Miller et al.: Distribution, abundance, and growth of Micropogonias undulatus 101 of growth in estuaries along the Atlantic coast and 102- 250 mm TL in the Gulf of Mexico (Knudsen and Herke, 1978), but only a few studies have reported seasonal growth rates (Hansen. 1969; Knudsen and Herke, 1978). Length-frequency based growth rate estimates for the first year of growth for YOY along the Atlantic coast have ranged from 0.32 to 0.41 nim/d ( Knudsen and Herke, 1978). However, these were based on the entire year, including the larval and early juvenile period when analysis of otolith daily growth rings indicates much slower growth rates of 0.18-0.41 mm/d during the fall and winter months (Nixon and Jones, 1997). Length-frequency data from estuarine nursery areas clearly indicate that most growth occurs during the spring and summer months (Haven, 1957; Chao and Musick, 1977; Ross, 1988; Able and Fahay 1998). Despite various studies of YOY Atlantic croaker in some areas of the Atlantic and Gulf coasts, there is relatively little known of their early life history near the northern part of their range in the MAB and this is especially true for Delaware Bay. Our four-year study used extensive collections in Delaware Bay and in adjacent tidal marsh creeks to describe the timing of Atlantic croaker ingress, their seasonal abundance and size, growth rates, and the timing of their egress out of the marshes. Methods Study sites Delaware Bay is the estuary of the Delaware River and encompasses about 1878 km^ of open water along the southern edge of New Jersey and the northern edge of Delaware (Fig. 1). It has a relatively deep area (10-30 m) in the middle of the lower bay, bordered by narrow shoals and flanked by extensive tidal flats and salt marshes, which contain an additional 85.5 km- of open water in tidal creeks bordered by approximately 640 km- of marsh- plain area. Depending on the amount of river discharge, salinities range from 30-31%f at the mouth of the bay, to 1-10'^f in the lower Delaware River (Table 1; Cronin et al., 1962; Garvine et al., 1992). Ichthyoplankton survey Catch data from an ichthyoplankton survey (Table 2) was used to analyze the distribution, abundance, and size of larval Atlantic croaker in Delaware Bay and the lower Delaware River from April to October 1996. Sampling was performed during daylight hours once a month in April and October and twice a month from May to September Each sampling period included one tow at 70 randomly selected stations distributed among eight designated sam- pling zones (Fig. 1). Samples were collected with a 1-m diameter plankton net (0.5-mm mesh) deployed with a depressor in single stepwise oblique tows from the surface to the bottom. Tows were made at a speed of 1.4-1.9 knots for four to six minutes in the direction of the tidal flow. Up to 50 individuals were measured to the nearest millimeter total length (TL) from each sample. DELAWARE 4 15 km Scale Figure 1 Locations of salt marsh tidal creek sampling sites (1996-99) and designated sampling zones (1-8) in the Delaware Bay and in the lower Delaware River for the ichthyoplankton (April-October 1996) survey and the otter trawl (April-October 1996-98) survey. The heavier line across the bay indicates the boundary between the upper (5-8) and lower ( 1-4) sampling zones and marsh sites. Otter trawl survey We used catch data from a three-year otter trawl survey (Table 2) to analyze the distribution, abundance, and growth of settled YOY Atlantic croaker in Delaware Bay and the lower Delaware River Sampling was performed during daylight hours twice a month from April to Octo- ber in 1996 and once a month in 1997 and 1998, at 40 stations divided up among the same eight sampling zones of the ichthyoplankton survey (Fig. 1). Station locations were selected by using a stratified random sampling design from a grid of 1002 stations, excluding the stations over the deepest water near the mouth of the bay in zone 1. There were eight stations sampled each month in zone 3, six in zone 4, and four in all the other zones. Trawling was done with 4.9-m otter trawls (6 mm stretched codend mesh), made against the prevailing direction of the tide at a speed of 1.8 m/sec for 10 minutes. Up to 100 individu- als were measured from each sample. For presentation 102 Fishery Bulletin 101(1) Table 1 Physical characteristics of marsh and adj Figure 1 for locations of individual sites. icent bay study sites located along the New Jersey shore of Delaware Bay, 1996-99. See Surface temp. Average surface Surface salinity Average surface Average surface Marsh site range (°C) temp. CO range (%f ) salinity (%c) dissolved oxygen (mg/L) Upper hay Mill Creek 5.0-29.0 19.5 0-8.4 2.8 7.4 Mad Horse Creek 0-31.0 19.2 0.7-23.0 9.1 6.3 Browns Run 7.0-3L3 20.5 0.2-14.0 7.0 5.4 Bay 7.0-28.0 20.2 1.5-17.8 10.2 6.3 Lower bay Commercial Township 8.0-30.0 19.8 4.5-22.9 17.0 6.8 Upper Moores Beach 2.0-30.0 18.7 10.0-23.8 17.2 6.2 Lower Moores Beach 5.0-29.0 19.0 4.0-25.0 18.9 7.0 Dennis Township 6.0-32.0 20.4 6.2-24.7 17.0 5.7 Bay 6.0-29.1 19.6 11.1-24.7 17.8 7.3 Table 2 Yearly catch per unit of effort (CPUE= number offish per tow or weir set) for the different types of gear in the marsh creeks or in Delaware Bay and the total number of Atlantic croaker collected by each type of gear. RUMFS = Rutgers University Marine Field | Station; EEP = Public Service Enterprise Group Estuary Enhancement Program. Total number Total number 1996 CPUE 1997 CPUE 1998 CPUE 1999 CPUE of tows/sets offish Source Marsh creeks Otter trawl (creeks) 2.4 3.8 19.1 3.9 4,654 36,295 RUMFS Otter trawl (bay) 164.6 15.3 29.3 49.5 336 12.755 RUMFS Weir 1.1 41.6 46.8 98.7 443 20,714 RUMFS Delaware Bay Otter trawl 4.6 2.8 19.7 — 1,438 13,497 EEP 1-m plankton net 1.7 — — — 957 1,638 EEP Total fish 8,671 12,350 43,942 19,936 7,828 84,899 and statistical analysis of some aspects of the data of the ichthyoplankton and otter trawl surveys in the bay, the upper four zones were combined into an upper bay region and the lower four combined into a lower bay region (Fig. 1 ). Marsh creek survey Tidal creek samphng was carried out at six salt marsh sites on the New Jer.sey side of Delaware Bay (Fig. 1, Table 1). Dennis Township, Commercial Township, and Moores Beach will be referred to collectively as the lower bay sites, and Browns Run, Mill Creek, and Mad Horse Creek will be referred to as the upper bay sites. The average depth of the trawling stations ( ].',i-2.6 m) and Secchi depth values (0.3-0.4 ml were similar at all sites. The upper bay sites in the mostly oligohaline region of the bay had average salinities of 2.8-9. 1'/Jr and the lower bay sites were in the mesohaline region with average salinities of 17.0-18.9%p (Table 1). We sampled each of the marshes (Fig. 1) monthly from April through November 1996-99 (Table 2). Small inter- tidal marsh creeks were sampled with weirs set at high tide and hauled at low tide, approximately six hours later Each weir (2.0 m x 1.5 m x 1.5 m, with 5.0 m x 1.5 m wings, 6.0-mm stretched mesh) consisted of a funnel-shaped net stretched across the channel with wings extended back onto the marsh surface from each end of the net. In cases when the creek did not drain completely the area in ("runt of the weir was seined into the weir Trawling in larger intertidal to subtidal marsh creeks took place aroimd high tide and consisted of four replicate two-minute tows per station, made against the current with a 4.9-m otter trawl (6-mm stretched codend mesh) towed at a constant engine RFM of 2500. Trawling station locations at each site were designed to sample fishes along Miller et al : Distribution, abundance, and growth of Micropogonias undulatus 103 the mouth to upper creek gradients (see Able et al., 2000; 2001; Able et al.M. Thus, at each of the marshes there were six trawling locations. These locations included two large subtidal creeks and two smaller creeks with lower sub- tidal and upper subtidal or intertidal sections in each of the latter. Additional trawling locations were established in the bay immediately outside the mouth of the large creek at the Dennis Township, Moores Beach, Commer- cial Township, and Mad Horse Creek study sites (Fig. 1). The fish collected at these bay stations were used in the length-frequency figures for the bay (exclusive of Novem- ber when there was no trawl survey sampling in the bay) and for the growth calculations, but not in the catch-per- unit-of-effort (CPUE) calculations for the bay. Atlantic croaker collected in each weir set and in each trawl were enumerated, and up to 50 individuals per weir set and 20 per trawl were measured to the nearest millimeter total length. Abundances (CPUE, number of fish per trawl) were compared between the upper and lower bay sites in Delaware Bay, among the six different marsh sites, and among years, by using the nonparametric Mann-Whitney f/-test, or the Kruskal-Wallis ANOVA of ranks for mul- tiple comparisons, and when differences were found, the Dunnis test was used (criteria for significance; P<0.05) to make pair-wise comparisons. Physical variables were measured at the end of each weir and otter trawl sample in the marshes and in Dela- ware Bay (Table 1). Temperature, salinity, and dissolved oxygen concentrations were measured with a hand-held salinity, temperature, and oxygen meter (YSI Model 85), by lowering the probe into the water and recording surface values. Water transparency was measured by lowering a Secchi disc into the water column until it was no longer visible and recording the corresponding depth in 0.1-m increments. Growth J ■ 1996 Ichthyoplankton Survey rT September and October kV CPUE of Atlantic croaker ^•V larvae and postlarvae n= 1,635 ^:^ Upper Bay Sampling \ ^Y Zones 5 - 8 o *\ \ • '^ \^^ •°'S». • -^^ CPUE o 0 • 1-5 \ o ° «o<^o» 2 oo" o • 6- 15 • > 15 \«o « « / \°^° / o o ( ^ 1 "b ^-^ Lower Bay Sampling \ ^ Zones 1-4 \ _^ Figure 2 Catch per unit of effort (CPUE) of larval and postlar\'al Atlantic croaker (Micropogonias undulatus) collected in the ichthyoplank- ton survey in September and October 1996. Growth rates for YOY Atlantic croaker were calculated for samples collected during the late spring through fall in the upper and lower regions of the bay in 1997 and 1998 and in the upper and lower bay marsh sites during 1997. 1998, and 1999. We compared growth using the progres- sion of the monthly median lengths in each area by com- puting the change in the median length of a cohort over a time period divided by the number of days in the period. This method was based on the following assumptions: 1 ) no new (small) recruits join the population during the calculation interval, and 2) no (large) individuals leave the population over the calculation interval. To best meet these assumptions, median growth rates were calculated by using the monthly length data from May to July when there was a minimum of movement offish between differ- ent areas, and then for longer-term monthly comparisons, from May to August, September, and October when fish were moving out of the marshes into the bay. The smaller- Able, K. W., D . M. Nemerson, and T. M. Grothues. In review. Evaluating salt marsh restoration in Delaware Bay: continued analysis offish response at former salt hay farms. size cohort present in the bay in June and July 1998 and in the marshes in 1999 was excluded from the growth cal- culations for those years. The linearity of the progression of median lengths was tested by using linear regression, and the resulting lines were compared between the upper and lower bay Results Distribution, abundance, and size during fall ingress and settlement Atlantic croaker lan'ae were collected only in the late summer and fall during the ichthyoplankton survey in Delaware Bay in 1996 ( Figs. 2 and 3 ). A few individuals were first collected in August (/z=3, CPUE=0.02 fish/tow), and then large numbers of larvae were collected through September («=639, CPUE=3.6) and October (n=996, CPUE=9.0), but they were absent from April to July The overall September-October CPUE was 9.0 fish/tow (range; 104 Fishery Bulletin 101(1) 10 I 0 I 150 August n = 3 c: Upper bay ■ Lower bay ■ > f September n = 494 October n = 905 10 15 20 25 Total length (mm) 30 ^35 Figure 3 Length-frequency distributions of larval and postlarval Atlantic croaker (Micropogonias undulatus) collected in the upper and lower regions of Delaware Bay during the ichthyoplankton sur\'ey in 1996^ 0-56) in the upper bay zones 5-8 and 4.9 (range: 0-36) in the lower bay zones 1-4 (Figs. 1 and 2), and these CPUE values were significantly different between zones (P=0.03). At least one individual was collected in each of the eight zones in both September and October; the highest two- month combined CPUE occurred in the uppermost zone 8 (CPUE=13.4), followed by zone 5 (CPUE=9.6), and the lowest occurred in zone 3 (CPUE=0.1). Larvae were 4-10 mm during August (all in zone 2i, predominantly 2-24 mm in September, and 5-28 mm in October (Fig. 3) — the small- est individuals being caught in the lower bay. Benthic YOY Atlantic croaker of a variety of sizes first appeared in substantial numbers in September in the otter trawl surveys in both the bay (Fig. 4) and marshes (Fig. 5) at lengths >5 mm, and with modes of 15-30 mm for the primary cohort. Exceptions occurred in the bay in 1997, when they were not collected by the trawl survey until October and when they were not collected during September at two of the three upper bay marsh sites each year. The CPUE of benthic YOY Atlantic croaker was usually highest during October in the lower bay marshes (Fig. 6). This pattern of abundance is illustrated by the much high- er four-year overall CPUE of recently ingressed YOY, espe- cially at Dennis Township, Commercial Township, and Up- per Moores Beach (Fig. 7). The combined four-year CPUFI values were significantly different (/'<0.001 ) at each of the six sites, and the CPUE values at the Dennis Township site were significantly greater than at all the sites except for Commercial Township. Similarly, Commercial Town- ship was different from all sites except Dennis Township and Upper Moores Beach, and Upper Moores Beach also was different from Browns Run in the upper bay. Recently settled YOY Atlantic croaker were also caught in the weirs in small intertidal marsh creeks during Sep- tember, October, and November in all three years; the ma- jority were collected at the Dennis Township marsh in the lower bay (Fig. 8). The monthly CPUE (fish per set) in the weirs at Dennis Township was greatest in October 1997 and November 1999 (the weirs were not in place until October 1996) and the largest total number was collected during 1999. The combined four-year CPUE values for 1996-99 at each of the six sites were significantly different (P<0.001), and the CPUE values at the Dennis Township site were significantly greater than those at all the sites, except Commercial Township. The monthly CPUE values during ingress in the bay also were highest in October, but in contrast to the marsh sites were usually higher in the upper part of the bay (Fig. 6). The combined CPUE values for September and October were significantly different between the upper and lower bay regions in 1998 (P<0.001) and 1997 (P=0.048), but not in 1996 (P=0.51). The combined CPUE values for Septem- ber and October for each year ( 1996-98) were significantly different among years (P<0.001) and were different be- tween 1996 and 1997, and between 1996 and 1998. A second, smaller cohort of YOY Atlantic croaker ap- peared in the bay in -June and July 1998 and in the tidal Miller et al : Distribution, abundance, and growth q\ Micropogonias undulatus 105 1996 1997 1998 0 25 50 75 100 125 150 175 200 225 250 275 300 Total length (mm) Figure 4 Length-frequency distributions of log (« + l) transformed numbers of Atlantic croaker ^Micropogonias undulatus) collected by otter trawl between April and November in Delaware Bay from 1996 to 1998. None were collected in the bay from April to August of 1996. creeks in June 1999 (Fig 4). In the bay these were as small as 15 mm in June 1998 and had a mode of 26-30 mm. They were even more abundant in the bay during July 1998 and had a mode of 41-45 mm. Individuals of this cohort were collected at five of the eight zones in the bay during June and July but were rare in subsequent months (Fig. 4). A smaller size cohort of YOY (;!=69 fish) also appeared in the marshes (Fig. 5) and in the associated bay stations in June 1999. Distribution, abundance, and habitat use during summer residency Young-of-the-year Atlantic croaker were abundant in Del- aware Bay and in the adjacent marsh creeks from April through the fall egress of each year, except in 1996, when trawling in both the marshes and the bay caught no YOY until 26 individuals (115-200 mm) were collected in the bay in September and October (Fig. 4). In contrast, the 106 Fishery Bulletin 101(1) 1996 1997 1998 1999 0 25 50 75 100 125 150 175 200 225 250 275 300 Total length (mm) Figure 5 Lciiftth-frcquency distributions of log l/i + ll transformed numbers of Atlantic croaker iMkropoffonias undulatus) collected by otter trawl (April to November) in salt marsh creeks along the northern shore of Delaware Bay from 199(t to 1999. None were collected in the marshes from April to August of 1996. marsh creek surveys found YOY in both the larfje and small creeks from April to September during 1997, 1998, and 1999. Typically in the years after 1996, the CPUE was greatest from April to June at the lower bay marsh sites and then decreased after July to an almost total absence offish toward the end of the fall egress out of the marshes in November (Fig. 6). The overall CPUE at each marsh site for all three year classes combined (April to November in 1997, 1998 and 19991 was highest at Dennis Township and Commercial Township and at Upper Moores Beach in the lower bay and lowest at Lower Moores Beach and at the tipper hay sites iF'ig. 7). As a result, the monthly Miller et al : Distribution, abundance, and growth o\ MIcropogonlas undulatus 107 Upper bay zones Lower bay zones 35 30 25 20 15 ^ 10 5 0 1996-1997 Year Class 1996 n = 2.733 1997 n = 357 8 9 10 4 5 6 7 8 9 10 60 uj 40 Zl Q. " 20 \ 1997 Year - 1998 1998 ^'^^^ l. = 6,117 1997 T n = 578 i / J"}J 89 10 456789 10 1998-1999 / ^ Year Class 1998 n = 3.612 8 9 10 Month Upper bay marshes ' Lower bay marshes 1996- 1997 Year Class 8 9 10 11 4 5 6 7 8 9 10 11 1999-2000 Year Class 1999 i V n = 3.468 8 9 10 11 Month Figure 6 Monthly average catch per unit of effort (CPUE ±SEM) of young-of-the-year Atlantic croaker (Micropogonms undulatus) collected by otter trawl during the fall ingress and from April to November for three year classes in the upper and lower regions of Delaware Bay (left panels: CPUE=fisli/10 min. tow) and of four year classes in tidal creeks at the upper and lower bay marsh sites I right panels: CPUE=fish/2 min. tow). No data are presented from April to July of 1996 because Atlantic croaker were absent from the bav and marshes until October CPUE at the lower bay marshes was consistently more than twice as high as that in the upper bay (Fig. 6). The combined four-year CPUE values for YOY caught during April-October 1997-99, at each of the six sites were sig- nificantly different (P<0.001), and the CPUE values at the Dennis Township site were significantly greater than at each of the other sites. The CPUE values at Commercial Township and Upper Moores Beach in the lower bay also were greater than those at all the upper bay sites. Young-of-the-year Atlantic croaker also used small in- tertidal creeks in the marshes from April to August, where they were collected in weirs. They were most abun- dant at the Dennis Township site in the lower bay in all three years where they were present from May to July in 1997 and from April to August in 1998, 1999 (Fig. 8). The monthly CPUE (fish per set) in the weirs at Dennis Town- ship was greatest in June of both 1997 and 1998. Com- pared to the total catch in the weirs at Dennis Township in all three years (n=3994), far fewer were caught in the weirs at the other sites in the lower bay during all three years (« = 152) and fewer still at the sites in the upper bay (;)=9i. The CPUE of YOY Atlantic croaker in Delaware 108 Fishery Bulletin 101(1) 25 September - November 1996-1999 20 n = 36,295 15 ^ 10 1 - ■ CPUE O en l.ll ^. 50 April - November 40 1997- 1999 J n = 24,440 30 - 1 20 1 10 0 J l-.A DT LMB UMB CT BR MHC MC Lower Bay Sites Upper Bay Sites Figure 7 Combined overall average catch per unit of effort (CPUE= fish/2 min. tow |±SEM]) of young-of-the-year Atlantic croaker {Micropogonias undulatus) collected by otter trawl at the Dennis Township (DT), Lower Moores Beach (LMB), Upper Moores Beach (UMB), Commercial Town- ship (CT), Browns Run (BR), Mad Horse Creek (MHC) and Mill Creek (MC) marsh sites during the fall ingi'ess of four years 11996-99) and of post-ingress croaker (April to November 1997-99). Bay was higher in the upper bay, which has muii sedi- ments in most areas, and was much higher in 1998 than in 1997. The monthly CPUE in the upper bay peaked in July or August, but in the lower bay it peaked in October of both years (Fig. 6). The combined CPUE values in the upper and lower bay zones were different between the two regions in both 1997 and 1998 (P<0.001) and the catches within each region were different between the two years (P<0.001). During the summer most YOY were collected in areas of Delaware Bay that had muddy sediments (Fig. 9). In the upper bay zones 7 and 8, which likely have mostly pure mud sediments, YOY were collected at 829f of the stations. In contrast, they were absent in the deeper, large central area of the lower bay that has predominantly sandy and gravelly sediments. However, in the shallow portion of the lower bay, sandy mud, muddy sand, and gravelly mud sediments appear to be distributed on both sides of the bay, and YOY were almost exclusively collected over or near these substrates from April to August, Growth Although YOY Atlantic croaker showed rapid growth during the summer, there was no evidence of growth during the winter. The median growth rates for YOY 10000 ' 1997 1000 100 10 1 0,1 1997 ■■ I — I n = 803 All 1996-97Year class 1997-98 Year class n = 4,001 H -H 10000 I 1998 1000 1 100 1 n = 3,167 10 a. o 1 0 1 10000 - 1999 1000 \ 100 10 1 \l 1997-98Year class 1998-99 Year class n= 1,544 ^JJIJ 1998-1998Year class 1999-2000 Year class n = 24 n = 9,378 0.1 i_Ll APR MAY JUN JUL AUG SEP OCT NOV Montti Figure 8 Catch per unit of effort (CPUE, fish/set) of four year classes of young-of-the-year Atlantic croaker (Micropogonias undulatus) caught each year in the two monthly weir sets across small intertidal creeks at the Dennis Township marsh site in the lower bay during the fall ingress (white bars) and during spring and summer of the following year (black bars). Atlantic croaker calculated for two- to five-month periods beginning in May were fast and ranged from 0.5 to 1,.5 mm/d (Table 3), They were slightly higher in the bay than in the marshes (avg,=l,2 mm/d in the bay and 0,9 mm/d in the marshes) and were lowest in 1998 when Atlantic croaker were most abundant. The growth rates dropped off in the marshes when calculated from May to September or October (Table 3), The lowest early summer growth rates occurred in the tidal creeks in the lower bay in 1998 when the CPUE was the highest. The gi-owth rates in the upper and lower regions of Delaware Bay were similar in each year, but as in the lower bay marshes, the values were lower in 1998 when YOY were much more abundant. Linear regressions of the median lengths used to calculate these growth rates showed that median length was strongly cor- related to date (P=0.02-0.001) and illustrated the slightly slower growth rates at the lower bay sites in both 1997 and 1998 (Fig. 10). These pairs of regression lines were not significantly different (ANCOVA) for upper and lower bay regions of either the marsh sites in 1997 (P=0.1), or in the bay in 1998 (P=0.6), except in 1998, when a lower growth rate was indicated at the lower bay marsh sites (P=0.03). In 1999, a similar linear progi-ession of median lengths was observed in the lower bay marshes (r-=0.98), but sample sizes were too small in the upper bay for growth-rate cal- culations. Although there was no sampling in the winter, the length-frequency distributions indicated that most fish collected in April in the bay and marshes were the same Miller et al : Distribution, abundance, and growth o\ Miaopogonias undulatus 109 Delaware Bay Bottom Trawl Survey April-August 1997 and 1998 CPUE of Atlantic croaker n = 5,178 B 250 mm caught in October 1997 and 1998. The baywide trawling survey did not provide samples in November, so it was impossible to determine the timing of egress of the remaining YOY out of the bay, but very few age-1 fish were present in the bay or marshes by spring of the next year. Discussion Ingress and settlement Young-of-the-year Atlantic croaker ingress into bay and marsh nursery areas starting in the fall of each year in Delaware Bay and in other estuaries along the Atlantic coast. The majority appeared in September, October, and November during our study and in previous collections in Delaware Bay (Able and Fahay, 1998), Chesapeake Bay (Haven, 1957; Chao and Musick, 1977), and North Carolina (Ross, 1988). However, the fall ingress of this cohort was not evident until October in Georgia (Dahlberg, 1972) and December in South Carolina iBearden, 1964). The sudden appearance of significant numbers of larger fish (50-75 mm) in September 1996 in both the bay and marshes and to some extent in the marshes in 1999 sug- gests that individuals that experienced different growth rates or came from different spawning events sometimes occurred simultaneously in Delaware Bay. 110 Fishery Bulletin 101(1) Table 3 Estimated daily growth rates of young-of-the-year ^tlantic croaker (Micropogonias undulat us ) based on the monthly progression of median lengths in the upper and lower regions of Delaware Bay in 1997 and 1998 (see Fig. 1) and in tidal creeks in the marsh sites adjacent to the upper and lower bay in 1997, 1998. and 1999. Growth rate ca culations were made for periods of two to five months, 1 with each period starting in May. Calculations for locations with sample sizes <10 fish were excluded. Average Median length Median growth Habitat Year Location in bay collection date Sample size (mmTL) rate (mm/d) Delaware Bay 1997 Upper bay 17 May 25 33 — 7 July 71 103 1.37 5 Aug 88 143 1.38 2 Sep 16 183 1.39 2 Oct 10 190 1.14 Lower bay 26 May 52 43 — 16 Aug 64 154 1.35 20 Sep 14 187 1.23 4 Oct 96 218 1.34 1998 Upper Bay 13 May 275 41 — 15 Jul 873 121 1.27 10 Aug 433 144 1.16 6 Sep 45 177 1.17 Lower Bay 15 May 326 57 — 19 Jul 220 140 1.28 16 Aug 490 164 1.15 10 Sep 131 182 1.06 Marsh Creeks 1997 Upper Bay 21 May 17 46 — 17 Jul 13 115 1.21 20 Aug 16 154 1.19 14 Oct 33 120 0.51 Lower Bay 26 May 505 52 — 23 Jul 334 110 1.00 26 Aug 84 132 0.87 21 Sep 99 161 0.92 19 Oct 13 150 0.67 1998 Upper Bay 8 May 142 41 — 8 Jul 348 115 1.21 5 Aug 41 143 1.15 2 Sep 23 153 0.96 Lower Bay 13 May 467 50 — 14 Jul 851 100 0.81 11 Aug 174 125 0.83 9 Sep 126 142 0.77 1999 Lower Bay 18 May 148 49 — 18 Jul 128 114 1.07 15 Aug 12 151 1.15 According to the sizes of individuals captured by plank- ton net in the water column, versus those collected by otter trawl on the bottom, settlement may occur over a broad size range, i.e. approximately 10-40 mm TL. Scale formation in Atlantic croaker begins at 14-16 mm SL and is completed at 31-38 mm SL during this time (Bridges, 1971) and is an indicator of transformation between larval and juvenile stages. Alternatively, collection of overlap- ping sizes in water column and bottom samples may imply frequent vertical movements as could occur during tidal stream transport (see Weinstein et al., 1980, for recent ex- amples). These movements would provide an appropriate mechanism for small YOY to reach the bay and the lower Delaware River as has been suggested for larval Atlantic croaker in Chesapeake Bay (Norcross, 19911. The length-freiiuency data from our study and from pre- vious studies along the Atlantic coast indicate that a sec- ond, less-abundant cohort of YOY Atlantic croaker often Miller et a!.: Distribution, abundance, and growth o\ Micropogonias undulatus 111 enter nursery areas during the spring and sum- mer This second cohort (between 10 and 45 mm) was observed in June 1998 and 1999 during our study and in May or August (20-30 mm) in the York River of the Chesapeake Bay (Chao and Mu- sick, 1977). Similarly, a second mode was usually apparent from April through August during three years in North Carolina creeks and bays (Ross, 1988), and in May in Georgia (Dahlberg, 1972). In South Carolina, a second smaller cohort began appearing in March and subsequently became the dominant mode in June and July (Bearden, 1964). In addition, the larger-size individuals that have appeared during the fall months simultaneously with the ingressing fall cohort during our study and in the Chesapeake Bay (Haven, 1957; Chao and Musick, 1977), may be individuals of this late-arriving second cohort that did not enter the Chesapeake and Delaware bays until fall. These late arrivals to nursery areas in the Chesapeake and Delaware bays may be individu- als that were spawned close to or south of Cape Hatteras in late winter because there is no evi- dence of spawning in late winter or spring north of Cape Hatteras in the MAB. Atlantic croaker larvae were caught only from August to Janu- ary 1977-1987 over the continental shelf in the MAB and while entering estuaries in central New Jersey (Able and Fahay, 1998), or from November to February in coastal Virginia (Cowan and Bird- song, 1985). In contrast, just south of Cape Hat- teras, larvae as small as 5.2 mm SL were present in collections made from October through mid- April within and offshore of the Newport River estuary in North Carolina in both 1972-73 and 1973-74 (Lewis and Judy, 1983). Small larvae also were collected in the same estuary from November through mid-April in 1985-1986 (Warlen and Burke, 1990) and 1991-92 (Hettler et al., 1997) and in the Cape Fear es- tuary from mid-March to Mid-April in 1978 (Weinstein et al., 1980). Together, these studies indicate that late winter spawning occurs and suggests that it takes place south of Cape Hatteras. Analysis of otolith microstructure of lai^val and juvenile Atlantic croaker from the MAB indicates that later spawned larvae and juveniles have slower growth rates (Warlen, 1982; Nixon and Jones, 1997), which may account for the much smaller size of the later-arriving cohort when it enters the Chesapeake and Delaware bays during the late spring and early summer. Ross (1988) suggested that there may be two groups of Atlantic croaker that overlap and mix in North Carolina. The first group, occurring from North Carolina southward through the northern Gulf of Mexico, with a tendency to- ward high mortality, lower longevity, early maturation, re- sults from winter spawning (White and Chittenden, 1977; Barger, 1985) and mostly spring recruitment to estuaries. The second group ranges from North Carolina to about New Jersey and may exhibit lower mortality, higher longevity, greater size at age, late summer-fall spawning, mostly fall recruitment, and greater size at maturity (Wallace, 1940; • Upper bay sites ' ' Lower bay sites May Jun Jul Aug May Jun Jul Aug Date Figure 10 Linear regressions and goodness-of-fit measures of the monthly median total lengths of young-of-the-year Atlantic croaker (Micropo- gonias undulatus) caught at the upper bay (open circles) and lower bay (black circles) marsh sites and regions of Delaware Bay from May to August in 1997 and 1998 (see Table 3). The coefficient of determination is shown in the upper left for the upper bay regression lines and in the lower right for the lower bay. There is no regression for the lower bay because sample sizes in this region during June and July of 1997 were too small. Morse, 1980; Barbieri et al., 1994a). However, the group of larger, older Atlantic croaker observed by Ross (1988) apparently has been absent in Chesapeake Bay in recent years (Barbieri et al., 1994b). Lankford et al. (1999) did not find statistically significant genetic differences between fall- spawned YOY Atlantic croaker from north of Cape Hatteras and spring-spawned YOY from south of Cape Hatteras, but YOY from the Gulf of Mexico were genetically discrete from those from the Atlantic coast. This lack of marked genetic differences north and south of Cape Hatteras is not sur- prising if there is southward migration of adults from the MAB during winter as has been suggested (Haven, 1957). Although, spawning and recruitment to nursery areas does appear to occur later in the South Atlantic Bight (Bearden, 1964) and in the Gulf of Mexico (Pearson, 1929; Suttkus, 1955; Hansen, 1969), more research is needed to determine if there are significant biological differences between adults in these two areas and if the late arriving YOY in the north originate from spawning at or south of Cape Hatteras. Habitat use Young-of-the-year Atlantic croaker in this study used the entire range of marsh creek habitats, i.e. small intertidal 112 Fishery Bulletin 101(1) creeks and large subtidal creeks within the study area. Intensive tag and recapture studies in marsh creeks at the Dennis Township site found that YOY were resident for periods of up to 78 days from July through October 1998 (Miller and Able, 2002). As a result, our interpretations of habitat use and gi-owth may be representative for much of the summer and fall in Delaware Bay marsh creeks. In the deeper water of the bay, YOY were collected throughout the whole range of salinities but were most abundant over the predominantly pure mud sediments in the lower Delaware River and over areas with mud sedi- ments elsewhere in the lower bay. This pattern is evident elsewhere because YOY have been reported to be most abundant over soft mud sediments in Apalachicola Bay in the Gulf of Mexico (Kobylinski and Sheridan, 1979). As in Delaware Bay, YOY have been collected over the full range of salinities in South Carolina (Bearden, 1964; Miglarese et al.. 19821 and Georgia (Dahlberg, 1972). However, labo- ratory experiments suggest that lower salinities are meta- bolically less costly for YOY (Moser and Gerry, 1989; Pe- terson et al., 1999) and that in some areas of Chesapeake Bay, YOY are most abundant in regions with low salinities (<18'7„) (Haven, 1957). Habitat use and survival in the winter may vary be- tween estuaries. Young-of-the-year Atlantic croaker appear to overwinter in estuaries in the Gulf of Mexico (Pearson, 1929; Suttkus, 1955; Hansen, 1969; Knudsen and Herke, 1978) and in the South Atlantic Bight (Bearden, 1964; Dahlberg, 1972; Bozeman and Dean, 1980), but in the MAB there is probably significant overwinter mortality in years with particularly cold winters. The YOY appear to over- winter in some estuarine habitats in the York River region of Chesapeake Bay in most years (Haven, 1957; Chao and Musick, 1977) and in deeper areas of the bay (Welsh and Breder, 1923), but in some years YOY have been observed to experience winter mortality based on their subsequent disappearance after a cold period (Massman and Pacheco, 1960) and on direct observations of mass mortalities and collections of dead YOY in years with unusually cold winters (Joseph, 1972; Chao and Musick, 1977, Wojcik, 1978). Further, analysis of long-term recruitment indices for Atlantic croaker from 1979 to 1993 indicates that the YOY of this species may have experienced winter mortal- ity due to low water temperatures in 30'7f of the years in Chesapeake Bay and 74*7^ of the years in Delaware Bay I Lankford and fargett, 2001 ). Overwintering mortality apparently occurred in Dela- ware Bay in 1996 when water temperatures in the region dropped below 3°C and remained below 4°C for an extend- ed period of time. The NOAA Buoy 4409, located in the ocean just south of the mouth of Delaware Bay, recorded water temperatures at about 2-4°C for 18 days during January and February 1996, which is at or below the ap- proximate survival temperature of 3°C determined in lab- oratory experiments (Lankford and Targett, 20011. This apparently resulted in a total absence of YOY throughout the bay and in marsh creeks during the spring and sum- mer, which is not surprising because temperatures in the estuary were likely cooler than in the ocean. In contrast, during the winters preceding the relatively high catch years of 1997 and 1998, water temperatures at the same location never dropped below 4.4°C during the winter of 1996-97 or below 5.6°C during 1997-98. Growth Growth rates that we calculated in both upper and lower regions ofthe Delaware Bay (two years) and in the marshes (three years) ranged from about 0.8 to 1.4 mm/d from May to July. The strong linear correlation between median length and date suggested that the average growth rates were relatively constant during the summer from May to August before egress from marshes. Seasonal growth rates of YOY Atlantic croaker in other estuaries along the Atlan- tic coast may be similar to those in Delaware Bay, but the way in which they were calculated influences the values. Knudsen and Herke ( 1978) reviewed the apparent growth rates of YOY Atlantic croaker from a variety of sources but presented growth rates only for the entire first year of growth, which were all less than 0.5 mm/d for studies along the Atlantic coast and in the Gulf of Mexico. How- ever, these estimates included both larval and overwinter- ing periods; therefore they probably underestimated the growth rates during the summer when growth rates are highest. Monthly modal progression in published lengths indicate relatively fast growth rates during the summer in estuarine areas south of Delaware Bay Our calculation of modal progression in lengths from May to July in vari- ous parts ofthe York and Pamunkey rivers of Chesapeake Bay suggested growth rates of approximately 1.3 and 0.7 mm/d in 1952 and 1953, respectively (Haven, 1957) and of 0.9 mm/d in 1972 (Chao and Musick, 1977). Similarly cal- culated values for May to July for fish from shallow creeks in North Carolina indicated growth rates of 0.6, 0.8, and 0.9 mm/d in 1979, 1980, and 1981. but the 1979 estimate is likely to be an underestimate because many of the larger fish appeared to be moving into deeper habitats during that time period (Ross, 1988). In our study, the growth rates remained relatively high when calculated through October (1.1-1.3 mm/d) in the bay but dropped off in the marshes (0.5-0.7 mm/d), potentially reflecting the egress of larger YOY out ofthe marshes into the bay. Data from the Gulf of Mexico suggest slower growth rates of YOY Atlantic croaker in some areas but egress of larger fish out ofthe sampling area may also bias these estimates. Hansen (1969) used length-frequency data to determine growth rate estimates of 0.3 mm/d from January through August in the Pensacola Estuary on the Florida gulf coast in both 1964 and 1965 but noted the highest growth rates were in July (0.6 mm/d). Knudsen and Herke (1978) es- timated gi-owth of YOY in a semi-impounded marsh in Louisiana using recaptured individuals sprayed with fluorescent pigment during winter and spring and found rates of 0.4-0.5 mm/d for fish marked in late January and early February and recaptured into March. Rates for those marked mid-F'ebruary to late March and recaptured into May were 0.8-0.92 mm/d. A previous study at the same location, using the same techniques, estimated that fish marked from December to March and recaptured into June grew at about 0.47 mm/d (Arnold! et al., 1974 ). The monthly Miller et al.: Distribution, abundance, and growth oi Micropogonias undulatus 113 length-frequency data from Lake Pontchartrain, Louisiana (Suttkus, 1955), indicated a constant but slow growth rate of 0.3 rrtm/d from February to September 1954, and no increase in growth rate during the summer As a result of the above, it appears that growth rates may be faster, and thus countergradient in more northern populations, as suggested for Menidia (Conover and Present, 1990), but care should be taken in interpreting growth rates from the literature, especially those based on modal progression. Egress Young-of-the-year Atlantic croaker have a regular pattern of egress out of tidal creeks and estuaries in the MAB during the late summer and fall after reaching lengths of about 100-250 mm. As we observed in the Delaware Bay system, the majority left the marsh creeks from August to October at lengths <200 mm. The larger indi- viduals appeared to leave the marshes first, as has been observed elsewhere (Haven, 1957; Yakupzack et al., 1977), and almost all had left by November. However, the CPUE increased in Delaware Bay in October of both years, and this may have been caused by fish egressing out of the marshes into the bay. Large individuals remained in Delaware Bay longer than in the marshes and substantial numbers of fish 150-300 mm were present in the bay in September and October This finding suggests that egress from the tidal creeks caused the disappearance of Atlantic croaker there, and not gear avoidance, because large fish continued to be caught in the bay. The exact timing of egress of the majority of Atlantic croaker out of the bay is unclear due to lack of sampling throughout the bay after October However, previous collections in Delaware Bay have shown no evidence of any individual >100 mm from November to March (Able and Fahay, 1998), suggesting that egress out of the bay is finished by November in some years. The same pattern of egress out of nursery habitats in the fall has been observed in Chesapeake Bay (Haven, 1957), but in some years there were substantial numbers offish present into November (Chao and Musick, 1977). Very few of each year class reappear in collections during the spring and summer of the next year in either Chesa- peake or Delaware bays (Haven, 1957; Chao and Musick, 1977; Able and Fahay, 1998) and therefore the fate of these individuals is unknown. Fall egress also occurs out of estuaries in the South At- lantic Bight and the Gulf of Mexico, but in contrast to the Chesapeake and Delaware bays, more Atlantic croaker ap- pear to either remain through the winter or re-enter these habitats in some areas in late winter or early spring. In North Carolina, egress out of tidal creeks was mostly com- pleted by November, but this same year class was present again as age-1 fish in the bays in March, April, and May when sampling resumed (Ross, 1988). A similar pattern of egress from estuaries was observed in South Carolina, but the reappearance of age-1 fish in February was even more prominent and they continued to be collected until fall (Bearden, 1964). In the Gulf of Mexico, some age-1 fish have been observed to remain in estuarine habitats for an additional year in Lake Pontchartrain, Louisiana (Suttkus, 1955), or reappear from January to April after leaving the study area in December in coastal Texas (Pearson, 1929). In summary, this study presents the first compre- hensive examination of YOY Atlantic croaker seasonal- ity and habitat use in Delaware Bay and the adjacent marshes. Although patterns of habitat use and season- ality are similar along the east coast, some divergence from the seasonal patterns in Delaware Bay are evident in estuaries in the South Atlantic Bight and the Gulf of Mexico. Growth estimates appear to be the most di- vergent of any characteristics examined — faster growth rates occurring in the more northern estuaries such as Delaware Bay. Acknowledgments Numerous individuals from the Rutgers University Marine Field Station participated in the field sampling or helped with data analysis. We would particularly like to thank Ralph Bush, Bertrand Lemasson, Steven Teo, and James Chitty. John Balletto and Ken Strait provided background information and logistical support. Jonathan Sharp provided data on sediments in Delaware Bay. Financial support was provided by the Estuary Enhance- ment Program of Public Service Enterprise Group. Literature cited Able, K. W., and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight, 342 p. Rutgers University Press, New Brunswick. NJ. Able, K.W., D. Nemerson, R. Bush, and P. Light. 2001. Spatial variation in Delaware Bay (U.S.A.) marsh creek fish assemblages. Estuaries 24(31:441-452. Able, K.W., D. M. Nemerson, P. R. Light, and R. O. Bush. 2000. Initial response of fishes to marsh restoration at a former salt hay farm bordering Delaware Bay. In Concepts and controversies in tidal marsh ecology (M. P. Weinstein and D. A. Kreeger eds. ), p. 749-773. Kluwer Academic Publishers, The Netherlands. Arnoldi, D. C, W. H. Herke, and E. J. Clairain Jr. 1974. Estimate of growth rate and length of stay in a marsh nursery of juvenile Atlantic croaker, Micropogonias undu- latus (Linnaeus I, "sandblasted" with fluorescent pigments. Gulf Caribb. Fish. Inst. 26:158-172. Barbieri, L. R., M. E. Chittenden. Jr, and C. M. Jones. 1994a. Age, growth, and mortality of Atlantic croaker, Micro- pogonias undulatus, in the Chesapeake Bay region, with a discussion of apparent geographic changes in population dynamics. Fish. Bull. 92:1-12. Barbieri, L. R., M. E. Chittenden Jr., and S. K. Lowcrre-Barbieri. 1994b. Maturity, spawning, and ovarian cycle of Atlantic croaker, Micropogonias undulatus. in the Chesapeake Bay and adjacent coastal waters. Fish. Bull. 92:671-685. Barger, L. Y. 1985. Age and growth of Atlantic croakers in the northern Gulf of Mexico, based on otolith sections. Trans. Am. Fish. Soc. 114:847-850. Bearden, C. M. 1964. Distribution and abundance of Atlantic croaker. Mi- 114 Fishery Bulletin 101(1) cropogonias undiilatiix. in South Carolina. Contrib. Boars Bluff Lab., South Carolina 40:1-23. Bozoman, E. L.. Jr., and ,]. M. Dean. 1980. The abundance of estuarine lar\'al and juvenile fish in a South Carolina intertidal creek. Estuaries 3:89-97. Bridges, D.W. 1971. The pattern of scale development in juvenile Atlan- tic croaker (Micropogonias undulatus). Copeia 1971(2): 331-332. Chao. L. N., and J. A. Musick. 1977. Life history, feeding habits, and functional morphol- ogy of juvenile sciaenid fishes in the York River Estuary, Virginia. Fish. Bull. 75:657-702. Conover, D. O., and T. M. C. Present. 1990. Countergradient variation in growth rate: compensa- tion for length of the growing season among Atlantic silver- sides from different latitudes. Oceologia 83:316-324. Cowan, J. H., and R. S. Birdsong. 1985. Seasonal occurrence of larval and juvenile fishes in a Virginia Atlantic coast estuary with emphasis on drums (Family Sciaenidae). Estuaries 8:48-59. Cronin, L. E., J. C. Daiber. and E. M. Hulbert. 1962. Quantitative seasonal aspects of zooplankton in the Delaware River estuary. Chesapeake Sci. 3(2):63-93. Currin, B. M., J. P. Reed, and J. M. Miller 1984. Growth, production, food consumption, and mortality of juvenile spot and croaker: a comparison of tidal and non- tidal nursery areas. Estuaries 7:451-459. Dahlberg, M. D. 1972. An ecological study of Georgia coastal fishes. Fish. Bull. 70:323-353. Garvine, R. W., R. K. McCarthy, and K.-C. Wong. 1992. The axial salinity distribution in the Delaware Estu- ary and its weak response to river discharge. Estuar Coast. Shelf Sci. 35:157-165. Hansen, D. J. 1969. Food, growth, migration, reproduction, and abundance of pinfish, Lagodnn rliomhuidcs, and Atlantic croaker. Mi- cropogonias undulatus. near Pensacola, Florida, 1963-65. Fish. Bull. 68:135-146. Haven, D. S. 1957. Distribution, growth, and availability of juvenile croaker, Micropogonias undulatus. in Virginia. Ecology 38:88-97. 1959. Migration of the croaker, Micropogonias undulatus. Copeia 1959:25-30. Hettler, W. F. D. S. Peters, D. R. Colby, and E. H. Laban. 1997. Daily variability in abundance of larval fishes inside Beaufort inlet. Fish. Bull. 95:477-493. Joseph, E. B. 1972. The status of the sciaenid stocks of the middle Atlan- tic coast. Chesapeake Sci. 13:87-100. Knudsen, E. E., and W. H. Herkc 1978. Growth rate of marked juvenile Atlantic croaker.s, Mu-ropogonias undulatus, and length of stay in a coastal marsh nursery in southwest Louisiana. Trans. Am. Fish. Soc. 107:12-20. Kobylinski, G. J., and P. F. Sheridan. 1979. Distribution, abundance, feeding and long-term Huc- tuations of spot, Leiastomus xanthurus, and croaker, Micro- pogonias undulatus, in Apalachicola Bay, Florida, 1972- 1977. Contrib. Mar Sci. 22:149-161. Lankford, T E., Jr, and T. E. Targett. 2001. Low-temperature tolerance of age-O Atlantic croak ers: recruitment implications for U.S. Mid-Atlantic estu- aries. Trans. Am. Fish. Soc. 130:236-249. Lankford. T E. Jr.. T E. Targett, and P M. Gaffney 1999. Mitochondrial DNA analysis of population structure in the Atlantic croaker, Micropogonias undulatus, (Perci- formes: Sciaenidae). Fish. Bull. 97:884-890. Lewis, R. M., and M. H. Judy. 1983. The occurrence of spot, Leioslomus xanthurus, and At- lantic croaker, Micropogonias undulatus, larvae in On- slow Bay and Newport River Estuary, North Carolina. Fish. Bull. 81:405-412. Massman, W. H., and A. L. Pacheco. 1960. Disappearance of young Atlantic croakers from the York River, Virginia. Trans. Am. Fish. Soc. 89: 154-159. Miglarese, J. V, C. W. McMillan, and M. H. Sealy Jr 1982. Seasonal abundance of Atlantic croaker (Micropogo- nias undulatus) in relation to bottom salinity and tempera- ture in South Carolina estuaries. Estuaries 5:216-223. Miller, M. J. and K.W. Able. 2002. Movements and growth of tagged young-of-the-year Atlantic croaker, Micropogonias undulatus, in restored and reference marsh creeks in Delaware Bay. J. Exp. Mar. Biol. Ecol. 267:15-38. Morse, W. W. 1980. Maturity, spawning and fecundity of Atlantic croaker, Micropogonias undulatus, occurring north of Cape Hat- teras. North Carolina. Fish. Bull. 78:190-195. Moser, M. L., and L. R. Gerry. 1989. Differential effects of salinity changes on two estua- rine fishes, Le!ostomus.va;i//i!/r!;s and Micropogonias undu- latus. Estuaries 12:35-41. Nixon. S. W., and C. M. Jones. 1997. Age and gi-owth of lai-val and juvenile Atlantic croak- er, Micropogonias undulatus, from the middle Atlantic Bight and estuarine waters of Virginia. Fish. Bull. 95:773- 784. Norcross B. L. 1991. Estuarine recruitment mechanisms of larval Atlantic croakers. Trans. Am. Fish. Soc. 120:673-683. Norcross, B. L., and H. M. Austin. 1988. Middle Atlantic Bight meridional wind component effect on bottom water temperatures and spawning distri- bution of Atlantic croaker. Cont. Shelf Res. 8:69-88. Pearson, J. C. 1929. Natural history and consei-vation of the redfish and other commercial sciaenids on the Texas coast. Bull. L'. S. Bur Fish. 44:129-214 Peterson, M. S., B. H. Comyns, C. F Rakocinski, and G. L. Fulling. 1999. Does salinity affect somatic growth in early juvenile Atlantic croaker, Micropogonias undulatus (L.)? J. Exper. Mar Biol. Ecol. 238:199-207. Reiss. S. R.. and J. R. McConaugha. 1999. Cross-frontal transport and distribution of ich- thyoplankton associated with Chesapeake Bay plume dynamics. Cont. Shelf Res. 19:151-170. Ross, S. W. 1988. Age, gi'owth, and mortality of Atlantic croaker in North Carolina, with comments on population dynamics. Trans. Am. Fish. Soc. 117:461-473. Shenker, J. M.. and J. M. Dean. 1979. The utilization of an intertidal salt marsh creek by larval and juvenile fishes: abundance, diversity and tempo- ral variation Estuaries 2:154-163. Miller et al : Distribution, abundance, and growth of Micropogonias undu/atus 115 Suttkus, R. D. 1955. Seasonal movements and growth of the Atlantic croak- er [Micropogonias undulatus) along the east Louisiana coast. Gulf Caribb. Fish. Inst. 7:151-158. Wallace, D. H. 1940. Sexual development of the croaker, Micropogonias iinclulatus, and distribution of early stages in Chesapeake Bay. Trans. Am. Fish. Soe. 70:475-482. Warlen, S. M. 1982. Age and growth of larvae and spawning time of At- lantic croaker in North Carolina. Proc. Annu. Conf , S.E. Assoc. Fish Wild. Agencies 34:204-214. Warlen, S. M., and J. S. Burke. 1990. Immigration of larvae of fall/winter spawning marine fishes into a North Carolina estuary. Estuaries 13:453- 461. Weinstein, M. P. 1979. Shallow marsh habitats as primary nurseries for fishes and shellfish. Cape Fear River, North Carolina. Fish. Bull. 77:339-357. Weinstein, M. P., S. L. Weiss, R. G. Hodson, and L. R. Gerry. 1980. Retention of three taxa of postlarval fishes in an intensively flushed tidal estuary, Cape Fear, North Carolina. Fish. Bull. 78:419-434. Welsh, W. W., and C. M. Breder 1923. Contributions to the life histories of Sciaenidae of the eastern United States coast. Bull. U.S. Bur. Fish. 39:141- 201. White, M. L., and M. E. Chittenden Jr 1977. Age determination, reproduction, and population dy- namics of the Atlantic croaker, Micropogonias undulatus. Fish. Bull. 75:109-123. Wojcik, F. J. 1978. Temperature-induced croaker mortality. Coast. Oce- anogi'. Climatol. News 1:5. Yakupzack, P M., W. H. Herke, and W. G. Perry 1977. Emigration of juvenile Atlantic croaker, Micropogo- nias undulatus, from a semi-impounded marsh in south- western Louisiana. Trans. Am. Fish. Soc. 106:538-544. 116 Abstract — Goldband snapper {Pristi- pomoides multidens) coWccted (Tom com- mercial trap and line fishermen off the Kimberley coast of northwestern Aust- raHa were aged by examination of sec- tioned otoliths ( sagittae I. A total of 3833 P. multidens, 80-701 mm fork length (98-805 mm total length), were exam- ined from commercial catches from 1995 to 1999. The oldest fish was estimated to be age 30+ years. Validation of age estimates was achieved with marginal increment analysis. The opaque and translucent zones were each formed once per year and are considered va- lid annual growth increments (the translucent zone was formed once per year between January and May). A strong link between water tempera- ture and translucent zone formation was evident in P. multidens. The von Bertalanffy growth function was used to describe growth from length-at-age data derived from sectioned otoliths. No significant differences in length- at-age were found between sexes and growth parameters were L^, = 598 mm, K = 0.187/yr, t„ = -0.173 (r2=0.76). Re- gression models of estimated age as a function of otolith and fish measure- ments indicated a significant relation- ship between estimated age and otolith weight (r^=0.94). Total instantaneous mortality (Z) estimates generated from catch-at-age data off! multidens from the northern demersal scalefish fishery INDSF) were 0.65 for 1995-96, 0.87 for 1996-97, and 0.71 for 1997-98. Esti- mates of the annual instantaneous rate of natural mortality (M) were 0.10-0. 14. The NDSF population of P. multidens is considered to be exploited above opti- mum levels on the basis of these mortal- ity estimates. The protracted longevity, moderately slow growth and low natural mortality rates of P. multidens predis- poses this species as one vulnerable to overfishing, thus cautious management strategies will be required. Furthermore, capture of P. multidens from depths of 60 meters or greater results in a high mor- tality of fish because the physoclistous ruptures causing internal hemorrhaging and hence there is a low probability of survival of any fish returned to the sea. Thus traditional harvest strategies in- volving size limits will bo inappropriate for these fish. Conversely, hai"vest strate- gies that include appropriately targeted spatial fishery closures may provide a useful additional means of preserving the spawning stock biomass of these fish and protect against recruitment over- fishing. Manuscript accepted 21 August 2002. F'ish. Bull: 116-128 12003). Age validation, growth, mortality, and additional population parameters of the goldband snapper iPristipomoides multidens) off the Kimberley coast of northwestern Australia Stephen J. Newman lain J. Dunk Western Australian Marine Researcti Laboratories Department ol Fishenes Government of Western Australia P.O Box 20 North Beach, Western Australia 6920, Australia E-mail address (for S J Newman) snewmanm'fish wagovau The goldband snapper iPristipomoides multidens. Day), known also as gold- banded jobfish. Day's jobfish and large- scaled jobfish, is widely distributed throughout the tropical Indo-Pacific Ocean region from Samoa in the Central Pacific to the Red Sea in the Western Indian Ocean and from southern Japan south to Australia (Allen, 1985). Along Western Australia, P multidens is found as far south as Cape Pasley (34°S) and is landed in commercial quantities from the Ningaloo Reef area (23°30'S) north- wards (Kailola et al., 1993; Newman, unpubl. data). They inhabit hard bot- tom areas and areas of vertical relief and large epibenthos from depths of 60 to at least 245 m and are concentrated in depths from 80 to 150 m (Allen, 1985; Newman and Williams, 1996). Pristipomoides multidens is a com- mercially important species throughout much of its range, forming an important part of the landed catch in both arti- sanal and developed fisheries (Dalzell and Preston, 1992; Newman, 2001). In Western Australia this highly valued resource is marketed whole, usually fresh on ice, and transported by road from regional ports to markets in most state capital cities. It is occasionally ex- ported. In the Kimberley region, within the northern demersal scalefish fishery (NDSF), P. multidens has composed on average 37.7% of the landed catch from 1995 to 1999 (contributing on average 255 metric tons (t)/year). In terms of val- ue to fishermen, it is second only to the red emperor snapper (Lutjanus sehcw). Information on the biology of P. mul- tidens is limited. The juvenile habitats of P. multidens have not been identi- fied, although Newman (unpubl. data) obtained juveniles from uniform sedi- mentary habitat with no relief. In pre- vious studies, several age determina- tion techniques were used to determine the age of P multidens but there were limited attempts at age validation (Ed- wards, 1985; Mohsin and Ambak, 1996; Richards'). The accurate determina- tion of fish age is the key to estimat- ing growth rates and mortality. Errors in determining fish age can result in ambiguous demographic parameters and provide misleading impressions of the production potential of fish stocks (Newman et al., 2000a). There is a lack of reliable information on the longevity, growth parameters, mortality rates, and population characteristics of P. multidens, despite its ecological and commercial importance. This work represents the first com- prehensive study of age, growth, and mortality of a population of P. mul- tidens based on age estimates from sectioned otoliths and contributes to the management of these stocks. The objectives of this study were to validate aging and to provide age, growth, mor- tality and population characteristics of P. multidens from the Kimberley region of Western Australia that are based on age estimates from sectioned otoliths. • Richards, A. H. 1987. Aspects of the bi- ology of some deep water bottomfish in Pa- pua New (Guinea with special reference to Pristipomoides multidens (Day). Report 87-01, 31 p. Fisheries Research and Sur- veys Branch, Department of Primary In- dustry, Port Moresby, Papua New Guinea. Newman and Dunk: Age validation, growth, mortality, and additional population parameters of Pristipomoides multidens 117 122* 123' 124- 125' ly 128" MWW ^?Derby i^l\ { ) \ ,->" •proome f' T A Western Australia A Figure 1 Location of the northern demersal scalefish fishery (NDSF) off the Kimberley coast of northwestern Australia showing the 50-m, 100-m, and '200-m depth contours. The NDSF is bounded in the west by the 120''E latitude line, to the north by the boundary of the Australian fishing zone (AFZ), and to the east by the border with the Northern Territory. Fishing primarily occurs in depths of 80-140 m. A further objective was to investigate the relationship between estimated age and the measurements of both oto- lith and fish dimensions to assess the applicability of these measurements in predicting the age of this species. Materials and methods Commercial landings of P. multidens from the NDSF off the Kimberley coast of Western Australia were sampled from 1995 to 1999. Samples were acquired opportunisti- cally from July 1995 to December 1996, whereas samples obtained from January 1997 to December 1999 were col- lected on a monthly basis among all vessels in the fleet. All specimens were captured with fish traps at depths of 60 to 200 m from 12°-20°S latitude (Fig. 1). Additional specimens were attained from research vessel cruises with fish traps. All fish were measured to the nearest mm total length (TL), fork length (FL) and standard length (SL), weighed to the nearest g total weight (TW) and cleaned weight (CW), and where possible, sex was determined by examination of the gonads. Cleaned weight is defined as the TW after re- moval of the gills and viscera. Length measurements were used to derive conversion equations with linear regression models [TL=a + h(FL), FL=a + b tTL). FL=a + 6 (SL) and SL=a + b (FL)\. Length-weight models The relationships between FL and both TW and CW were described by the power function where W = weight (TW or CW, g); and L = FL(mm). These relationships were fitted to log-transformed data and the parameters were back-transformed (with correc- tion for bias) to the above form. 118 Fishery Bulletin 101(1) Analysis of covariance («=0.05) was used to determine if there were significant differences in the weights-at-length (FL) relationships between sexes for P multidens. Length and weight data were transformed to natural logarithms to satisfy assumptions of normality and homogeneity. Multiple comparisons were performed with Tukey's hon- estly significant difference (HSD) test. Trends in mean length and weight offish over time were assessed by using analysis of variance (a=0.05). Otolith preparation and analysis Otolith removal, measurement, and preparation followed the procedures and protocols described in Newman et al. (1996), Newman et al. (2000b), and Newman and Dunk (2002). All age estimates were based on the analysis of thin transverse sections of otoliths. These thin sections were examined under a dissecting microscope at 10-30x magnification with reflected light on a black background. The otoliths from eight juvenile P. multidens (80-140 mm FL) were examined for daily bands with a different technique. One sagitta per fish was embedded in epoxy resin and a thick transverse section (>500 pm) was cut. The section was then ground and polished from each side to a level near the core (perpendicular to the long axis of the otolith) by hand with ebony paper (1000 grade) and lapping film (9 and 3 pm). A polished thin transverse sec- tion approximately 100 pm thick was produced. The sec- tion was then examined with a compound microscope. Age validation Marginal increment analysis, routinely used to validate fish age, relies on the assumption that if a translucent zone is laid down once a year, there should be a clear pattern of peri- odic growth on the edge of the otolith during the year. Mar- ginal increment analysis is appropriate only if all fish in the population lay down the translucent zone at the same time. Thus, an annulus consists of a single opaque and a single translucent cycle within a r2-month period. The opaque zone is believed to form during periods of slow growth. Marginal increment analysis usually implies measure- ment of marginal growth and hence many researchers have measured the width of the edge of the otolith sec- tion over an annual cycle. This measurement approach has an advantage in that it should be possible to plot growth of the edge over time to validate that only a single translucent mark is laid down each year. However, in P. multidens, it can be difficult to determine a consistent location to measure on the otolith because of the inherent variability of their otoliths; hence this technique was not used in the present study Edge type analysis was adopted for the marginal incre- ment analysis of P. multidens and edge types were clas- sified according to Pearson (1996) as either translucent, narrow opaque (opaque area less than half of the previous opaque zone), or wide opaque (opaque area greater than half of the previous opaque zone). Sectioned otoliths offish of all ages were examined under a dissecting microscope with reflected light on a black background. Age determination Because the peak spawning period off! multidens occurs in late March, all fish were assigned a birth date of 1 April to assure proper year-class identification. Ages were assigned from counts of annual growth increments con- sisting of alternating opaque and translucent rings from sectioned otoliths (opaque rings were counted). Annual growth increments were counted in the ventral lobe of the otolith from the primordium to the proximal surface, as close as was practicable to the ventral margin of the sulcus acousticus. Annual growth increments were counted with- out reference to fish length or date of capture. Each otolith section was examined on four separate occasions. When the counts differed, otolith sections were re-examined. In most cases that required resolution, the fourth and final count was used for analysis of age and growth because by this time considerable experience had been gained in the interpretation of the otolith structure. Otoliths with structural irregularities (such as unusual calcification, deterioration of the ventral lobe, or poorly defined annual growth increments) were considered indecipherable and were excluded from analysis offish age. Counts were compared and the precision of age esti- mates were calculated with the average percent error (APE) of Beamish and Fournier (1981). Greater precision is achieved as the APE is minimized. The relationship be- tween fish length (FL) and age and otolith dimensions was assessed with linear regression techniques. Timing of translucent zone formation in P. multidens and mean sea surface temperatures (SST was assumed to reflect the temperature at depth) were compared by scal- ing values from the two data sets. The scaling process al- lowed direct comparison of each series and any time lags of one in relation to the other. Using the scaling score = 1 - ((maximum data value - data value) -^ range), where (in the month of November) mean SST for the month was 29°C, the maximum for the year (data set) was 29.7 and the range of the data values was 3.7, we calculated the scaled SST = 1 - ((29.7-29) -=■ 3.7) = 0.81; in addition the scaled '7c frequency of otoliths with translucent edge types = 1 - ((67-20) -^ 67) = 0.30. Growth and mortality models The von Bertalanffy growth function (VBGF) was fitted to estimates of length-at-age with nonlinear least squares esti- mation procedures. The VBGF is defined by the equation L, = L„[l- exp[-K{t - 1^)]], where Lf = mean length offish of age t; L , = asymptotic mean length; /f = is a rate constant that determines the rate at which Lf approaches Lj, t = age of the fish; and ^11 = the hypothetical age at which the mean length is zero if it had always grown in a manner described by the VBGF. Newman and Dunk: Age validation, growth, mortality, and additional population parameters of Piistipomoides multidens 119 Table 1 Length-weight relationships for P. multidens off the Kimberley coast of northwestern Australia parameters a and b of the relationship W = aL'', the sample size (n ), and the regression r- value in mm and the weight is total weight |TW] or cleaned weight ICW] in g). . Estimates were lengths used are obtained for the fork length [FL| Group a h n r2 P. multidens (all fish-TW) 2.483 X lO"^ 2.9501 3680 0.983 P. multidens (all fish— CW) 2.356 X lO"^ 2.9425 3073 0.983 P. multidens (male— TW) 2.156 X 10^^ 2.9737 1963 0.985 P multidens (female— TW) 2.825 X 10^^ 2.9281 1671 0.987 The von Bertalanffy growth curves for both sexes were compared with the likelihood ratio test of Cerrato (1990). Estimates of the instantaneous rate of total mortal- ity (Z) were obtained from catch-at-age data from the NDSF. Annual catch in weight was converted to annual catch in numbers-at-age by the use of age-frequency data standardized by fishing effort to obtain catch-per-age class. Catch in weight was converted to catch in numbers based on the mean weight of P. multidens observed in the sampled catch each year. Mortality estimates were then derived between successive years by obtaining the natural logarithm of the catch per age class (e.g. age 7) in year t and subtracting the natural logarithm of the catch per age class (e.g. age 8) in year ^ + 1 for all fully recruited age classes. Mean total Z was then calculated across all fully recruited age classes. Instantaneous natural mortal- ity rates (M) were derived by using the general regression equation of Hoenig (1983) for fish, where logg Z = 1.46 - 1.01 logg t,fi(^jf- itf^nx=^^^ maximum age in years). The Hoenig equation has been shown to provide a reasonable approximation of M in tropical demersal fishes (Hart and Russ, 1996; Newman et al., 1996; 2000b). The annual percentage removal was estimated by annual percentage = [F/Z (l-e-^)! x 100%. Exploitation rates (E) were derived from the estimates of Z and F as defined by the equation E = F/Z iF=the instantaneous rate of fishing mortality derived from the relationship F=Z-M). Reference points for target (optimal) and limit fishing mortality rates (F , and F,,„„,) were calculated for P. multidens by using the estimate of natural mortality (M), because F,^, = 0.5 M (Walters, in press) and F,,„,„ = 2/3 M (Patterson, 1992). Results A total of 3833 P. multidens (ranging in size from 80 to 701 mm FL [10.6-5770 gTW] ) were examined for age analysis. Of the fish collected, 2063 were males ranging from 245 to 671 mm FL and from 296 to 5195 g TW, and 1751 were females ranging from 284 to 701 mm FL and from 450 to 5770 g TW. Length conversion equations were derived for total length: TL = (1.12xFL) -t- 21.84 (n=2137, 7--^=0.995); fork length: FL = (0.89xrL) - 16.61 (/!=2137, r2=0.995); FL = (1.12xSL) -I- 6.44 (;?=2148, /-=0.992); and standard length: SL = (0.89xFL) - 2.14 (n=2148, r-=0.992). Length-weight models Length-weight relationships were calculated separately for males, females, and for both sexes combined (Table 1). The relationship between TW and FL is presented in Figure 2. ANCOVA of TW-at-FL and CW-at-FL were both significantly different between sexes (TW: F=42.56; df: 1, 3234; P<0.001; CW: F=94.29; df: 1, 2652; P<0.001); males were larger than females. The length-frequency distribu- tion for male and female P. multidens is shown in Figure 3. Temporal trends were evident in the mean length and weight of P. multidens over time. Mean FL was signifi- cantly different among years from 1995 to 1999 (ANOVA; F=31.29; df: 1, 4193, P<0.001), with (1995=1996=1997) > (1998=1999). Mean TW was also significantly different among years from 1995 to 1999 (ANOVA: F=89.33; df: 1, 3295, P<0.001), with 1995 > 1996 > 1997 > (1998=1999). Age validation Otoliths displayed alternating opaque and translucent zones. A consistent annual trend was evident; the trans- lucent zone was laid down from January to May and the opaque zone formed from June to December. The trend in thin opaque zone formation in June and July was replicated in both 1997 and 1998. Figure 4 clearly demonstrates that the opaque and translucent zones are laid down once a year and represent valid annual growth increments. Because the marginal increment analysis involved random sampling across all age classes in the sampled population, the validation of annual growth increments can be expected to hold across all age classes. In addition, the formation of the translucent zone in the sagittal otoliths of P. multidens and the annual cycle of sea surface temperatures in the Kimberley region of northwestern Australia were found to be closely related (Fig 5). Otolith structure, analysis, and functionality The sagittae of P multidens are somewhat laterally com- pressed, elliptical structures. The distal surface is concave and the rostrum and postrostrum are somewhat pointed. The sagittae are characterized by variable growth reticu- lations along the dorsal edge from the postrostrum to 120 Fishery Bulletin 101(1) 6()(K) . / / = 0 4.S y ^ 5000 ^X - 4000 0) g M' S 3000 o 1- j^F 2000 ^^F 1000 ^^^° ° Male JV^^^^ * Female 0 100 200 300 400 500 600 700 Fork length (mm) Figure 2 Relationship between fork length and total weight for P. multidens off the Kimberlev coast of northwestern Austraha. SO (I 40 SO - 120 - Male ;i = 22SI -B»- 220 2WI 300 340 3S0 420 4(iO 500 540 5SII h20 660 7110 Length class (10mm) Figure 3 Length-frequency distribution (10-mm longth classes) of male and female P. inulti dens sampled for age determination. the antiro.struni and along the ventral edge from the postrostrum to the rostrum. A curved sulcus crosses the proximal surface longitudinally, and the depth of the sulcal groove increase.^ with (ish age. The precision of otolith readings of P. multidens was relatively high lAl^E of lOA'/i). Given the variability en- countered among otoliths, this APE reflects a moderately high level of precision among otolith readings and indi- cates that the aging protocol adopted is replicable. Otolith length and breadth were useful predictors offish length in P. mullulcns. accounting for more than 17' i of the variability (Table 2). In contrast, otolith weight and, in par- ticular, height were poor predictors offish length (Table 2). Otolith weight was the best predictor offish age for f! multi- Newman and Dunk: Age validation, growth, mortality, and additional population parameters of Pristipomoides multidens 121 yo _ 80 -•- Wide Opaque. -O- Translueent, —tr- Thin Opaque % Frequency o o = o ' '/^^ /'^'°^"n '•-. / 30 :n 10 0 J J A .S O N D J F M A M J ,1 A IWonths (1997-1998) Figure 4 Marginal increment analysis of P multidens sagittal otoliths using an edge type class- ification system (edge types: wide opaque, thin opaque, and translucent). 1.(1 -A- Scaled SST 2^'°/'°^'*^ •-Q- .Sealed Translucent a^^/ ■ \". 0.8 1 V units / * V Scaling 4- 1 p \ °\ / ^ °\ .O 0.2 ■•V / •■■° ^ 0.0 b--^' \ J J A S O N D J I- M A M J J A Months (1997-1998) Figure 5 Comparison of the translucent zone formation on the sagittal otoliths of P miiltidoia with sea surface temperatures in the Kimberley region of northwestern Australia. Values are standardized to take into account any time lags and to allow comparison | of each time series. dens, accounting for Q'iA'yc of the variability in age (Table 2, Fig. 6). Otolith height was also a useful predictor of fish age, accounting for 88'X of the variability in age. In con- trast, otolith length and breadth were poor predictors of age for P. multidens (Table 2). Growth and mortality models The von Bertalanffy growth curve was fitted to FL-at-age for all P. multidens (Fig. 7), and separately for each sex (Table 3>. Growth in FL of P. multidens is relatively fast to 122 Fishery Bulletin 101(1) M) o 27 o Male g „ i Female o o Unknown o8 ^ ^ 24 u 21 S IS 01 1 s * '^^^^^m^ < '"IrJS^^^^A 12 ^.^^^P^^ 9 oa»^^^^ 6 ^^^^^^^^ 3 #■ 0° ■ 0,0 0,1 0,2 0,3 (1.4 0.5 0.6 0.7 O.S (1.4 1.0 1.1 1.2 Otolith weight (g) Figure 6 Relationsh ip between otolith weight and age of P. mtiltidens estimated from sectioned otoliths. Table 2 Comparisons among regression form v = a ses, fish length (FLl a (SEl of the estimate otolith dimensions and length and age of P multiden^. The predictive equations are of the simple linear + bx (OW=otolith weight; OL=otolith length; 0B= otolith breadth; OH=otolith height). For regression analy- nd age were used as the dependent variables (all regressions were significant at P<0.001).The standard error s a measure of the dispersion of the observed values about the regression line. Dependent variable Independent variable Sample size Equation r2 SEof estimate FL OW 2590 FL = (315.37 X OW) + 339.46 0.68 37.57 FL OL 2493 FL = (33.47 X OL) - 111.43 0.77 32.08 FL OB 3745 FL = (50.43x 05) -124.36 0.83 28.48 FL OH 3988 FL = (118.39x0//)+ 175.14 0.53 48.36 Age OW 2408 Age = (21.86 xOWl- 0.68 0.94 0.94 Age OL 2305 A?e = (1.77 xOD- 21.68 0.58 2.63 Age OB 3469 A?e = (2.40 X OB) -19.14 0.60 2.49 Age OH 3652 A^c = (8.52 xO//)- 12.81 0.88 1.36 age 9 but is much reduceci in age cohorts beyond 9 years of age. Parameters of the VBGF are li.sted in Table 3. FL-at- age of P. multich'ns was not significantly different between sexes (log-likelihood=0.9836, test statistic=1.001, P>0.05; no significant differences were found among parameters of the VBGF; see also Fig. 7). Generalized VBGFs of P. multi- dens from previous studies were compared to that derived from our study (Fig. 8). The maximum observed age of P. multidcns in the Kim- berley region was 30 years. Given that the P. multidcns resource in the Kimberley region has been exploited lor over 20 years, it is possible that in an unfished population the longevity of P. multidcns may be closer to 40 years. These two estimates of maximum age in P. multidens were applied to the Hoenig (19831 equation in order to derive an estimate of M. Consequently, M was considered to be in the range of 0. 104-0. 139, representing an annual survi- vorship of 87-90'^f for an unfished population. This range of M estimates for P. multidcns is similar to that observed for other long-lived lutjanid species in the Indo-Pacific region (Newman et al., 1996, Newman et al., 2000a; New- man and Dunk, 2002). The longevity of female and male P. multidens was somewhat similar at 27 and 30 years, respectively. The age structures of P. multidens in the commercial catch differed among years. The 1995 sample had a peak in year Newman and Dunk: Age validation, growth, mortality, and additional population parameters of Pristlpomoides multidens 123 700 bOO . .* ^ oo°". ^^^^^^^Pfe'cp.- f*°^' ° SOO E 400 ^^^pp- ^°-- © 300 ^^ o "" 200 / 1 o Male 100 f '■ '■=""" 0 2 4 6 S 10 12 14 16 IS 20 22 24 26 2X 30 32 Age (years) Figure 7 Length-at-age and von Bertalanffy growth curve for P. multidens off the Kimberley coast of northwestern Australia. class 5 and relatively strong age classes 6, 8, and 10, but abundance per age class declined rapidly to age 20, after which few fish were found to be older (Fig. 9). The 1996 and 1997 samples were somewhat similar. In 1996 rela- tively strong year classes were present from age 5 through to age 11, and abundance per age class declined rapidly to age 26 (Fig. 9). One year later, the 1997 sample had relatively strong year classes present from age 6 through to age 12 (Fig. 9), providing further evidence of the annual formation of growth increments. The 1998 sample had relatively strong 6, 7, and 8 age classes, and abundance per age class declined rapidly to age 24 (Fig. 9). The 1999 sample was similar to the 1998 sample with relatively strong 6, 7, and 8 age classes, and abundance per age class declined rapidly to age 20 ( Fig. 9 ). Age classes 9 through 12 were somewhat eroded in the 1998 and 1999 samples in comparison to the 1996 and 1997 samples. In all years, abundance per age class declined rapidly to age 20, and fish older than 20 years were not well represented in the catch over the five years of catch sampling. In most years there was a strong mode of age-6 individuals present and this mode may reflect the age at full recruitment to the sampling gear (fish traps). Pristipomoides multidens less than age 6 were in gen- eral not fully recruited to the sampled population and were therefore excluded from the mortality estimates derived from catch-at-age data. The year-specific total an- nual rate of mortality, Z, of P. multidens in the NDSF, was 0.65 for 1995-96 (fish aged 6-21 years), 0.87 for 1996-97 (fish aged 6-21 years), and 0.71 for 1997-98 (fish aged 6-21 years), representing an annual percentage removal of approximately 38%, 49%, and 41%, respectively, for each Table 3 Growth parameters derived from the von Bertalanffy growth function and population characteristics of P. mul- tidens off the Kimberley coast of northwestern Australia (r!=sample size , FL is in mm. and age (t) is in years). Parameters Male Female Total n 1879 1600 3479 i„ 594.49 603.23 598.08 K 0.1868 0.1867 0.1873 'o -0.3601 0.0018 -0.1730 r^ 0.7394 0.7875 0.7630 n 2281 1916 4573 '' ^mean 501.5 493.5 495.1 '' ^m in 245 284 80 '' '^max 671 701 701 n 1872 1597 3833 hnean 10.24 9.54 9.73 hnin 3 4 0.35 ^max 30 27 30 year (Table 4). In addition, exploitation rates were 0.79, 0.84, and 0.80, respectively. The optimum fishing mortality rate, F^^,, for P. multidens was estimated to be 0.052-0.069, and the limit reference point, F,,„,„, was estimated to be 0.069-0.092 (see Table 4). These results indicate that only approximately 6% of the 124 Fishery Bulletin 101(1) 675 600 .525 E E. 450 c 0) .^75 .WO 225 150 75 Briiuard el al. 1 14X4) growth curve Richards ( 1987) growth curve Edwards ( 1985) growth curve This study 10 12 14 16 18 20 Age (years) 24 26 28 30 .^2 Figure 8 Comparison of generalized von BertalanfTy growth cun'es for P. multidens from pre- vious studies with that derived from this studv. Table 4 Summary of total mortality (2) estimates for P. multidens derived from catch-at-age data based on ages determined from sectioned otoliths. Estimates of fishing mortality If) are derived by subtraction because Z = F + M and are compared to estimates of optimum fishing mortality rates. Year Z P F,. F,.,..., 1995-96 0.649 1996-97 0.869 1997-98 0.710 0.510-0.545 0.052-0.069 0.069-0.092 0.730-0.765 0.052-0.069 0.069-0.092 0.571-0.606 0.052-0.069 0.069-0.092 available stock of P. multidens can be harvested on an an- nual basis in a sustainable manner and that annual har- vest rates should not e.xceed 107( of the average stock size. Discussion Sagittal otoliths were determined to be valid structures for age determination in P. multidens. The edge-type clas- sification system of three edge types used in this study is capable of indicating whether the opaque zone has just been formed or whether a new translucent zone is ready to form. The use of marginal increment analysis (MIA) of individuals of all ages exhibits a clear trend and demon- strates conclusively that annual growth increments are formed once per year Annual growth increments were most conspicuous in the ventral lobe of the sagittal oto- liths. However, we observed that experience is a critical iactor in increasing the agreement and hence [precision of successive counts of annual growth increments in P. multidens. The spring-summer peak in opaque zone formation ob- served in our study is in accordance with the peak in opaque zone formation identified by Fowler (1995) and Beckman and Wilson ( 1995) for tropical fishes. The translucent zone (the period of fast growth in the otoliths) is formed in the summer months (January to May ) and the opaque zone ( the slow growth period) is formed in the winter-spring months (June to December). Translucent zones are relatively thin. Declining sea-surface temperature (which was assumed to reflect water temperature change at depth) was associated with the onset of opaque zone formation in the otoliths of P. multidens. Furthermore, reproduction is unlikely to play a significant role in the timing of translucent zone forma- tion in P. multidens because spawning occurs primarily in the March -April period. These results indicate that water temperature, doubtless in association with other factors, provides a stimulus that Influences the endolymph fluid Newman and Dunk; Age validation, growth, mortality, and additional population parameters of Phstipomoides multidens 125 chemistry of these fish, culminating in the formation of an- nual growth increments. Female and male fish older than 20 years of age were uncommon in the landed catch. Fish of both sexes between 5 and 12 years of age were common in the landed catch. The maximum age of P. multidens observed in our study was much greater than that recorded previously. Richards^ reported a maximum age of 14 years in Papua New Guinea from counts of daily rings on otoliths, whereas Brouard et al. (1984) recorded a maximum age of only 8 years in Vanuatu with a similar method. Edwards (198.5) analyzed vertebrae and scales of this species in the Timor Sea and re- ported a maximum age of 14 years. In contrast, Mohsin and Ambak (1996) estimated a maximum age of only 5 years from the east coast of peninsular Malaysia with length- frequency analysis. Variation in the longevity estimates of earlier works is related to the aging methods used and their biases. For example, growth increments in vertebrae are often difficult to detect despite the presence of numerous discontinuities in bone growth (Marriott and Cappo, 2000). Alternatively earlier longevity estimates may have been drawn from sample populations biased by gear selectivity or from populations with varying degrees of exploitation. Otolith weight was a good predictor of age in P. multidens, accounting for 94% of the variability in age. The strong lin- ear relationship between otolith weight and fish age from a very large sample size implies that otolith weight may be used as a proxy for age. The coefficient of determination of the regression model is affected by the degree of colinearity of the independent variables. The high r^ value observed in our study provides the basis for a first-order age approxi- mation. Thus, the potential exists for an age-otolith-weight key to be derived for P. multidens, as for an age-length key, whereby the age composition of the landed catch in future years may be obtained by weighing large numbers of oto- liths. However, the accuracy and precision of adopting this monitoring strategy remains to be tested. The fit of the regression model for the otolith weight- age relationship was much more precise than the fit of the fork-length-age relationship as described by the von Ber- talanffy growth model. Considerable variation in length was observed within most age groups for both sexes. The large variation in length at a given age makes it difficult to accurately determine the age of P multidens from length data alone. For example, fish ranging in length from 450 to 550 mm FL may vary in age from 5 to 30 years. This vari- ability may explain the very low estimate of maximum age obtained by Mohsin and Ambak ( 1996), which was derived with length-frequency analysis. Growth was most rapid through age 9 for both sexes. From age 9 onwards somatic growth slows with increasing age. The estimation of growth parameters is dependent upon adequate sampling across the length range of any species. The fish sampled in our study ranged in length from 80 to 701 mm FL, covering most of the length range of P multidens. Therefore, it is unlikely that the growth parameters of P. multidens are biased because of inad- equate sampling across the length range. Despite methodological differences in age estimation, the estimates of K derived from the studies of Richards' 1995 n = 328 IS 20 22 24 26 28 30 32 1996 n=984 Hn-n^ 4 fi S :() 22 24 2f> 28 .^0 32 1997 n.702 „nnn^ s 1(1 i; 14 199S 11=1126 OQDii 14 16 IK 211 24 26 2S M) ^2 1999 n=573 24 26 2s Ml M Age (years) Figure 9 Age-frequency distributioii.s ol P. multidens in the northern demersal scalefish fishery from ISO,') to 1999. 126 Fishery Bulletin 101(1) (A'=0.188), Ralston (1987; Ar=0.188), and Edwards (1985; /i'=0.219) were somewhat similar to that observed in this study (A'=0.187). However, the asymptotic lengths report- ed by Richards' and Ralston (1987) were larger than our estimates, which were again similar to those of Edwards, 1985). In contrast, the estimates of A' derived by Brouard et al. (1984), Brouard and Grandperrin (1985), and Mohsin and Ambak ( 1996), which ranged from 0.28 to 0.50, provid- ed overestimates of the growth potential off! multidens as observed in our study. Clearly, methodological differences in age estimation have the potential to unduly influence growth parameter estimation and may provide misleading impressions of the production potential of these fishes. The similarity of growth in length-at-age between sexes indicates that there is little trade off in energetic invest- ment into reproductive activity after sexual maturity at the expense of somatic growth as there is for some Lutja- nus species (Newman et al., 1996, 2000b). Information on energy partitioning in the Pristipomoicles is not known. However, females with a large body size would be repro- ductively "fitter" if they could accommodate a large mass of hydrated eggs prior to spawning, especially in a mul- tiple male, multiple female spawning system. The long life span of P. multidens and other lutjanid species (Loubens, 1980; Newman et al., 1996, 2000a; New- man and Dunk, 2002; Rocha-Olivares, 1998) may be an evolutionary adaptation that supports iteroparity. Many demersal reef fish are highly fecund, but egg and larval survivorship is low; therefore, spawning over numerous years may be necessary to maintain stable populations. In addition, numerous years of reproductive output may also be required to contend with environmental variability (e.g. the incidence of cyclones. El Nino-La Nina cycling), which may substantially influence recruitment success. Extended periods of high exploitation results in decreases in the spawning stock biomass and constriction of the age structure of fish populations, and thus diminishes the number of effective spawnings. Any reduction in the number of effective spawnings may result in a decrease in ecological fitness and hence limit the adaptive capacity of the species to combat environmental or anthropogenic induced stress. Variation in life expectancy due to fishing pressure has the potential to bias estimates of M used in our study. To account for any Af associated difference, a range of M es- timates have been considered in our study. Pristipomoides multidens were fully recruited to the commercial fishery in the NDSF by age 6. Catch-at-age data showed relative- ly consistent estimates of Z among years from 1995-96 through to 1997-98 and a relatively broad age structure in the landed catch. Fishery management implications Throughout much of its range P. multidens composes a significant proportion of the demersal catch of tropical multispecios fisheries. Within these multispccies fisheries P. multidens is taken as part of the directed targ(>t catch or as a part of the retained catch. In fish trawl-based fish- eries, P. multidens can be harvested at all stages of their life history from juvenile to adult, making them especially vulnerable to overexploitation. In contrast, fisheries that use trap and line methods of capture (using bait to attract fish) only have the capacity to harvest fish in the subadult- to-adult phase of their life history. Hence, the method of capture and harvest strategy adopted has the capacity to influence the sustainable exploitation of the P. multidens resource. Of particular relevance to fishery managers is the ca- pacity that fish trawl-based fisheries have in being capa- ble of continuing to function and to be economically viable (driven by the more productive, lower value species) while populations of higher valued species such as P. multidens become depleted. Thus, careful monitoring of the P. multi- dens resource will be required, particularly in trawl-based fisheries. Fishery managers need to be responsive to the intrinsic vulnerability off! multidens to overhai-vesting as a corollary of its life history characteristics. Furthermore, fish such as P. multidens, which have low rates of natural mortality, low growth potential, extended longevity, ma- ture relatively late in life and are either dead or moribund as a consequence of internal hemorrhaging when the physoclistous is ruptured during capture, are likely to be particularly sensitive to exploitation pressure. The appar- ent low survival rate for released fish in the fishing depths of the NDSF fieet indicates that the traditional use of legal minimum sizes to increase survival to spawning sizes and hence increase overall yields is not a practical option. Populations off! multidens have a low productive capac- ity and hence are vulnerable to overfishing as a conse- quence of slow gi-owth, extended longevity, late maturity, and low rates of natural mortality The demersal fish re- sources of the NDSF, of which P. multidens is a significant part, is currently being managed with an innovative total allowable effort system that allocates individually trans- ferable effort units equitably to each licensee. However the highly mobile, efficient, and wide-ranging capacity of the NDSF fleet may require more complex management ar- rangements to maintain future breeding stock levels. The incorporation of appropriately targeted spatial or temporal (or both spatial and temporal) closures within the existing effort management framework is likely to provide an addi- tional useful and robust mechanism to maintain spawning stock biomass and protect against recruitment overfishing. In the wider Indo-Pacific region, fishery managers should consider han'est strategies of low frequency or low intensi- ty in conjunction with targeted spatial or temporal closures to protect the spawning stock biomass of these fishes. Harvest strategies such as setting fishing mortality at or hear natural mortality (F=A/) were often prescribed prior to the 1990s (GuUand, 1970). Recently, the adop- tion of harvest strategies such as setting F = F„ , were thought to be quite conset-vative, but usually resulted in F = M harvest strategies (Walters, in press). Following the meta-analysis of Myers et al. (1999), who examined stock-recruitment cui-ve slopes expressed as maximum reproductive rates per spawner at low spawner biomass, Walters (in press) has reported that optimal fishing mor- tality rates are substantially lower than natural mortality rates for most species and stocks. Furthermore, Patterson Newman and Dunk: Age validation, growth, mortality, and additional population parameters of Phstipomoides multidens 127 (1992) reported that fishing mortahty rates above 2/3 M are often associated with stock decHnes, whereas fishing mortahty rates below this level have resulted in stock re- covery. Consequently, exploitation rates for long-lived reef fishes need to be very conservative. The declines evident in the length and weight of fish in the landed catch over the duration of our study support the finding of the high levels of F. These data support the estimates of the annual percentage removals that indicate that the NDSF population of P. multidens is currently exploited above optimum levels. The age structure of the P. multiderjs stock within the NDSF currently consists of close to 30 age classes (ages 2 to 30 years). Therefore, de- pletion of the spawning stock biomass of these fishes will result in long population recovery times and the economic loss associated with recovering and rebuilding these fish- eries may persist longer A minimum of 30 years would be required for the fished population to recover in terms of both virgin spawner biomass and age structure. The results of our study provide the basis for a more detailed age-structured stock assessment for this species. Acknowledgments The authors gratefully acknowledge funding from the Fisheries Research and Development Corporation (FRDC) for this project. This work was undertaken as part of FRDC Project 97/136. The comments and suggestions of Rod Lenanton and Jim Penn and three anonymous reviewers contributed gi'eatly to this manuscript. Logis- tical support was provided by the Department of Fisher- ies, Government of Western Australia. The authors are thankful to the fishermen of the NDSF for the provision of samples and to the fish wholesalers of Perth (Attadale Seafoods Pty Ltd., Kailis Bros., New West Foods [W.A.I Pty. Ltd., Festival Fish Wholesalers) and Broome (Fortes- cue Seafoods) for access to specimens from northwestern Australia. Jerry Jenke provided invaluable support in all areas, Richard Steckis was responsible for maintaining the databases used for this project, and Peta Williamson assisted with the development of the figures. Literature cited Allen, G. R. 1985. FAO species catalogue. Snappers of the world. An annotated and illustrated catalogue of lutjanid species known to date. FAO Fisheries Synopsis 125, vol. 6, 208 p. FAO, Rome. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determmations. Can. J. Fish. Aquat. Sci. 38:982-983. Beckman, D. W., and C. A. Wilson. 1995. Seasonal timing of opaque zone formation in fish otoliths. In Recent developments in fish otolith research (D. H. Secor, J. M, Dean, and S. E. Campana, eds.), p. 27-44. Univ. South Carolina Press, Columbia, SC. Brouard, F., and R. Grandperrin. 1985. Deep-bottom fishes of the outer reef slope in Vanuatu. South Pacific Commission 17th regional technical meeting on fisheries (Noumea, New Caledonia, 5-19 August, 1985). SPC/Fisheries 17/WP.12, 127 p. [Original in French.] Brouard, F, R. Grandperrin, M. Kulbicki, and J. Rivaton. 1984. Note on observations of daily rings on otoliths of deepwater snappers. ICLARM (International Centre for Living Aquatic Resources Management) Translations 3, 8 p. ICLARM, Manila, Philippines. Cerrato, R. M. 1990. Interpretable statistical tests for growth comparisons using parameters in the von Bertalanffy equation. Can. J. Fish. Aquat. Sci. 47:1416-1426. Dalzell, P., and G. L. Preston. 1992. Deep reef slope fishery resources of the South Pacific. A summary and analysis of the dropline fishing survey data generated by the activities of the SPC Fisheries Progiamme between 1974 and 1988. Inshore Fisheries Research Project Technical Document 2, 299 p. South Pacific Commission, Noumea, New Caledonia. Edwards, R. C. C. 1985. Growth rates of Lutjanidae (snappers) in tropical Australian waters. J. Fish. Biol. 26:1-4. Fowler A. J. 1995. Annulus formation in otoliths of coral reef fish — a review. In Recent developments in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds.), p. 45-63. Univ. South Carolina Press, Columbia, SC. Gulland, J.A. 1970. The fish resources of the ocean FAO Fisheries Tech- nical Paper 97, 425 p. Hart, A. M.. and G. R. Russ. 1996. Response of herbivorous fish to crown of thorns star- fish AcanthaMer planci outbreaks. III. Age, growth, mor- tality and maturity indices of Acanthurus nigrofuscus. Mar Ecol. Prog. Ser 136:25-35. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 82:898-902. Kailola, R J., M. J. Williams, R C. Stewart, R. E. Reichelt, A. McNee, and C. Grieve. 1993. Australian fisheries resources, 422 p. Bureau of Resource Sciences, Department of Primary Industries and Energy, and the Fisheries Research and Development Cor- poration, Canberra, Australia. Loubens, G. 1980. Biologic de quelques especes de poissons du lagon Neo-Caledonian. III. Croissance. Cahiers de I'lndo-Paci- fique 2:101-153. Marriott, R., and M. Cappo. 2000. Comparative precision and bias of five different ageing methods for the large tropical snapper Luljanus johnii. Asian Fish. Sci. 13:149-160. Mohsin, A. K. M., and M. A. Ambak. 1996. Marine fishes and fisheries of Malaysia and neigh- bouring countries, 744 p. Universiti Pertanian Malaysia Press, Serdang, Selangor Darul Ehsam, Malaysia. Myers, R. A., K. G. Bowen, and N. J. Barrowman. 1999. Maximum reproductive rate offish at low population sizes. Can. J. Fish. Aquat. Sci. 56:2404-24 19. Newman, S. J. 2001. Northern demersal scalefish interim managed fishery status report. In Slate of the Fisheries Report 1999-2000 I J. W. Penn, ed.), p. 61-64. Fisheries Western Australia, Perth, Western Australia. Newman, S. J., and I. J. Dunk. 2002. Growth, age validation, mortality, and other popula- 128 Fishery Bulletin 101(1) tion characteristics of the red emperor snapper, Lutjanus sebae (Cuvier, 1828). off the Kimberley coast of North- western Austraha. Estuar Coast. Shelf Sci. 55 (1):67- 80. Newman, S. J., and D. McB. Williams. 1996. Variation in reef associated assemblages of the Lut- janidae and Lethrinidae at different distances offshore in the central Great Barrier Reef Environ. Biol. Fishes 46: 123-128. Newman. S. J.. M. Cappo, and D.McB. Williams. 2000a. Age. growth, mortality rates and corresponding yield estimates using otoliths of the tropical red snappers, Lut- janus erythropterus, L. malabaricus and L. sebae, from the central Great Barrier Reef Fish. Res. 48 (1):1-14. 2000b. Age, growth and mortality of the stripey, Lutjanus carponotatus (Richardson) and the brown-stripe snapper, L. vitta (Quoy and Gaimard) from the central Great Bar- rier Reef Australia. Fish. Res. 48 (3):263-275. Newman, S. J.. D. McB. Williams, and G. R. Russ. 1996. Age validation, growth and mortality rates of the trop- ical snappers (Pisces: Lutjanidae), Lutjanus adetii (Castel- nau, 1873) and L. quinquelineatus (Bloch, 1790) from the central Great Barrier Reef, Australia. Mar Freshwater Res. 47(4):575-584. Patterson, K. 1992. Fisheries for small pelagic species: an empirical ap- proach to management targets. Rev. Fish Biol. Fish. 2(4): 321-338. Pearson, D. E. 1996. Timing of hyaline-zone formation as related to sex, location, and year of capture in otoliths of widow rockfish, Sebastesentomelas. Fish. Bull. 94 (11:190-197. Ralston, S. 1987. Mortality rates of snappers and groupers, /h Tropi- cal snappers and groupers: biology and fisheries manage- ment (J. J. Polovina, and S. Ralston, eds.), p. 375^04. Westview Press, Boulder, CO. Rocha-Olivares, A. 1998. Age, growth, mortality and population characteristics of the Pacific red snapper, Lutjanus peru. off the southeast coast of Baja California, Mexico. Fish. Bull. 96:562-574. Walters, C. J. In press. Stock assessment needs for sustainable fisheries management. Bull. Mar. Sci. 129 Abstract — An assessment of the total biomass of shortbelly rockfish t.Sebastes jordani) off the central California coast is presented that is based on a spatially extensive but temporally restricted ich- thyoplankton survey conducted during the 1991 spawning season. Contempor- aneous samples of adults were obtained by trawl sampling in the study region. Daily larval production (7.56 x 10"^ lar- vae/d) and the larval mortality rate (Z=0.11/d) during the cruise were es- timated from a larval "catch curve." wherein the logarithm of total age-spe- cific larval abundance was regressed against larval age. For this analysis, lar- val age compositions at each of the 150 sample sites were determined by examination of otolith microstructure from subsampled larvae (n=2203), which were weighted by the polygonal Sette-Ahlstrom area surrounding each station. Female population weight-spe- cific fecundity was estimated through a life table analysis that incorporated sex-specific differences in adult g^rowth rate, female maturity, fecundity, and natural mortality (M). The resulting statistic (102.17 larvae/g) was insensi- tive to errors in estimating M and to the pattern of recruitment. Together, the two analyses indicated that a total biomass equal to 1366 metric tons (t)/d of age-l+ shortbelly rockfish (sexes combined) was needed to account for the observed level of spawning output during the cruise. Given the long-term seasonal distribution of spawning ac- tivity in the study area, as elucidated from a retrospective examination of California Cooperative Oceanic Fisher- ies Investigation (CalCOFI) ichthyo- plankton samples from 1952 to 1984, the "daily" total biomass was expanded to an annual total of 67,392 t. An attempt to account for all sources of error in the derivation of this estimate was made by application of the delta- method, which yielded a coefficient of variation of 19%. The relatively high precision of this larval production method, and the rapidity with which an absolute biomass estimate can be obtained, establishes that, for some species of rockfish {Sehastes spp.), it is an attractive alternative to traditional age-structured stock assessments. An approach to estimating rockfish biomass based on larval production, with application to Sebastes jordani* Stephen Ralston James R. Bence Maxwell B. Eldridge William H. Lenarz Southwest Fisheries Science Center National Marine FIshenes Service 1 10 Shaffer Road Santa Cruz, California 95060 E mail address (for S Ralston, contact auttior); Steve RalstoniS)noaa gov Manuscript accepted 20 September 2002. Fish. Bull. 101:129-146 (2003). Shortbelly rockfish (Sebastes jordani) is an underutilized species that is dis- tributed from Vancouver Island to northern Baja California (Eschmeyer, 198.3), although it is especially abun- dant along the central California coast. Based on a swept-area bottom trawl survey of demersal rockfish, Gunder- son and Sample ( 1980) estimated there were 24,000 metric tons (t) of shortbelly rockfish in the Monterey International North Pacific Fishery Commission (INPFC) area (35°30'N-40°30'N). This biomass estimate was far greater than that of any other species of rockfish in any area and, significantly, it did not include the midwater portion of the stock. Hydroacoustic estimates of short- belly rockfish biomass in the shelf-slope area between Ascension Canyon and the Farallon Islands (37°00'-38°00'N), a distance of only 110 km, have ranged from 153,000 to 295,000 t (Nunnely'). Although at present there is no di- rected fishery for this species (Low, 1991), much is known of its biology. Ear- ly work by Phillips (1964) provided ba- sic information about the length-weight relationship, growth as estimated from scale annuli, spawning seasonality (i.e. parturition), fecundity, maturity, and the food habits of shortbelly rockfish. Lenarz (1980) later studied shortbelly rockfish growth using ages from whole otoliths and provided preliminary cal- culations of the effect of fishing on the stock. He also demonstrated marked spatial variation in age and length composition along both latitudinal and depth gradients. Growth was re-esti- mated by Pearson et al. (1991) using ages determined from broken and burnt otoliths. From the hydroacoustic biomass estimates cited above and an estimated range for the natural mortal- ity rate (0.20-0.35 yr), they concluded that the maximum sustainable yield (MSY) of the shortbelly rockfish stock in the Ascension Canyon-Farallon Is- lands area was 13,400-23,500 t. Shortbelly rockfish is one of the few Sebastes spp. that can be readily identi- fied at all life history stages. Descrip- tions of the early life stages of short- belly rockfish, from preflexion larvae through the pelagic juvenile stage, were provided by Moser et al. ( 19771. Extend- ing that work, MacGregor (1986) pro- vided a summary of the spatiotemporal distributions of shortbelly rockfish larvae taken in California Coopera- tive Oceanic Fisheries Investigation (CalCOFI) cruises conducted in five different years. His results showed that 99.1"^^ of all shortbelly rockfish lar\'ae (-4-10 mm) were captured within 90 km of shore and that 65. 4"^* were sampled during the month of Febru- ary. Moreover, a strong peak in larval abundance (i.e. 34.7Cf of the coastwide total) was concentrated in the vicin- * Contribution 111 of the Santa Cruz Labor- atory, Southwest Fisheries Science Center, National Marine Fisheries Service, Santa Cruz, CA 95060. ' Nunnely, E. 1989. Personal commun. Alaska Fisheries Science Center, 7600 Sand Pomt Way N.E., Bin C 15700. Seattle, WA 98115-0070. 130 Fishery Bulletin 101(1) ity of Pioneer Canyon (CalCOFI line 63). Later research by Laidig et al. (1991) resulted in the development of a detailed growth model for young-of-the-year shortbelly rockfish, from extrusion through the late pelagic juvenile stage (-180 d), and verified the feasibility of a daily ag- ing protocol by validating a one-to-one correspondence between counts of daily increments and elapsed time in days. More recent work by Ralston et al. (1996) showed that larval shortbelly rockfish can be accurately aged by using optical microscopy. For the year 2000 the Pacific Fishery Management Council (PFMC) revised the shortbelly rockfish acceptable biological catch (ABC) downwards from 23,500 to 13,900 fyr (PFMC-). The new ABC is based on the low end of the estimated MSY range presented in Pearson et al. (1991); it was reduced due to a probable natural decline in stand- ing stock during the 1990s arising from poor ocean condi- tions (MacCall, 1996). The original range, however, was derived by using quite variable data from unpublished hy- droacoustic surveys and Pearson et al. (1991) considered it a strictly preliminary estimate. Given that the biomass of shortbelly rockfish along the central California coast was once thought to be very large (Gunderson and Sample, 1980; MacGregor, 1986), and that the species is still proba- bly the single largest rockfish contribution available to the west coast groundfish fishery, data are needed to estimate the size of the stock more precisely than results available from these previous investigations. The goal of this study was to develop an analytical ap- proach to estimate the total biomass of shortbelly rockfish in the region of Pioneer and Ascension Canyons and to gather field data to evaluate the method. Successful appli- cation of the method to shortbelly rockfish would provide to the PFMC information useful for management. On a more fundamental level, it would also assist in developing fishery-independent survey techniques capable of assess- ing other, more highly exploited, species of rockfish. The assessment approach The basic premise of egg and larval surveys is that it is easier to estimate the absolute abundance of ichthyo- plankton than it is to estimate that of adults (Saville, 1964; Gunderson, 1993). This is especially true when the spatial distribution of adults exhibits some type of size- or age-specific pattern. Shortbelly rockfish is one such spe- cies (Lenarz, 1980) and obtaining a representative sample of the adult population is challenging (Lenarz and Adams, 1980). Conversely, early life history stages (i.e. eggs and preflexion larvae) can be sampled effectively with stan- dard plankton nets (Smith and Richardson, 1977). Due to the direct coupling between egg [iroduction and spawning biomass, mediated through population weight-specific ■^ PFMC (Pacific Fishery Management Council). 1999. Status of the Pacific coast proundfish fi.shcry through 1999 and recom- mended acceptable biological catches for 2000, 44 p. + .54 tables and 5 figs. Pacific Fishery Management Council. 21.30 SW Fifth Ave.. Portland, OR 97201. fecundity(^[eggs/gl), egg and larval surveys have proven successful for estimating spawning biomass in many applications (e.g. Houde, 1977; Parker, 1980; Richardson, 1981; Lasker, 1985; Armstrong et al., 1988; Hunter et al, 1993). Members of the scorpionfish genus Sebastes are distinc- tive because they are primitive viviparous livebearers (Wourms, 1991), resulting in parturition of advanced yolksac larvae (Bowers, 1992). This reproductive strategy lends itself to a larval production stock assessment be- cause the age of all spawning products can be accurately determined from otolith microstructure (Laidig et al., 1991; Ralston et al., 1996). In contrast, in egg surveys, egg age is back-calculated to the time of spawning by 1 ) defining a series of developmental stages, 2) estimating the relationship between stage-specific developmental rates and temperature, 3) assigning a thermal history to each egg, and 4) determining the time required to account for embryo development from spawning to the obsei-ved stage (see for example Lo, 1985; Moser and Ahlstrom, 1985). Because a distinctive extrusion check forms on the otoliths of Sebastes larvae at the time of parturition (Ralston et al., 1996), a rockfish lai-val production estimate does not require information on temperature-dependent developmental rates and the ambient thermal history of egg samples. In the approach presented in the present study, a spa- tially extensive but temporally restricted ichthyoplankton survey was conducted. The age composition of larvae in each plankton tow was determined by subsampling the catch, aging the subsample, and expanding the subsample age composition back to that of the tow total. The total age-specific abundance of larvae in the study region was calculated by weighting the larval catch-at-age in each plankton sample by the polygonal area around it (see Sette and Ahlstrom, 1948). Characterization of the declin- ing trend in total larval abundance at age with an expo- nential mortality model allowed estimation of the produc- tion rate of day-0 larvae and the larval mortality rate at the time the ichthyoplankton cruise was conducted. Information on adult reproduction was obtained from contemporaneous data collected during two separate cruises conducted during the spawning season. In particu- lar, the following functional relationships were estimated from the data collected: 1 ) weight as a function of total length, 2) sex-specific weight at age (i.e. male and female von Bertalanffy growth equations), 3) fecundity as a func- tion of total weight, 4) maturity as a function of age, and 5) the population sex ratio. From these relationships, a life table was constructed, based on an estimate of natu- ral mortality (/yr), that yielded estimates of population weight-specific fecundity (O). As defined here, population weight-specific fecundity includes the biomass contribu- tions to total population size from males and immature females. Together, these estimates (daily larval production and population weight-specific fecundity) can be used to cal- culate the "daily" total biomass of fish in the population required to produce the obsei-ved abundance of lai^vae. The long-term mean seasonal distribution of shortbelly Ralston et al : An approach to estimating rockfish biomass from larval production 131 rockfish larvae and spawning activity was then determined by a retrospective analysis of all CalCOFI ichthyoplankton samples collected in the vicinity of Pioneer Canyon. Based upon the timing of our lar- val survey with respect to the long-term mean distribution of spawning activity, total "annual" biomass was calculated by expansion. Finally, the robustness of the total biomass estimate was evaluated through a simulation study and sensitiv- ity analyses. Methods Trawl sampling of adults 38- 3 37- lOO-fathom contour 124 Specimens of adult shortbelly rockfish were collected during February-March 1991 by the FV New Janet Ann and the RV Novodrutsk (Table 1). Shortbelly rockfish aggregations in and around Ascension and Pioneer Canyons (Fig. 1) were targeted from acoustic sui-veys. A total of 28 trawls (12 bottom and 16 midwater) were con- ducted over bottom depths ranging from 115 to 384 m, at an average net depth of 126 m (range: 18-210 m). The bottom trawl codend mesh size was 3.8 cm and the midwater trawl mesh was 5.1 cm. Dura- tion of the trawls ranged from 3 to 73 min (x=26 min). All landings were either fully weighed or subsampled, depending on the size of the catch. Landings were highly variable, ranging from 23 to 36,300 kg. Two subsamples were taken from each trawl landing. The first was used to determine the overall age, size, sex, and maturity composition of the catch; it was obtained by randomly selecting and examining 100-300 individuals from each trawl, depending on the availability of both time and fish. For these specimens, total lengths (TL) were measured to the nearest mm. gonads were examined to determine sex and to assign gi-oss maturity stages (see below), and otoliths were removed for later age determina- tion by the break-and-burn method (Pearson et al., 1991). A second smaller subsample of 25-50 specimens was also taken to estimate length-weight and fecundity relation- ships. For each of these fish.TL was measured, weight was determined to the nearest mg, the otoliths were extracted, and late vitellogenic ovaries were dissected from females and fixed in Gilsons fluid. We rated females in terms of maturity based on gross gonadal condition. A summary of the scale we used is the following: 1.0 = immature; 2.0-2.9 = vitellogenic oocytes (yolk deposition with associated oocyte and ovary enlarge- ment); 3.0-3.9 = fertilized eggs (embryos to hatched lar- vae); 4.0 = spent; and 5.0 = reorganization and recovery. Fecundity of female shortbelly rockfish was estimated gravimetrically from vitellogenic ovaries after 2-4 months Pioneer Canyon + + Ascension Canyon -I- + bongo station * adult trawl station * CalCOFI station 63.55 123 122 Longitude Figure 1 Map of the central California study region showing bongo net and adult trawl sampling locations. The annual spawning season was estimated by the long-term seasonal distribution of shortbelly rockfish larvae at CalCOFI station 63.55. Table 1 Summary infor nation of trawl collections for ad ults Trawl locations are shown as closed circles in Figure 1 NJA = RV 1 New Janet Ann NOV = RVNoi'odrutsk. Date Vessel Trawl type Bottom Midwater 14 Feb 1991 NJA 3 1 15 Feb 1991 NJA 2 1 22 Feb 1991 NJA 4 1 23 Feb 1991 NJA 3 1 15 Mar 1991 NOV 0 3 16 Mar 1991 NOV 0 3 17 Mar 1991 NOV 0 2 18 Mar 1991 NOV 0 2 19 Mar 1991 NOV 0 2 fixation with periodic stirring. Entire fixed ovaries from each female were blotted dry and weighed to the nearest 1.0 mg. Duplicate subsamples of both ovaries were weighed 132 Fishery Bulletin 101(1) to the nearest 0.1 mg and their egg contents counted with the aid of a dissecting microscope. The mean number of eggs per gram from each fish was then expanded to the total ovary weight to estimate annual fecundity (total eggs per individual female). To estimate population weight-specific fecundity, first the length and weight data were fitted to the power func- tion and the bias-corrected regression equation was used to estimate the weight of every fish that was aged. Next, for each sex, growth equations were obtained by fitting the weight and age "data" to the von Bertalanffy growth model ( Ricker, 1975 ). Maturity was quantified by fitting the logis- tic equation to the proportion of females that were mature within 5-mm-TL intervals, and fecundity was estimated by fitting fecundity and female weight data to the power function, with appropriate bias-correction. These various functional relationships were then combined in a life table analysis to determine the expected biomass per female recruit and the expected lifetime lai-val production per fe- male recruit. The ratio of these quantities is the estimated equilibrium cohort weight-specific fecundity (i.e. Oflarvae/ gl ) of female shortbelly rockfish. Given an estimate of total age-0 larvae (N,,), the female biomass responsible for the observed larval production can be estimated as A^q^"'. Finally, the combined sex biomass can be determined by expanding female biomass using weight-based cohort sex ratio estimates from the life table analysis. Ichthyoplankton sampling The primary set of ichthyoplankton samples used in our study was obtained by using bongo nets during a cruise of the NOAA RV David Starr Jordan (DSJ-9102) conducted in the winter of 1991. Sampling began at 1500 h on 8 Feb- ruary and ended at 0230 h on 15 February. During that 6V2-day period 150 stations were occupied in the region bounded from lat. 36°.30'N to lat. 38°00'N and offshore to a maximum distance of 130 km (Fig. 1). The study area included Pioneer and Ascension Canyons — two features in the continental slope known to harbor large numbers of adults (Lenarz, 1980; MacGregor, 1986; Chess et al., 1988). At these sites the sampling density was increased. Field and laboratory processing of the bongo net samples followed prescribed CalCOFI guidelines (i.e. Kramer et al., 1972; Smith and Richardson, 1977), with minor modi- fication. For example, at every fifth sampling station, the bongo frame was deployed with 333-pm and 505-)im mesh nets to determine the extent of extrusion of small larvae in the standard 505-|im mesh (Lenarz, 1972; Somerton and Kobayashi, 1989). After the nets were washed down, samples from both mesh sizes were preserved in 80% EtOH. At all other stations, the net frame was deployed with two 505-pm mesh nets; one sample was preserved in 80'/^ EtOH (to allow later age determinations from larval otoliths), and the other was preserved in lO'/f buffered for- malin. In addition, because Sehastes larvae were believed to occur only in the upper mixed layer (Ahlstrom, 1959), the maximum amount of wire deployed was 200 m, result- ing in a maximum depth fished ctiual to 140 m. Following splitting, sorting, identification, and enumeration of the larvae in the laboratory, abundance was expressed as the number of shortbelly rockfish per 10 m^ of sea surface. To estimate the age composition of the larval popula- tion, the sorted shortbelly rockfish larvae from each of the 150 EtOH-preserved bongo hauls were randomly subsam- pled for otolith microstructure examination. To determine the size of an age subsample (N^), based upon the total number of larvae occurring in a haul (A^^, ), we applied the following rule: 1) for A^,, less than or equal to 10, N^ = N/^, 2); for A^^, greater than 10 but less than or equal to 410, N^ = 10-1- 0.10 [AT; -10]; and 3) for A^;, greater than 410, N^ = 5o! Otoliths were extracted from each specimen in the haul subsample and individual ages determined by methods outlined in Laidig et al. (1991) and Ralston et al. (1996). The age composition of the larvae in each bongo sample was then estimated by expanding the percent age-fre- quency obtained from the subsample to the haul total (N/,). The estimated numbers-at-age of larvae in each haul were standardized to the number per 10 m^ of sea surface irij.^ for age T and haul i) by application of standard haul fac- tors (Kramer et al., 1972; Smith and Richardson, 1977). We expanded the n-j.^ to the entire survey area by using the method of Sette and Ahlstrom ( 1948). The Sette-Ahl- strom estimate is calculated by the following equation (see Kendall and Picquelle, 1990): ^T=Y,^^'^T" where for each of k hauls A, = the area that haul ( repre- sents (units of 10 m'-); and Nj, = the total abundance of lar- vae of age T in the entire survey area. The area for a haul (A, ) is defined as the area circumscribed by a polygon containing all points in space closer to a haul's location than to the location of any other haul. With this definition, we were able to write a simple computer program to calculate Sette-Alilstrom weights by dividing the study area into a fine grid and assigning each grid point to a haul. Note that this definition and procedure for obtaining Sette-Ahlstrom areas is equivalent to construct- ing polygons manually using perpendicular bisectors and measuring their areas (Sette and Ahlstrom, 1948). The mean daily larval production rate during the cruise was estimated as the bias-corrected antilogarithm of the y-intercept of the ordinary least-squares linear regression of log.lA^J against larval age. Moreover, the regression slope provides an estimate of the total instantaneous mor- tality rate of the larvae (Z |/d|). This calculation implicitly assumes that the age distribution of larvae was stationary throughout the 6V2-day period of the cruise. To determine if shortbelly rockfish larvae occur deeper in the water column than 140 m, a series of 1-m^ multiple- opening-closing-net with environmental sensing system (MOCNESS) tows was conducted aboard the RV David Starr Jordan (cruise l)SJ-9203i during the 1992 spawn- ing season. At that time, 21 tows were made in the area of Ralston et al : An approach to estimating rockfish biomass from larval production 133 Pioneer and Ascension Canyons from 1800 h on 21 Febru- ary to 0600 h on 23 February. Where bottom depth permit- ted, discrete depth samples were gathered using 505-pm mesh nets, sampling obliquely in the 0-40, 40-80, 80-120, 120-160, 160-200, 200-300, and 300-400 m depth inter- vals. Sampling was arrayed along seven onshore-offshore transects, each composed of three tows conducted at dif- ferent bottom depths, i.e. mid-continental shelf (110 m), the shelf-break (183 m), and well off the shelf (550 m). All samples were preserved in EtOH and after sorting, iden- tifying, and enumerating the larvae in the laboratory, we expressed abundances as the number of shortbelly rock- fish larvae per 1000 m'^ water sampled. Spawning seasonality At its inception, this assessment was intended to be an application of the fecundity reduction method described by Lo et al. (1992, 1993). However, samples from the Febru- ary and March adult trawl surveys showed that a higher proportion of females had completed spawning in the ear- lier cruise in comparison with the later cruise, an indica- tion that sampling was not representative during one or both of the cruises. Consequently, the fecundity reduction method was abandoned and an alternative approach was devised. Instead, we estimated the seasonal distribution of spawning activity based on the temporal distribution of preflexion shortbelly rockfish larvae in samples col- lected as part of the CalCOFI program from December to April from 1952 to 1984 (see Ahlstrom et al, 1978). We then used this seasonal spawning pattern to expand our estimate of the daily spawning biomass from the short period represented by our 1991 plankton samples to the entire year. To estimate the seasonal spawning distribution, we first identified the appropriate samples from CALCOFI station 63.55 in the vicinity of Pioneer Canyon by using results from MacGregor ( 1986) as a guide. Plankton samples at this location (Fig. 1) were re-examined and the total number of preflexion shortbelly rockfish larvae were enumerated from Sebastes subsets. These samples amounted to 41 plankton tows (bongo and ring nets) taken in 21 different years. Next, we calculated the mean density of preflexion lar- vae for each month, assigned these densities to the mid- point day of the month, and used nonlinear least-squares regression to fit the normal curve to approximate the sea- sonal spawning pattern, i.e. N,(t)- ¥ ^t-iij' where N „, = the estimated density of preflexion larvae on calendar day t (for December t is nega- tive); /J = the expected value of the seasonal distribu- tion of preflexion larvae; the standard deviation i and f = a "nuisance" scaling constant. a = the standard deviation of the distribution; and To determine the seasonal distribution of age-0 larval production that generates the seasonal distribution of preflexion larvae, we assumed that the preflexion larval period has a duration of 15 days (Laidig et al., 1991) and that larvae experience the estimated preflexion mortality rate (see above). As with preflexion larvae, we approxi- mated the seasonal distribution of age-0 larval production with a normal curve, with mean;/f, and standard deviation a^. Given particular values for/^„ and a^^ we calculated the corresponding relative numbers of age-0 larvae produced during each day of the spawning season, and the integrat- ed seasonal distribution of preflexion larvae, i.e. N,(t). ^£iVo«- i)e- We started the estimation with trial values of //q and a^ and then recursively adjusted the parameter estimates until the mean and standard deviation of the inferred sea- sonal distribution of preflexion larvae (A'',,,) converged on the empirical estimates of u and a . Lastly, based on the timing of the larval survey within the long-term mean spawning distribution, total biomass was calculated by simple expansion. Specifically, the midpoint of the 1991 bongo survey was 11 February (i.e. calendar day 42). Consequently we calculated A, which is the proportion of annual spawning under the age-0 larval production curve that occurs from day 41.5 to 42.5. Total population biomass was then estimated by multiplying the estimate of "daily" biomass by Ilk. This calculation im- plicitly assumes that the seasonal progression of spawn- ing has been stable over years. Therefore, the sensitivity of the biomass estimate to a violation of this assumption was evaluated by profiling over a range of values for the mean date of spawning (;/,,), which has a substantial effect on A. Precision of the biomass estimate The determination of total biomass requires the estimation of numerous statistical relationships, each with its own parameter set. The results of fitting these functions were then combined algebraically to produce the final biomass estimate. We calculated the precision of the final biomass estimate by using the delta method (Sober, 1982. p. 8), i.e. v\g(e)\ = Y^v\e,\\ — \ +2^^cov[0„0,]^ he, de, where g(0) = the algebraic combination of functions used to produce the final biomass estimates; and 6 = the full set of estimated parameters. Application of this method requires estimates of variances for each parameter, covariances among parameters, and partial derivatives of estimated biomass with respect to each parameter (dB/dO). Partial derivatives of the final biomass estimate with respect to the parameters were cal- culated numerically by using central differencing by per- 134 Fishery Bulletin 101(1) turbing each parameter ±1% and calculating the resulting effect on biomass. In some instances, variance estimates (i.e. squared stan- dard errors) and covariances were extracted directly from computer output produced by the SAS (1987) procedures PROC REG and PROC NLIN. In the case of linear regres- sions, these are the usual estimates of these quantities (e.g. Draper and Smith, 1981), whereas for nonlinear re- gression, these are asymptotic variances and covariances (e.g. Seber and Wild, 1992). In addition, following the as- sumptions of normal-based regression, the residual error parameter estimates (ct^,,,^,) used in bias adjustments are independent of other regression parameters (i.e. covari- ances are zero) and n&l^_.Ja^^^ has a chi-square distribu- tion with k degrees of freedom, where k is the degrees of freedom associated with (Tj,^^^. (e.g. for a linear regression k=n-2). This chi-square distributed random variable, i.e. no:. 2k, was then used to estimate the variance of the mean square errors (Larsen and Marx, 1981 ) var(o--.,. x2k. 60 5.5 O) 2 en 5.0 K. 0.2843 0.00917 1.73% -1.58% '0.9 -0.78 0.11026 -0.91% 0.96% Male von Bertalanffy growth WL... 209.9 3.0872 -0.67% 0.67% H', = W„„(l-exp[-/f„(r-«o,o.)l)'' K„ 0.2432 0.00984 -2.05% 2.01% to.cr -1.48 0.15183 1.55% -1.54% Adult survival iog,(no' 3.194 0.30662 0.00% 0.00% log,,(A'j.) = log^,(no)-M(r) M 0.2616 0.02656 -1.46% 1.53% Fecundity at weight log,( 0) 3.8155 0.1732 18.91% -15.90% log^,(F) = \ogJ0) + 5(log,,[W1) 8 ' 1.1416 0.0366 19.88% -16.61% MSE 0.2972 0.01827 0.92% -0.91% Maturity at length V 2.888 0.0237 0.00% 0.00% S= 1 + v/|l+exp|-p(rL-TL')ll P 0.6046 0.10318 0.00% 0.00% TV 135.05 0.85515 -0.15% 0.00% Larval production and survival log,(Afo) -0.3775 0.1737 -15.97% 19.01% log^(Nj.) = log^iN g)-zm Z 0.1107 0.01157 0.70% -0.67% MSE 0,1960 0.05659 -2.79% 2.87% Spawning seasonality "V 5.2 xlO" 2.36 X 103 0.00% 0.00% Nj.t) = (H'l tjpVIjr) exp[- (/ - //^ )V2]/*, tt)Q = the .v-intercept of the growth curve (yr); and j3 = the allometric growth parameter estimated from the regression of weight on length (Ricker, 1975). The /3 parameter is often set equal to 3.0, implying isomet- ric growth, although /3 was fixed at 2.980 in this applica- tion (see above). Likewise, 535 adult male fish were aged, their weights estimated from measurements of TL, and the data fitted to the weight-based von Bertalanffy growth model. Regression results for both sexes are presented in Figure 3 and Table 2. A simple exponential mortality model is used to describe the observed pattern of shortbelly rockfish abundance with age. The model is of the form where Wq = female weight (g) W__Q = the asymptotic mean weight of females at a hypothetically infinite age (g); Kq - the instantaneous growth coefficient specific to females (/yr); T = age (yr); N.,. loe ~MT where N-j. = the number of individuals alive at age T(yr); Pq = the extrapolated number at age 7^=0; and M = the instantaneous rate of natural mortality (/yr). 136 Fishery Bulletin 101(1) 300 - 250 - 200 - 150 - S ° ° § S LaJr^rS-a ^ 0 QMS-^gB e e 0 B-'^o ° ° 8 ° ° o §1^ ° S e e J^ ° o ° o 100 - 50 - 1/ |/i o ° Females I : o 1 i^ Weigh Ol , 1 , , , , o 0 o o o 200 -_ 150 - 100 - 0 o o o ° ° o 8 llJTi « o 0 o^ ° 0 o/ « o 50 - n ^/ :/■ o ° o Males ; 8 8 o : ? D 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 . 1 1 1 1 1 . 1 I 1 1 0 5 10 15 20 25 Age (yr) Figure 3 Weight-based von Bertalanffy growth curves fitted to shortbelly rockfish data collected by trawhng during February-March 1991. In this application a term for fishing mortality is assumeci to be unnecessary because there is no commercial or recre- ational harv'est of shortbelly rockfish. The model was fitted by a weighted linear regression of log^.lA^j,) on T, where the statistical weights were derived from expansions of the aged subsamples to the full trawl catches. Results showed a declining trend in abundance with age (Table 2, Fig. 4), and an estimated adult natural mortality rate of 0.26/yr Like the weight-length relationship, fecundity is typi- cally related to fish size with the power function (Bagenal and Braum, 1968 1, which is linearized by logarithmic transformation, i.e. F = 0W'\ \og^,(F) = log,(0) + Sx log,,(W), where /■' = individual annual Iccuiulilv i larvae/female); log., I W = female weight (g); and and 5 = fitted parameters. In our study fecundity estimates were gathered from 531 females taken during the trawl surveys and the transformed data were fitted by simple regression (Fig. 5, Table 2). There was considerable residual variability in the fecun- dity at weight relationship (r2=0.65; a%^^. =0.29715) and, in this instance, the addition of a bias-correction term had a noticeable effect on back-transformed predictions of fecundity at weight on the arithmetic scale. Although the distribution of regi'ession residuals deviated significantly from normal (P=0.0001i, due in large part to negative skewness, the overall fit was deemed adequate. Predic- tions of fecundity at weight were ~25-35'^( less than the results presented in Lenarz ( 1980), although his equation is based on only the 10 data points provided in I'hillips 11964). Ralston et al.: An approach to estimating rockfish biomass from larval production 137 3 0 - 2 0 - u c cr 1.0 T 0.0 r c y a -1.0 T -2.0 -_ a -3.0 ^ -4.0 - 5 ' I ' ' ' I I ' 10 15 Age (yr) 20 '"I 25 Figure 4 Weighted linear regression of log^.-transformed shortbelly rockfish abundance on age (statistical weights based on expansions of aged subsamples to full trawl catches; all samples collected by trawling during February-March 1991). There was no evidence that the sex ratio of the fish sampled during the 1991 spawning season (n = 1121 from February and March cruises combined) varied with age. A two-way test for independence of age and sex yielded X^=27.55, df=23, P=0.23. Moreover, there was no evidence that the overall population sex ratio was other than 50:50 (A^^->=586. N.y=535: ;f-=2.32, df=l, P=0.14). From these re- sults, we concluded that females and males both enter the population at approximately the same rate and thereafter they experience a similar natural mortality rate. The maturity stage data were used to establish a ma- turity schedule for female shortbelly rockfish collected during the 1991 winter spawning season. Fish were first stratified by size class (5-mm-TL intervals), and for each size group, the mean coded ovarian stage was computed. The data were then fitted by nonlinear regression (SAS, 1987) to a logistic model of the form S=l+- v l + e -p'TL-TL't where S = the mean coded maturity stage; TL = total length (mm); and V, p, and TL' = fitted parameters. Results show (Fig. 6, Table 2) an abrupt change in ovarian condition at a length of 135 mm. At the time of sampling (February-March), virtually all fish above that size were gestating or had already released their larvae, whereas fish smaller than that cutoff size were almost exclusively immature. All females were therefore assumed to be reproductively mature if TL was greater than 135 mm. The proportion of fish in each age class that exceeded 135 mm TL was used to define an empirical age-based 12.0 - 11.0 - o (fecundity) o b i 8.0^ 7.0 - o <^ 0 Pi n b.U -j . 1 i 1 1 . . 1 . 1 1 . . . 1 1 1 1 1 1 1 1 , 1 I 1 r I T 1 3.0 3.5 4.0 4.5 5.0 55 60 lege (total weight [gm]) Figure 5 Fit of ordinary least-squares regression to log^-trans- formed fecundity and weight data from shortbelly rockfish sampled by trawling during February-March 1991. 100 150 200 Total length (mm) 250 Figure 6 Fit of the logistic maturity function to coded ovarian devel- opmental stage data. Circles are means, which are brack- eted by ±1.0 standard deviation (all samples collected by trawling during February-March. 1991 ). maturity ogive. Results indicated that 7.9^; of l-yr-old fish spawned, whereas 99. 0"^/ of 2-yr-old females reproduced. When coupled with some type of recruitment model, the four functional relationships given in Figures 3-6 can be used to estimate population weight-specific fecundity and a weight-based population sex ratio. These latter two vari- ables are presented in Table 3 as part of a life table projec- tion for shortbelly rockfish. In the table, age is increment- ed discretely in one year steps to a maximum life span of 30 yr, which extends well beyond the maximum observed age of 22 years (Pearson et al., 1991). All calculations were 138 Fishery Bulletin 101(1) Table 3 Life table for shortbelly rockfish (Sebastes joi ruary-March). ■dani) based upon samples of ad ults obtained during the 1991 spawning season (Feb- Age (yr) 9or o- numbers 9Wt (g) 9 Cohort Wt(g) Fecundity (larvae/9) Weight-specific fecundity (no. of larvae/g of female) Proportion mature Cohort fecundity (larvae) o-Wt (gl cr cohort Wt(g) 1 1.0000 15.9 15.85 1235 77.90 0.079 98 19.9 19.89 2 0.7698 41.0 31.54 3651 89.11 0.990 2783 39.6 30.50 3 0.5926 71.5 42.37 6893 96.42 1.000 4085 62.0 36.72 4 0.4562 102.4 46.72 10.390 101.45 1.000 4740 84.4 38.50 5 0.3512 130.8 45.93 13,736 105.03 1.000 4824 105.3 36.99 6 0.2704 155.3 41.98 16,710 107.61 1.000 4518 124.0 33.52 7 0.2081 175.6 36.55 19,227 109.50 1.000 4002 140.0 29.14 8 0.1602 192.0 30.76 21,291 110.89 1.000 3411 153.6 24.60 9 0.1233 205.0 25.28 22,943 111.93 1.000 2830 164.7 20.32 10 0.0950 215.1 20.43 24,245 112.70 1.000 2302 173.9 16.51 11 0.0731 223.0 16.30 25,258 113.27 1.000 1846 181.3 13.25 12 0.0563 229.0 12.89 26,040 113.70 1.000 1465 187.2 10.54 13 0.0433 233.6 10.12 26,640 114.02 1.000 1154 192.0 8.32 14 0.0333 237.2 7.91 27,098 114.26 1.000 904 195.8 6.53 15 0.0257 239.8 6.16 27,446 114.44 1.000 705 198.8 5.10 16 0.0198 241.8 4.78 27,710 114.58 1.000 548 201.1 3.98 17 0.0152 243.4 3.70 27,910 114.68 1.000 425 203.0 3.09 18 0.0117 244.5 2.86 28,061 114.76 1.000 329 204.5 2.39 19 0.0090 245.4 2.21 28,175 114.82 1.000 254 205.7 1.85 20 0.0069 246.1 1.71 28,261 114.86 1.000 196 206.6 1.43 21 0.0053 246.5 1.32 28,326 114.89 1.000 151 207.3 1.11 22 0.0041 246.9 1.02 28,375 114.92 1.000 117 207.9 0.85 23 0.0032 247.2 0.78 28,412 114.94 1.000 90 208.3 0.66 24 0.0024 247.4 0.60 28,440 114.95 1.000 69 208.7 0.51 25 0.0019 247.6 0.46 28,461 114.96 1.000 53 208.9 0.39 26 0.0014 247.7 0.36 28,476 114.97 1.000 41 209.1 0.30 27 0.0011 247.8 0.28 28,488 114.97 1.000 32 209.3 0.23 28 0.0009 247.8 0.21 28,497 114.98 1.000 24 209.4 0.18 29 0.0007 247.9 0.16 28,504 114.98 1.000 19 209.6 0.14 30 0.0005 247.9 0.13 28,509 114.98 1.000 14 209.6 0.11 Tola 1 411.37 42.028 347.65 based on a discrete time origin that was centered in the spawning season and it was assumed that samples were obtained at that time. For simpHcity, constant annual recruitment to the pop- ulation at age 1 is assumed, although this particular assumption is later relaxed and its specific effect on es- timates of population weight-specific fecundity is evalu- ated. To start the simulated population, recruitment was arbitrarily set equal to N, = 1.0/yr for both females and males. Then, given female weight at age (Fig. 3) and female numbers at age (Fig. 4), one can calculate age-specific female cohort biomass as the product of numbers and individual weights (Table 3), which when summed over all ages yields the total equilibrium female biomass (411.37 g/female recruit). Likewise, age-spe- cific cohort fecundity is calculated a.s female numbers at age, multiplied by estimates of individual female fe- cundity (Fig. 5), multiplied by the proportion of females that are mature (Fig. 6), which when summed over all ages classes yields the expected lifetime larval produc- tion of a cohort (42,028 lai^vae). The ratio of these two quantities (0=102.17 larvae/g of female) estimates the population weight-specific fecundity of female shortbelly rockfish. This population statistic can be compared with individual age-specific estimates presented in Table 3. These weight-specific fecundities range from 77.90 lar\'ae/g offemale for the mature 1-yr-old females, to 114.98 larvae/g of female for the oldest fish. It is also revealing to compare predicted weight-specific fecundity at age from the life table analysis (Table 3i with the observed data calculated directly from individual fish (Fig. 7). Results show a great deal of variability in the observed data, which is consistent with the extensive residual variability in fecundity (see Fig. 5, Table 2). Life table i)re(iictions were generally similar to, but slightly higher than, the observed data. This slight difference is Ralston et al ; An approach to estimating rockfish biomass from larval production 139 175 150 125 100 75 i 50 25 0 t iU ^ life table prediction o observed ' I ' ' 10 Age (yr) 15 20 Figure 7 Weight-specific fecundity and its dependency on age. Observed values are means of measured females, brack- eted by ±1.0 standard error and ±1.0 standard deviation (all samples collected by trawl during February-March, 1991). The solid line, which is not fitted to the "observed" points, is the predicted relationship from the life table analysis presented in Table 3. due to the bias correction that occurs when the fecundity relationship, which was fitted on the log-scale, is back- transformed to the arithmetic scale. Irrespective of the type of data, however, it is evident that weight-specific fecundity is only weakly dependent on age. This finding implies that changing the equilibrium age structure of the population, as mediated through an alteration in the natural mortality rate (M), will have little effect on the population's weight-specific fecundity. The assumption that recruitment is constant and uni- form does not appear to appreciably bias the estimate of population weight-specific fecundity derived from the life table analysis (102.17 lai-vae/g of female). Population weight-specific fecundity was also estimated in a Monte Carlo life table simulation that used a more realistic lognormal recruitment model (Fogarty, 1993). Annual recruitments in that simulation were determined as N^ = exp(;/ -I- aX). where N^ is the number of recruits at age 1, // = log^,|10,0001, a = 0.921, and X is a standard normal deviate (i.e. X~Af[0,l]). This level of variability in lognor- mal recruitment is comparable to that observed in the widow rockfish fishery (Bence et al., 1993: Hightower and Lenarz*), where 20-fold differences in recruitment have been observed in a 10-yr time period. Results of the simu- lation showed that fluctuating, lognormal recruitment can Hightower, J. E., and W.H.Lenarz, 1990. Status of the widow rockfish fishery in 1990. In Status of the Pacific coast ground- fish fishery through 1990 and recommended acceptable biologi- cal catches for 1991. Stock assessment and fishery evaluation, appendix vol. 2, 48 p. Pacific Fishery Management Council, Portland, OR. 800 -3 700 ^ 600 -. >, 500 H (J § 400 - CT : '^ 300 -. 200 100 ^ 0 p(91.3 < larvae/g < 106.4) = 0.90 > X = 101.06 n = 5000 -.nnn I T I I I T I jyVtVV'i"i"i"i"|"i"i"i"i"i"i"i"i"i"|' I I I ' r I I 70 80 90 100 110 Population weigtit-specjfic fecundity (larvae/g of female) Figure 8 Results of Monte Carlo simulation evaluating the effect of recruitment stochasticity on calculations of shortbelly rockfish population weight-specific fecundity. produce values of population weight-specific fecundity that range from 71 to 110 larvae/g of female, depending on the exact sequence of year classes and their resulting affect on age structure (Fig. 8). Even so, the mean of the sample distribution (.v = 101.06 larvae/g of female, n=5000) did not differ significantly from the life table calculation that had no recruitment variability. In addition, 90'^f of all the lognormal recruitment realizations were within ±10'^f of the constant recruitment result. By definition, population weight-specific fecundity rep- resents the number of larvae produced by one gram of female biomass (including immature 1-yr-olds). We ex- panded female biomass to total biomass including males. Results presented in Table 3 show that age-specific male cohort biomass, like that of females, is calculated as the product of numbers-at-age and individual weights-at- age, which when summed over all ages yields the total equilibrium male biomass (347.64 g/male recruit). Total population biomass (i.e. females-t-males) is then 759.01 g, of which females comprise 54.2'7f by weight. Thus, the total biomass is estimated by applying a 1.845 expansion factor to female biomass. Larval production Sampling with different mesh-size bongo nets (333 and 505 pm) allowed an assessment of whether a portion of smaller larvae retained by the smaller mesh was lost with the standard 505-pm mesh. Although shortbelly rockfish are relatively large and stout at parturition (-5.0 mm, Moser et al., 1977), undersampling of small, young larvae could seriously bias larval production estimates. Results presented in Figure 9 show, however, that the standard- ized catches of larvae ( number/| 10 m'-^l ) in the two net sizes were quite similar. A paired t-test of catches (333 minus 140 Fishery Bulletin 101(1) — I — I — I — I — I — I — I — r — I — I — I — I — 1 — I — I — I 3.0 4.0 5.0 60 70 loQc (catch witti 505-(.im net) Figure 9 Paired comparison of log^-transformed larval shortbelly rockfish catches taken in bongo nets with different mesh sizes (all samples collected during February 1991). Also shown is the line of equality. 505) resulted mx= -0.1458, n = 23J = -1.24, and P = 0.23, indicating no difference in the catch of the larger mesh net in relation to the smaller net. We conclude that the 505-pm net was an effective sampler of rockfish larvae. Examination of the standardized catches from the MOC- NESS tows conducted in 1992 revealed that shortbelly rockfish larvae were not caught in depths below the 120- 160 m interval (Fig. 10). The mean catch rate in that depth range was 16.8 larvae/1000 ni'^ which amounted to only l.S'/f of the combined average catch at all depths, i.e. 98.7% of all larvae were captured at depths < 120 m. Be- cause the bongo net was deployed to a maximum depth of 140 m in 1991, and because the mixed layer depth was de- pressed during MOCNESS sampling in 1992 due to strong El Nino conditions (Lynn et al., 1995), we concluded that the 1991 bongo net survey sampled the entire depth range where shortbelly rockfish larvae occurred. For this assessment, a total of 2203 shortbelly rockfish larvae were subsampled from the 505-pm bongo net catch- es and were aged by using optical microscopy (see Ralston et al., 1996). Results from that work indicated that the ages were quite precise (849^ agreement among three readers to within ±1 d) and there were, moreover, no increment in- terpretation differences between optical observations and those made with a scanning electron microscope (SEM). The horizontal distribution of very young (0-2 d) short- belly rockfish larv-ac was strongly associated with the continental shelf break (Fig. 11) and, latitudinally, with the Pioneer Canyon area. Due to the coincidence of this distribution with the locus of adult sampling sites (Fig. 1), we concluded that the trawl survey sampled the adults that produced the lai-vae captured in the iclithyoplankton survey. This finding supports the coupling of our popula- tion weight-specific fecundity statistic (102.17 larvae/g of female) with larval production to estimate the total bio- Sample depth (m) Figure 10 Mean catch rate of shortbelly rockfish larvae in seven depth strata taken by a MOCNESS sampler during March 1992. Means are bracketed bv ±1.0 standard error mass of shortbelly rockfish in the Pioneer Canyon region. Although not shown, older age classes of larvae tended to be more dispersed and to occur increasingly to the north- west— a pattern consistent with hydrographic conditions at the time of the survey (Sakuma et al., 1995). When Sette-Ahlstrom weights were used to expand age- specific standardized bongo net catches to the entire study region, the composite age-frequency distribution of short- belly rockfish larvae appeared to support a log^-linear model with constant exponential mortality (Z |/d| ), i.e. iV^ = AT^e^^r iog^.(7Vj,) = log,,(A^o) - ZT. where Nj = the total abundance of larvae of age Ttd); Z = the instantaneous larval mortality rate (/d); and 7V„ = the lai^val renewal rate (i.e. daily production of larvae). The model was fitted over the first 25 days of life, which largely represents the preflexion stage. In addition, the N-j. were first coded by scaling all obsei-vations to 1x10" and 0.5 was added to all larval ages as a continuity correction (see Ralston et al., 1996). Results show (Fig. 12, Table 2) a satisfactory fit, although there is some suggestion of an aberrant, serially correlated pattern in the residuals from age 0-9 d. The ,v-mtercept term (log.liV,,! = -0.3775), when back-transformed iwith bias-correction 1 yields N,, = 7.562x10'" age-0 lai-vae. Given estimates of daily larval production (7.562x10'" larvae) and population weight-specific fecundity (4>= 102.17 lai-vae/g of female), we calculated from the ratio of these two quantities that 740 t of female shortbelly rockfish spawned each day during the bongo net cruise (DSJ-9102), which is equivalent to a daily biomass of 1366 t/d (sexes combined). Ralston et al : An approach to estimating rockfish biomass from larval production 141 38- OJ ■D 37- O 124 00000 d

0.05). However, among the animals with an age equal to or greater than 24 years (7!=7), 85.7% were females. The oldest male was 29 years old and the oldest female was 30 years old (Fig. 2). Age estimates for S. guianensis individuals varied from 0 to 30 years. Although the age mode was in the 0 and 1 year classes (Fig. 2), 53.5% of all animals whose ages were estimated were seven years or more. This proportion remained relatively constant between the sexes; 55% of the males and 50% of the females were equal to or greater than seven years. Growth When applied to the present data, Schnute's model indi- cated that the von Bertalanffy growth equation fitted the length and age data of S. guianensis better (Table 1, a>0 and b>l; Schnute, 1981). Even though the predictive power of Schnute's model is greater than von Bertalanffy 's (see "Explained variance" in Tables 1 and 2), the latter is justi- fied by its historical use, and therefore has a greater value for populational comparison, and incorporates a better understanding of the biological meaning of its variables. Comparing the total lengths (TL) of the estuarine dol- phins of six years or more, we found no significant differ- 380 Fishery Bulletin 101(2) ence between the sexes (^test, P>0.05). However, it was ob- served that males possibly exhibit a discontinuity in growth around five years. The existence of a secondary growth spurt around this age was considered to be due to the onset of puberty in this species (Rosas and Monteiro-Filho, 2002), and necessitated the calculation of separate growth curves for each sex. It should be noted that sexual maturity of the dolphins here analyzed was determined by Rosas and Monteiro-Filho (2002) to occur at seven years in males. In order to estimate the fit of a two-step model, the sample was divided into two groups: 1) up to five years (prepuberty) and 2) older than five years (subadult and adults). The growth parameters obtained for males and females are given in Table 2. The results obtained by Borobia ( 1989) and Schmiegelow ( 1990) using the von Bertalanffy growth model are also indicated in Table 2 for comparison. The growth parameters obtained by the analyses of males up to five years of age and those older than five are presented Table 1 Schnute growth model parameters applied to Sotalia guianensis on the Parana coast. southern Brazil, "r," and "r2" are predetermined ages in years; "vj" and "y^" ^""^ estimated sizes at ages r, and T.,, in cm; "a" and "b" are adi- mensional parameters. "SQ" represents the residual sum of squares, and "Expl. var" represents the variance of the data explained by the model. Parameters Females Males All I^i 0 0 0 y, 93.11 86.04 89.53 h 28 28 28 ^2 181.72 190.45 185.70 a 0.14 0.07 0.13 h 7.64 9.45 7.86 SQ 1242.35 1898.75 4375.56 Expl. var. (%) 93.47 92.17 92.47 n 28 34 71 in Table 3. By dividing the sample in two, the fit of the von Bertalanffy model improved considerably (Tables 2 and 3). The growth curves of S. guianensis males and females obtained by the von Bertalanffy model are presented in Figure 3. The t-test applied to parameters a and b of the weight/ length regression equations for males and females revealed a significant difference (t=2,25; df=38; P<0.05). Therefore, this relationship was analyzed for the sexes separately and the equations obtained were 7W = 3. 156 X 10-6 X XL ^ 2836 TW = 8.974 X 10-5 X tL ^ "82 (males) (r=0.96) (females) (r=0.95). Discussion Age estimation Although there was no difference in the age estimation between teeth orientated in the buccal-lingual and ante- rior-posterior planes, we recommend the buccal-lingual plane to obtain easier on-center or close-to-center sections, which are essential for accurate age estimates. The differences found in counting GLGs in teeth from the anterior extremity and the median region of the tooth row of the same animal corroborate the results obtained by Hui ( 1980) for Tkirsiops truncatus. Therefore, we also do not recommend using teeth from the most anterior part of the mandible for age estimation in S. guianensis. The mean distance between the neonatal line and the end of the first GLG obtained in the present study (622.4 pm) was approximately double that obtained by Ramos (1997) (297.8 pm) for estuarine dolphins on the coast of Rio de Ja- neiro. The differences, however, must be analyzed carefully: the measurements carried out in our study were always made in the neck of the teeth, whereas those made by Ra- mos ( 1997 ) were from the base of the neonatal line. However, the differences may be related to the interpretation of the position of the first annual layer The accessory layers (no- nannual), observed between the neonatal line and the end Table 2 Von Bertalanff} growth model parameters applied to Sotalia guianensis on the Parana coast, southern Brazil, and parameters from the literature. " TL, "= asymptotic length ( cm), "k"= growth constant and '<„"= theoretical age at which the length of the animal is zero. "SQ" represen ts the residual sum of squares, and ' Expl var" represents the variance of the data explained by the model. Parameters Our study Borobia (1989) Schmiegelow (1990) Females Males All TL^ 177.31 179.10 179.53 187.21 182.6 k 0.66 1.00 0.79 0.20 0.41 'o -1.23 0.72 0.95 -4.05 -1.57 SQ 1944.25 3732.30 6942.93 — — Expl. var (%) 89.78 84.61 88.06 — — n 28 34 71 24 22 Rosas et al.: Age and growth of Sotalia gutanensis 381 of the GLG of the first year, frequently appear in a very conspicuous manner, especially in the tip of the tooth, and can be easily confused with annual layers. The assumption that accessory layers are annual could result in a duplication of the real age of young animals up to two years old, with significant consequences in the interpretations of populational bio- logical parameters ( Hohn, 1990 ). The ideal situation would be that a GLG deposition model already existed for the species being studied, thereby avoiding counting acces- sory layers as being annual (Hohn et al., 1989). In most odontocete species, includ- ing S. guianensis, accessory layers do not continue up to the end of the root of the tooth, in contrast to annual layers, which can be seen from the tip to the base of the root of the tooth. However, to identify ac- cessory layers it is necessary that the sec- tions selected for age determination are central, or close to the center of the pulp cavity (Pinedo and Hohn, 2000). Off-center sections can be used for age estimation, but reading errors increase markedly and consequently induce unreliable age esti- mates (Pinedo and Hohn, 2000). The reasons for GLG deposition in teeth are unknown (Hohn et al., 1989). Howev- er, several reasons have been suggested, including seasonal variations in growth rate, genetic physiological cycles, dietary changes, hormonal influences, and intrin- sic factors on the metabolism in general (Boyde, 1980; Klevezal, 1980; Scheflfer and Myrick, 1980). Although all these factors could be influential, variations in the diet certainly play a significant role. According to Klevezal ( 1996), a descriptive record of the dietary changes of an animal during the year should initially be looked for in structures that have a large degree of sensitivity, such as teeth. It is known that dentine reacts to the introduction of fluoride, calciferol and a series of other components in the organism, forming layers with different degrees of mineralization (Klevezal, 1996), which is known as a calcium-traumatic reaction of dentine. Therefore, it is possible to find a record of dietary changes in the dentine, starting from weaning (Klevezal, 1996). We believe that the accessory layer in the dentine found at approximately 419.6 pm from the neonatal line, could be a record of the end of weaning in the estuarine dolphin. It was observed in all the teeth of individuals older than 6.7 months and could be a hypomineralized layer caused by a reduction of calcium in the body due to the absence of milk in the diet (Klevezal, 1996). The other accessory layer found closer to the neonatal line (mean of 248.9 pm) was not observed in all animals and the interpretation of this layer remains uncertain. It may be related to the beginning of weaning, as has been suggested for the bottlenose dolphin 200- 180- A . . fi ^^-^"A A A A 160- ^^-b-tr-n 140 - i 120- f 100- o oU T- £ 0 CJ) c "en ° 200 180 160- 5 10 15 20 25 30 B \ ° 0 oo oo ^ o o o o °p 140 - / 120- / 100- : 0 5 10 15 20 25 30 Age (years) Figure 3 Growth curve of male (A) and female (B) Sotalia guianensis on the Parana coast, southern Brazil. Table 3 Von BertalanfTy growth model parameters for male Sotalia guianensis on the Parana coast, southern Brazil. "TL,"= asymptotic length (cm), "*"= growth constant and '7,j"= theoretical age at which the length of the animal is zero. "SQ" represents the residual sum of squares, and "Expl. var" represents the variance of the data explained by the model. Parameters Up to 5 years More than 5 years k SQ Expl.var (%) n 159.64 2.02 -0.38 510.99 94.20 15 186.41 0.53 -1.40 1013.98 50.90 19 382 Fishery Bulletin 101(2) (T. truncatus) (Hohn*). This hypothesis still needs to be confirmed. However, all the S. guianensis individuals that were still nursing, but which already had remains of solid food in their stomachs («=5), had only an accessory layer that is closer to the neonatal line — they did not have the layer that we are assuming marks the end of weaning. According to Rosas (20001, there was no significant difference in incidental catches between mature and im- mature individuals of S. guianensis caught on the coast of Parana, suggesting a similar vulnerability of young and adult estuarine dolphins to fisheries. Because the animals analyzed in our study were the same ones used by Rosas (2000), this lack of significant difference between mature and immature individuals can suggest a representative age distribution of the individuals analyzed. Because the maximum estimated age in our study was 30 years, and because the dolphins here analyzed were incidentally caught in fishing nets, it seems reasonable to assume that the longevity of the estuarine dolphin may be 30-35 years. This hypothesis is also corroborated by the study carried out by Ramos (1997) with S. guianensis on the coast of Rio de Janeiro State (southeastern Brazil). Although the age of the oldest male observed in our study was 29 years, the frequency of males older than 21 years was less than 3%, which is extremely low when compared with the frequency of 21.5% for females older than 21 years. These results suggest a greater life expectancy for females, which is also corroborated by a study carried out by Ramos (1997) in Rio de Janeiro. Growth The use of Schnute's model is helpful in deciding which growth model should be used. Even though the researcher can usually decide which model is most appropriate by looking at the data, subtle differences in data distribution could cause one or another model to be more adequate. Use of a generic model allows this choice without intervention of the researcher and avoids any unconscious bias towards or against any model. The discontinuity of growth in male S. guianensis in our study could have been due to the small sample size or may have been due to a second growth spurt, which has already been observed in the total length oiStenella attenuata (Per- rin et al., 1976), Lissodelphis borealis (Ferrero and Walker, 1993), and Phocoenoides dalli (Ferrero and Walker, 1999), and in the weight of male Tursiops truncatus (Cockroft and Ross, 1990). The k value obtained for male S.guianensis up to five years was very high, meaning that asymptotic length in this phase of life was reached quickly. The ces- sation of growth exhibited by the model for males up to 5 years probably is not true in the biological sense but could be an artifact created by the model and the small sample size. Most probably there is a marked reduction in growth with the start of sexual maturation and a greater investment in the weight or reproductive apparatus (or ' Hohn, A. A. 1999. Personal comimiii licMiiDirt l.ahoralory, Sc)ulli(>ast Fishiories Scienct' ('enter. National Marine Fi.slieries Ser\'ice, 101 I'river.s Island Road. Bcaulort, NC 28516-9722. both). The hypothesis of a greater investment in weight is supported by the observed difference in the weight-length coefficient between males and females. Additionally, sexual investment of male estuarine dolphins is very high — testes of adult males can reach up to 32 cm in length and weigh up to 3.3% of the total body weight (Rosas and Monteiro- Filho, 2002). After the secondary growth spurt in males, the final asymptotic length did not differ very much from that in females. Previous growth studies carried out by Borobia (1989), Schmiegelow (1990), and Ramos et al. (2000) with the estuarine dolphin did not mention the existence of a second growth spurt in males, possibly because the authors did not analyze the growth of males and females separately. According to Ramos et al. (2000), male and female data were combined because of the absence of sexual dimor- phism in the body size of adults of this species. Borobia (1989) and Schmiegelow (1990), who also used the von Bertalanfiy model, obtained different values for the growth equation parameters (Table 1). The sample used by Borobia ( 1989 ) did not have many individuals in ages 1 and 2, and none in the 0 age class. The absence of animals that "an- chor" the beginning of the curve could result in low estimates of ^ and ?Q. Additionally, Borobia ( 1989) examined individuals from different locations along the distribution of the species and thus did not take into consideration possible geographi- cal variations. The results obtained by Schmiegelow (1990) are similar to those of our study, probably because both of them used animals from the same region. Ramos et al. (2000) analyzed the growth of S. guian- ensis using the Gompertz growth model and obtained an asymptotic length (191.7 cm) which was much greater than that obtained in our study and in previous studies (Borobia, 1989; Schmiegelow, 1990) (Table 1). This differ- ence could be due to 1) the small number of individuals older than 12 years (/!=3) in their sample; or 2) a difference in asymptotic lengths between southeastern and southern Brazil populations. Similar differences have been observed between asymptotic lengths of Pontoporia hlainvillei from Rio de Janeiro (southeastern Brazil) and Sao Paulo and Parana (same area of the present study), where larger in- dividuals were found in Rio de Janeiro (Ramos et al., 2000; Rosas, 2000). Therefore, it is possible that environmental variables could be responsible for larger sizes in the area studied by Ramos et al. (2000), both for S. guianensis and for f! hlainvillei. Although no significant difference was observed in the asymptotic length between adult males and females, the differentiated growth in time between the two sexes is probably responsible for the difference observed in the weight-length relationship. In most species, the length exponent (0) of the weight- length relationship is usually close to 3 (Santos, 1978). The estimated values of this exponent for the estuarine dolphin (3.2 for males and 2.6 for females) suggest that the longi- tudinal and transversal body growth in this species follows a similar pattern. Our results suggest that it is important to study growth by analyzing the sexes separately, because there may be dif- ferential growth between the sexes before the adult age. Rosas et aL: Age and growth of Sotalia guianensis 383 Acknowledgments We sincerely thank the fishermen of Vila da Barra do Superagtii and Ilha das Pe^as (Parana coast) for the infor- mation they provided and their help in collecting the inci- dentally caught dolphins. We thank Fundagao O Boticario de Prote^ao a Natureza and the MacArthur Foundation for financial support, and IBAMA/PR, especially Guadalupe Vivekananda, head of the National Park of Superagtii. We also thank Maria Cristina Pinedo, who allowed us to use her laboratory and equipment for the age estimations, and Kesa K. Lehti, who translated the manuscript from Portuguese into English. Renata Ramos and an anonymous referee provided critical and insightful comments on the manu- script. Coordenagao de Pessoal de Nivel Superior (CAPES) provided a fellowship to the senior author This study is part of a dissertation presented by Fernando C. Weber Rosas, sub- mitted in partial fulfillment for a Ph.D. degree in Zoology at the Universidade Federal do Parana, Curitiba, Brazil. Literature cited Borobia, M. 1989. Distribution and morphometries of South American dolphins of the genus Sofa/(a. MSc. thesis, 81 p. McDon- ald College, McGill University. Montreal, Quebec, Canada. Boyde, A. 1980. Histological studies of dental tissues of Odontocetes. In Age determination of toothed whales and sirenians (W. F. Perrin and A. C. Myrick, eds.), p. 65-87 Rep. Int. Whal. Comm., Special Issue 3. Cockcroft, V. G., and G. J. Ross. 1990. Age, growth, and reproduction of bottlenose dolphins Tursiops truncatus from the east coast of Southern Africa. Fish. Bull. 88:289-302. da Silva, V. M. F, and R. C. Best. 1996. Sotalia fluviatilis. Mammalian Species 527:1-7. Ferrero, R. C, and W. A. Walker 1993. Growth and reproduction of the northern right whale dolphin, Lissodelphis borealis, in the offshore water of the North Pacific Ocean. Can. J. Zool. 71:2335-2344. 1999. Age, growth and reproductive patterns of Dall's por- poise (Phocoenoides dalli) in the Central North Pacific Ocean. Mar Mamm. Sci. 15(2):273-313. Hohn, A. A. 1990. Reading between the lines: Analysis of age estimation in dolphins. In The Bottlenose dolphin (S. Leatherwood and R. R. Reeves, eds.), p. 575-585. Academic Press, New York, NY. Hohn, A. A.; M. D. Scott, R. S. Wells, J. C. Sweeney, and A. B. Irvine. 1989. Growth layers in teeth from known-age, free-ranging bottlenose dolphins. Mar Mamm. Sci. 5 (41:315-342. Hui,C.A. 1980. Variability of dentine deposition in Tursiops truncatus. Can. J. Fish. Aquat. Sci. 37:712-716. Klevezal, G. A. 1980. Layers in the hard tissues of mammals as a record of growth rhythms of individuals. In Age determination of toothed whales and sirenians ( W F. Perrin and A. C. Myrick, eds.), p. 89-94. Rep. Int. Whal. Comm. Special Issue 3. 1996. Recording structures of mammals. Determination of age and reconstruction of life history, 274 p. A.A. Balkema, Rotterdam, Netherlands. Molina, D. M., and J. A. Oporto. 1993. Comparative study of dentine staining techniques to estimate age in the Chilean Dolphin, Cephalorhynchus eutropia (Gray 1846). Aquat. Mamm. 19 ( 1 ):45-48. Monteiro-Filho, E. L. A., L. R. Monteiro, and S. F dos Reis. 2002. Skull shape and size divergence in the dolphins of the genus Sotalia : a tridimensional morphometric analysis. J. Mamm. 83(1):125-134. Norris, K. S. 1961. Standardized methods for measuring and recording data on the smaller cetaceans. J. Mamm. 42 (4):471-476. Perrin, W. F, J. M. Coe, and J. R. Zweifel. 1976. Growth and reproduction of the spotted porpoise, Stenella attenuata, in the offshore eastern tropical Pacific. Fish. Bull. 74(2 ):229-269. Pinedo, M. C, and A. A. Hohn. 2000. Growth layer patterns in teeth from the franciscana, Pontoporia blainvillei: developing a model for precision in age estimation. Mar Mamm. Sci. 16 (l):l-27. Ramos, R. M.A. 1997. Determina?ao de idade e biologia reprodutiva de Pon- toporia blainvillei e da forma marinha de Sotalia fluviatilis (Cetacea:Pontoporiidae e Delphinidae) no norte do Rio de Janeiro. M.Sc. thesis, 95 p. Universidade Estadual do Norte Fluminense. Campos dos Goytacazes, Rio de Janeiro, Brasil. Ramos, R. M. A., A. P M. Di Beneditto, and N. R. W. Lima. 2000. Growth parameters of Pontoporia blainvillei and Sota- lia fluviatilis (Cetacea) in northern Rio de Janeiro, Brazil. Aquat. Mamm. 26 (l):65-75. Rosas, F C. W. 2000. Interagoes com a pesca, mortalidade, idade, reprodu9ao e crescimento de Sotalia guianensis e Pontoporia blainvil- lei (Cetacea, Delphinidae e Pontoporiidae) no literal sul do Estado de Sao Paulo e literal do Estado do Parana, Brasil. Ph.D. diss., 145 p. Universidade Federal do Parana. Curi- tiba, PR, Brasil. Rosas, F C. W, and E. L. A. Monteiro-Filho. 2002. Reproduction of the estuarine dolphin (Sotalia guia- nensis) on the coast of Parana, southern Brazil. J. Mamm. 83(2):507-515. Santos, E. P. 1978. Dinamica de popula?6es aplicada a pesca e piscicul- tura, 129 p. Editora de Humanismo, Ciencia e Tecnologia "HUCITEC" Ltda, Sao Paulo. Scheffer, V. B., and A. C. Myrick. 1980. A review of studies to 1970 of growth layers in the teeth of marine mammals. In Age determination of toothed whales and sirenians (W. F Perrin and A. C. Myrick, eds.), p. 51-63. Rep. Int. Whal. Comm., Special Issue 3. Schmiegelow, J. M. M. 1990. Estudo sobre cetaceos odontocetes encontrados em praias da regiao entre Iguape (SP) e Baia de Paranagua (PR) (24°42'S-25°28'S) com especial referenda a Sotalia fluviatilis (Gervais, 1853) (Delphinidae). M.Sc. thesis, 149 p. Univer- sidade de Sao Paulo, Institute Oceanografico, Sao Paulo. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J Fish. Aquat. Sci. 38:1128-1140. Watson, L. 1988. Whales of the world. A handbook and field guide to all the living species of whales, dolphins and porpoises, 302 p. Hutchinson, London. 384 Abstract— Offshore winter-spawned fishes dominate the nekton of south- eastern United States estuaries. Their juveniles reside for several months in shallow, soft bottom estuarine creeks and bays called primary nursery areas. Despite similarity in many nursery char- acteristics, there is, between and within species, variability in the occupation of these habitats. Whether all occupied habitats are equally valuable to indi- viduals of the same species or whether most recruiting juveniles end up in the best habitats is not known. If nursery quality varies, then factors controlling variation in pre-settlement fish distribu- tion are important to year-class success. If nursery areas have similar values, interannual variation in distribution across nursery creeks should have less effect on population sizes or production. I used early nursery period age-specific growth and mortality rates of spot {Leios- tomus xanthurus) and Atlantic croaker iMicropogonias undulatus) — two domi- nant estuarine fishes — to assess relative habitat quality across a wide variety of nursery conditions, assuming that fish growth and mortality rates were direct reflections of overall physical and biologi- cal conditions in the nurseries. I tested the hypothesis that habitat quality varies for these fishes by comparing growth and mortality rates and distribution patterns across a wide range of typical nursery habitats at extreme ends of two systems. Juvenile spot and Atlantic croaker were collected from 10 creeks in the Cape Fear River estuary and from 18 creeks in the Pamlico Sound system. North Carolina, during the 1987 recruitment season (mid-March-mid-June). Sampled creeks were similar in size, depth, and substrates but varied in salinities, tidal regimes, and distances from inlets. Spot was widely distributed among all the estuarine creeks, but was least abundant in the creeks in middle reaches of both systems. Atlantic croaker occurred in the greatest abundance in oligohaline creeks of both systems. Instantaneous growth rates derived from daily otolith ages were generally similar for all creeks and for both species, except that spot exhibited a short-term growth depression in the upriver Pamlico system creeks — perhaps the result of the long migration distance of this species to this area. Spot and Atlantic croaker from upriver oligohaline creeks exhibited lower mortality rates than fish from downstream polyhaline creeks. These results indicated that even though growth was similar at the ends of the estuaries, the upstream habitats provided conditions that may optimize fitness through improved survival. Manuscript accepted 25 October 2002. Manusript received 31 December 2002 at NMFS Scientific Publications Office. Fish. Bull. 101:384-404 (2003). The relative value of different estuarine nursery areas in North Carolina for transient juvenile marine fishes Steve W. Ross NC National Estuarine Research Reserve 5600 Marvin Moss Ln. Wilmington, North Carolina 28409 E-mail address: rosss(guncwil.edij Offshore winter-spawned (OWS) fishes are a major component of the nekton of southeastern United States and Gulf of Mexico estuaries. Their larvae migrate across the shelf, enter estuaries, and the majority of juveniles reside for several months in shallow, soft bottom estuarine creeks and bays called primary nursery areas (PNAs). Very high concentrations of fishes in these PNAs suggest that they are valuable habitats, perhaps because they are good sources of food and shelter (Boesch and Turner, 1984; Mclvor and Odum, 1988; Miltner et al., 1995). Despite similarity in some PNA physical characteristics, there is vari- ability in habitats occupied (between and within species), especially with regard to salinity, tidal influence, acces- sibility (i.e. distance from inlets), and, perhaps, food and predator regimes (Weinstein, 1979; Ross and Epperly, 1985). Assessing the relative value of all PNA habitats to individuals of the same species is increasingly important (Weinstein, 1982; Sogard, 1992; Guin- don and Miller, 1995; Beck et al., 2001). If PNA value varies, do most of the recruiting juveniles end up in the best habitats (Thresher, 1985)? Understand- ing variation in habitat quality during a major early life history phase should yield insight into causes of variability in year-class strength, particularly if juve- nile fish distributions vary interannu- ally. If PNA quality varies, then factors controlling variation in presettlement distribution are important to year class success because animals could be trans- ported to habitats of unpredictable quality. If nursery areas have similar value, interannual variations in dis- tribution across nursery creeks should have less effect on ultimate population sizes or production. General estuarine distributions of two dominant OWS fishes, spot (Leio- stomus xanthurus) and A.i\a.ntic croaker (Micropogonias undulatus), exhibit con- sistent patterns throughout their rang- es. Juvenile Atlantic croaker routinely concentrate in oligohaline creeks or bays (Weinstein, 1979; Mercer, 1987a) — a pattern that suggests that the upstream regions are most valuable to this species. Spot, however, are more ubiquitously and variably distributed through the shallow PNAs (Ross and Epperly, 1985; Mercer, 1987b), perhaps indicating less dependence on a particu- lar estuarine region. Despite these gen- eralities, both species can be present in large numbers in almost any estuarine creek or bay over the full salinity range (e.g. Nelson et al., 1991). In general, ju- veniles of both species seem to avoid (or are unsuccessful in) more open water areas of estuaries during the early part of the nursery period. The main purpose of this paper is to assess relative habitat value for two dominant members of the OWS fish group, spot and Atlantic croaker, across a wide variety of North Carolina PNA conditions. I assumed that fish growth and mortality rates were direct reflec- tions (integrators) of overall physical and biological conditions in PNA habi- tats. Therefore, 1 used early nursery period age-specific growth and mortal- ity rates of spot and Atlantic croaker, in addition to distribution data, to assess relative habitat quality, testing the hy- pothesis that habitat quality varies for these fishes across a broad range of typ- ical PNAs in two very different estua- rine systems. Growth and mortality can be influenced by fish density; however, Ross (1992) found similar growth and mortality rates for spot and Atlantic Ross: Relative value of different estuarine nursery areas for juvenile marine fishes in North Carolina 385 76' '30' Upper TS'SO" 35-30' PAMLICO f SOUND y 35°00: 78°0(t 77"5a' Figure 1 Pamlico Sound (A) and Cape Fear Estuary (B) in coastal North Carolina. Gen- eral areas (e.g. upper, middle, lower) and sampling locations during March-June 1987 (solid dots) are labeled in the enlargements. Numbers correspond to station descriptions in Table 1. croaker across wide ranges of densities in these systems. The similarity in these rates imply that PNAs were below carrying capacities. Thus, I did not consider density as a variable affecting growth or mortality for the following comparisons of PNA quahty. Methods Study area To encompass the greatest variability possible in estuarine habitats, I sampled nursery creeks in two widely separated, geophysically different North Carolina estuarine systems: 1) Pamlico Sound and River and 2) the Cape Fear River ( Fig. 1 ). Each system was partitioned into general areas (e.g. upper, middle, lower) and stations were selected to repre- sent these areas. Stations were located in creeks through- out both systems that previous sampling (Weinstein, 1979; Ross and Epperly, 1985; NC Division of Marine Fisheries^ indicated were consistently productive for juvenile marine fishes during the spring-summer season. All creeks were similar in depth, size, and sediment type. The greatest phys- ical differences between stations were the salinities, tidal regimes, and distances from the nearest inlets (Table 1). Pamlico Sound is a shallow lagoon estuary whose hydrography is controlled by wind (Giese et al., 1979; Pietrafesa et al., 1986a; Pietrafesa and Janowitz, 1988). NC (North Carolina) Division of Marine Fisheries. Unpubl. data. Program 120 Nursery Area Survey, P.O. Box 769, More- head City, NC 28557. 386 Fishery Bulletin 101(2) Table 1 Distances to nearest inlets (D in km), salinity (%<:) ranges and means, mean depths (m). and sediments of stations sampled from mid-March through mid-June 1987 in the Pamlico Sound and Cape Fear systems. Sediment symbols are m = mud. fs = fine sand. g = grass, and fs-m = fine sand and mud. Area D Salinity range (mean) Mean depth Sediment Pamlico Sound Lower "Coastguard" Creek 3 15.5-23.0(18.9) 0.9 m "Doctors" Creek 4 15.0-24.5(18.6) 0.7 m "Tom Bragg Slough" 7 13.0-22.0(17.4) 0.9 fs-m Royal Point Bay 8 17.0-20.4(18.6) 0.7 fs Mid-Lower Oyster Creek 26 11.0-15.0(13.5) 0.9 m Southwest Prong 33 13.5-23.0(17.8) 1.0 m' Merkle Hammock Creek 41 11.0-14.5(15.1) 0.9 fs-g' Codduggen Creek 46 9.0-14.5(12.3) 0.9 fs-m' Middle Caffee Creek 46 10.8-18.5(13.4) 1.8 m Oyster Creek 48 10.8-17.8(13.2) 1.2 m' unnamed Creek 51 10.9-18.0(13.7) 1.6 fs-m "Swan" Creek 60 10.0-15.9(12.4) 1.1 m Tooley Cr 61 10.2-17.1(12.5) 1.3 m^ Head Rose Bay 64 10.0-13.3(11.1) 2.5 m Upper Mallard Creek 92 1.3-7.2(3.5) 1.2 m Flatty Creek 99 0.9-4.0(2.6) 1.0 m Broad Creek 101 0.3-8.7 (2.4) 1.4 m Little Creek 104 0.6-3.9(2.3) 1.0 m Cape Fear estuary Lower Molasses Creek 4 15.4-29.0(21.5) 1.0 m Piney Pt. Creek 6 15.0-29.9(22.5) 0.6 m Dennis Creek 7 11.8-31.0(21.7) 1.0 m-fs Dutchman Creek 7 12.6-24.6(17.7) 0.6 m Middle Town Creek 29 0.0-10.2(3.7) 2.0 m Mott Creek 29 0.0-13.4(5.6) 1.0 m Upper Jackeys Creek 36 0.0-6.9(1.7) 0.5 m Toomers Creek 44 0.0-2.3(0.3) 3.0 m Horseshoe Bend 43 0.0-2.3(0.4) 0.6 m Smith Creek 44 0.0-4.8(1.5) 1.3 111 ' See Ross and Epperly (1985) for additiona sediment data. Eighteen stations were located in creeks in four areas along an approximately 100-km transect from Oeracoke Inlet to the upper Pamlico River (Fig. 1). Four polyhaline stations were located on Portsmouth Island (lower area). Water depths there were largely controlled by semidiurnal lunar tides (range usually <0.7 m, Giese et al., 1979); however, on one occasion I observed that northerly winds (>37 km/h) moved large quantities of water into these creeks. The mid- lower area consisted of four stations on Cedar Island, which exhibited less depth variation (dampened lunar tides) than creeks on Portsmouth Island. Six creeks were sampled in the middle area: three each in Rose and Swanquarter bays. Tidal influence was negligible here (Pietrafesa et al., 1986a). The creeks in the above three areas were largely surrounded by Spartino and Juricus marsh grasses. The upper area was represented by four creeks (oligohaline or freshwater) surrounded by a mixture of woodlands and patches of marsh. Water levels and currents here were almost entirely controlled by winds or river flow (or both) (Hobbie, 1970; Pietrafesa et al., 1986a). The Cape Fear River is more typical (compared to the Pamlico Sound system) of United States East and Gulf Ross: Relative value of different estuanne nursery areas for luvenile marine fishes in Nortfi Carolina 387 coasts estuaries (i.e. a drowned river valley). Diurnal lu- nar tides (average range about 1.5 m) were a dominant feature throughout the study area (Welch and Parker, 1979; Pietrafesa and Janowitz, 1988). Ten stations were located along a 50-km transect of the Cape Fear system (Fig. 1). In the lower estuary, four polyhaline creeks were sampled on the west side of the inlet (Oak Island). Two creeks were sampled on opposite sides of the middle of the estuary and four oligohaline creeks were sampled in the upper estuary near Wilmington. All stations in this system were sur- rounded by Spartina marshes. Field sampling All stations were sampled during daylight with two one- minute tows (68.6 m each) of a small-mesh trawl (3.2-m headrope length, 6.4-mm bar mesh wings and body, 3.2-mm tail bag mesh). Catches from the two tows were combined for the station sample. Surface and bottom salinities (near- est %c) and water temperatures ( nearest °C ) were recorded after each sample. Mean salinities and temperatures for each area were analyzed for differences by using <-tests for all possible combinations of area pairs. Sampling was designed to provide biological data during the time of early residency in the nursery creeks, but before significant emigration. Most recruitment of young juvenile fishes into these creeks has ended by late-April, and some fishes begin to emigrate by June— July (Weinstein, 1979; Ross and Epperly, 1985; author's pers. obs.). To minimize the influence of emigration on the calculation of growth and mortality rates, sampling occurred during seven periods, every other week from mid-March through mid-June 1987 (about 14 d between samples). Synoptic samples over this large region were generated by assigning areas to four crews for trawling during the same period of each sample week. Newly recruiting OWS juvenile fishes of the 1987 year class were sorted from the catches and preserved in the field in 100% ethyl alcohol. About one month after col- lection the fishes were identified, counted, and standard lengths (SL) were measured to the nearest mm. Analyses were limited to spot and Atlantic croaker, the two most abundant species. Catch per unit of effort (CPUE) was calculated by dividing the total number of individuals of a species captured by the number of trawl tows in a time period or area. Subsamples of these fishes representing several collection dates and all areas were measured for SL, blotted, and weighed to the nearest 0.01 g and were used to develop a weight-length relationship using linear regression. Differences in weight-length relationships be- tween areas were assessed by using analysis of covariance (covariate=logSL) in a general linear model procedure (SAS Institute, 1988). Otolith aging Subsamples for aging were randomly selected from early (early and mid- April) and late (mid and late May) dates and from downstream (lower) and upstream (upper) areas in each system. Sagittae were removed from these fishes, mounted on microscope slides with thermoplastic cement, and polished (often on both sides) until thin sections were obtained. Otoliths were viewed with oil immersion and transmitted polarized light at magnifications between 500 and 625x, and images were projected through a video system to a screen. Rings, presumed to be daily, were counted. The formation of daily rings has either been validated (Peters et al., 1978; Baldevarona, 1987; Siegfried and Weinstein, 1989) or assumed (Warien and Chester, 1985; Cowan, 1988) for the two species in the size ranges used here. Even so, the counted rings need not be deposited daily for growth rate comparisons, nor is it necessary to know their periodicity. It is required that groups offish being compared exhibit the same ring formation periodicity over the time and space of the comparison. Spot and Atlantic croaker do not form growth rings until after yolksac absorption, about four to five days after spawning (Peters etal., 1978; Warien, 1980). Therefore, to estimate actual ages for mortality calcula- tions, five days were added to the ring counts. Although it seems reasonable to assume that juvenile spot and Atlantic croaker form daily sagittae rings, the precision (repeatability) of ring counts and the ability to identify daily rings needs addressing. Before aging the samples used in this study, I examined several hundred spot and Atlantic croaker otoliths. Counts by myself and 2-3 other otolith readers were compared. Our ring identi- fications were compared to samples of known age spot and Atlantic croaker provided by the National Marine Fisheries Service (Warien^). These preliminary samples were used as a training device to ensure that daily rings were ac- curately identified and were not confused with shadows or subdaily rings. Subdaily rings may not even be resolvable at the magnifications (<625x) used in the present study (Campana et al., 1987; Isely^). After aging the samples used in this study, I re-aged a random selection of spot (without knowledge of previous age assignments) and obtained a mean difference in counts of 2.85 (SD=2.03,n=27). Because Atlantic croaker otolith rings were usually easier to count, I assumed that the above count difference was generally similar for this species. I assumed that the ages reported here had a count precision of ±3 days. I also assumed that any aging errors were randomly distributed throughout the samples and were not spatially or temporally biased. Growth and mortality Linear regression of the form Log,gSL = 6 -i- miage) was used to model growth. The slope of this line, m, is the instantaneous daily growth rate. Differences in growth rates between areas were assessed by using analysis of covariance (covariate=age) in a general linear model pro- cedure (SAS Institute, 1988). Absolute and relative daily growth rates were calculated by using values predicted with the age-SL regression equation (Ricker, 1975). For each sampling date, mean SLs were compared between all 2 Warien, S. M. 1989. Personal commun. National Marine Fisheries Service Beaufort Lab, Beaufort, NC 28516. ' Isely, J. 1989. Personal commun. Zoology Dept., NC State Univ., Raleigh, NC 27695. 388 Fishery Bulletin 101(2) 30 25 i. 20 2- c 0.05) between areas. Mean tempera- tures throughout the Pamlico system were not significantly different (P>0.05), except that the lower area was cooler than the others (P<0.05). Comparisons between systems revealed no significant differences (P>0.05) between middle or upper area temperatures. Lower Cape Fear creeks were Ross: Relative value of different estuanne nursery areas for luvenile marine fishes in Nortfi Carolina 389 o PAMLICO SOUND upper CAPE FEAR upper Figure 3 Mean catch per unit of effort for spot in general areas of the Cape Fear and PamHco Sound, NC, estuaries by sampling week, March^une 1987. Vertical bars represent plus and minus one standard error of the mean. The -V-axis abbreviations (from left to right) stand for March, April, April, April, May. May, June. significantly (P<0.0.5) warmer than those of the lower Pamlico. The cooler temperatures in the lower Pamlico may have resulted from sampling there at earlier times of the day or on different days of the sampling week. Salinity was more variable than temperature, particu- larly within the mesohaline and polyhaline areas (Fig. 2). Within both the Pamlico and Cape Fear systems all areas exhibited significantly different «-test, P<0.05) bottom salinities from each other. As expected, the Cape Fear estuary, with its larger, more channeled river flow and obvious tidal effects, was a more variable system than the Pamlico system. During this study mean salinities within the lower and middle areas of the Cape Fear system varied over a range of 11.8%f and 14.2%f, respectively, whereas mean salinities in all other areas (including the Pamlico) varied over a range less than 9%f . The lower Cape Fear creeks had significantly higher (P<0.05) salinities than those of the lower Pamlico; however, the middle and upper Pamlico areas had significantly higher (P<0.05) salinities than their counterparts in the Cape Fear, even though the upper Pamlico area was twice as far from an inlet as the upper Cape Fear. Salinities declined rapidly in the Cape Fear with increasing distance from the inlet; however, this relationship was more variable in the Pamlico System (Fig. 2). In both systems, overall mean salinity (S) was ac- curately predicted by distance (D, in km) from the inlet: Cape Fear: S=23.5 - 0.55(D), r2=0.95, n=10 and Pamlico: S=20.4 - 0.17(D), r-=0.92, n = l8. Leiostomus xanthurus Distribution Spot was the more widely distributed of the two species (Figs. 3-5). At the earliest sample date, small numbers of spot had accumulated in all areas of the Pam- lico system (Fig. 3). Peak abundance was observed in the lower and mid-lower areas by the second sample date and in the middle and upper areas by the third sample date. Although more spot were collected in the lower Pamlico, numerical differences between areas were not extreme (Figs. 3 and 4). Overall, the least numbers of spot occurred in the middle region (Figs. 3 and 4). In the Cape Fear system, overall spot abundance was similar between the upper and lower regions, and, as in the Pamlico, the least numbers were collected from the middle 390 Fishery Bulletin 101(2) 20 10 LOWER Ul 18 Mar n=123 MID-LOWER Li n=392 ~1 1 \ 1 1 1 1 r n=785J MIDDLE UPPER n = 1040 Standard length (mm) Figure 4 Spot length frequencies in general areas of the Pamlico Sound, NC, estuary by sampling week, March-.June 1987, Solid dots represent mean SL. Ross: Relative value of different estuarine nursery areas for luvenile marine fishes in North Carolina 391 20 10- 0 30 20- 10- 0 30 _ 20 -.O O ~ 10 o a) 0- Z 20- c Q) y 10- «) a. 0 20-1 10- 0 20 10- 0 20- 10- 0 LOWER 18Mar n=544 1 Apr n.ll05 15 Apr n-3515 29ADr n=1337 CAPE FEAR ESTUARY MIDDLE —I r -T soj. J. J. - T . . r T . , , n=2 13May n=1052 27May n=1734 n=970 n=520 n=230 n=253 n=50 i n=18 Am . UPPER I n=1 — 1 1 1 r i^ X n=1938 n=2705 Jl. n=109 n = 250 n=424 • n=2012 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Standard length (mm) Figure 5 Spot length frequencies in general areas of the Cape Fear, NC. estuary by sampling week, March-June 1987. Solid dots represent mean SL. area creeks (Figs. 3 and 5). Initial recruitment was high in the lower Cape Fear area, whereas it lagged behind in the middle and upper areas (Fig. 3). Nevertheless, all Cape Fear areas reached peak numbers by the second or third sampling dates (Fig. 3). Size distributions The bulk of the spot year class recruited to all creeks from mid-March to early April; however, some small fish (<25 mm) continued to enter the creeks through the end of May (Figs. 4 and 5). The smallest spot captured in both systems were always between 13 and 15 mm SL (Figs. 4 and 5). Because the trawl can collect spot and Atlantic croaker at least to 10 mm SL , 13-15 mm prob- ably represented the smallest recruitment sizes of spot to nursery habitats. Differences in mean spot SLs between areas steadily in- creased during the study from <1 mm (through mid- April) to 4.3 mm in the Cape Fear and from 1 to 6.2 mm in the Pamlico. Although mean spot SLs exhibited the small- est variations between areas in the Cape Fear system (Fig. 5), they were significantly different (paired ^tests, P<0.05) between areas on most sampling dates. No area in the Cape Fear had consistently larger or smaller mean spot SLs. After mid-April mean SLs of spot from all four Pamlico areas were larger than those from the Cape Fear. Within the Pamlico system (Fig. 4) spot in the mid-lower area had significantly larger (P<0.05) mean SL (except sampling weeks four and six), and fish in the middle area were always significantly smaller (P<0.05) than those in the other three areas. Growth Estimated spot ages ranged from 61 to 157 days (17-35 mm SL, /!=379). All of the age-SL relationships used to assess growth rates (Fig. 6) were highly significant (P<0.0001). Regression residuals were evenly distributed around zero, indicating that the exponential growth model 392 Fishery Bulletin 101(2) 40 30- (1) / / / PAMLICO / Upper / / '' / '' / / y y y y y y y / ' / ^ / y / y y y y y y -^,.'' Lo9,oSl=0.935 + 0.004(Age) T 1 1 1 1 — r2=0.93. n = 96 r 1 1 r Loa,QSL=0.875 + 0.005(A8e) r2 = 0.91,n.96 -I 1 1 1 1 f— 40- 30 20 10- CAPE FEAR Upper " Lo9|„Sl = 0.856»0.005(A9o) r^ = 0.94,n=91 CAPE FEAR Lower log,QSL=0.829+0.005(Agel r2 = 0.90,n = 96 ■l 1 r— -I 1 1 f- 60 80 100 120 140 160 60 80 100 120 140 160 Age (days) Figure 6 Spot growth rate curves (solid lines) based on otolith daily ages for upper and lower areas of the Pamlico Sound and Cape Fear, NC estuaries. Dotted lines represent 95 percent confidence intervals. (logj^SL = 6 -t- mkage)) was appropriate. Instantaneous daily growth rates (slopes of the regressions) were similar (analysis of covariance, P>0.05) between upper and lower areas in the Cape Fear and the lower Pamlico (Fig. 6). Therefore, a combined age-SL regression for all spot in the upper and lower Cape Fear estuary and the lower Pamlico area was developed: logipSL = 0.861 + 0.0048(a^e), r2=0.90, n=2S3. Analysis of covariance indicated that spot from the upper Pamlico region exhibited significantly slower (P>0.05) overall growth rates (Fig. 6) than fish from the other three areas. Age-specific absolute and relative growth of spot was predicted from the age-SL regression for the upper and lower ("ape Fear and lower Pamlico combined and the up- per Pamlico (Table 2). Predicted absolute growth rates in the Cape Fear and lower Pamlico areas increased from 0.16 mm/d between 60 and 65 days of age to 0.43 mm/d between ages 150 and 155 days of age, and the largest increase in absolute growth occurred between ages 95 and 105 days of age (Table 2). Relative growth remained constant around 1.13-1.14 %/d SL over the whole age range examined (Table 2). Although predicted sizes at ages were larger in the upper Pamlico area than those of the other three areas, the absolute growth rates were lower, increasing from 0.16 mm/d between ages 60 and 65 d to 0.39 mm/d between ages 150 and 155 d (Table 2). Absolute growth rates in this area also exhibited the largest increases around 100-105 days. Relative growth rates in the upper Pamlico were lower than in the other areas and averaged 1.01 9t/d SL (Table 2). The ages when absolute growth in all areas was greatest (95-105 d) translated to SL ranges around 21-23 mm. This SL range dominated the length frequencies in all areas during the first two weeks of April (Figs. 4 and 5). Water temperatures were steadily increasing in all areas prior to mid-April (Fig. 2). (Jrowth was also compared by using weight-length rela- tionships. These relationships for spot were highly signifi- cant (analysis of covariance, P<0.0001) and took the usual Ross: Relative value of different estuarine nursery areas for |uvenile marine fisfies in Nortfi Carolina 393 Table 2 Predicted age-specific mean standard lengths (SL), absolute (mm/d), and relative (%/d SL) grow th rates for spot from the upper and lower Cape Fear and lower Pamlico combined (CFR f LPAM land the upper Pamlico area (UPPAM). Age CFR + LPAM UPPAM Mean Absolute Relative Mean Absolute Relative (days) SL growth rate growth rate SL growth rate growth rate 60 14.09 15.60 65 14.89 0.16 1.14 16.39 0.16 1.01 70 15.74 17.22 75 16.63 0.18 1.13 18.09 0.17 1.01 80 17.58 19.01 85 18.58 0.20 1.14 19.98 0.19 0.99 90 19.63 20.99 95 20.75 0.22 1.14 22.05 0.21 1.00 100 21.93 23.17 105 23.17 0.25 1.13 24.35 0.24 1.04 110 24.49 25.59 115 25.88 0.28 1.14 26.88 0.26 1.02 120 27.35 28.24 125 28.91 0.31 1.13 29.68 0.29 1.03 130 30.55 31.19 135 32.28 0.35 1.15 32.77 0.32 0.99 140 34.12 34.43 145 36.06 0.39 1.14 36.18 0.35 1.02 150 38.11 38.02 155 40.27 0.43 1.13 39.95 0.39 1.03 Table 3 Weight-standard length (W-SL) relationships for spot and Atlantic croaker from all areas of the Cape Fear and Pamlico systems, March-June 1987. Area Formula r- n Spot Cape Fear Upper W= 10-« 2.5,5^4 02) 0.96 100 Cape Fear Middle W = 10-» 90(SL3 05) 0.97 134 Cape Fear Lower W= 10-5 00(SL3 12) 0.97 212 Pamlico Upper W = 10-6 "^CSZ-^-S') 0.98 456 Pamlico Middle W = 10-5 'stSL^es) 0.97 252 Pamlico Mid-Lower W = 10-« 12,Si,3.89) 0.98 246 Pamlico Lower W = 10-« 22(S/,3 95) 0.97 250 Atlantic croaker Cape Fear Upper W = 10-5 80(SL3,64) 0.97 138 Cape Fear Middle W = 10-5 30(SZ,3 31) 0.86 118 Cape Fear Lower W = 10-5 64(SL3«) 0.93 98 Pamlico Upper W = 10-5 55(SL3 «) 0.96 188 Pamlico Lower W = 10-»85{SL2 ^SS) 0.88 46 curvilinear form (Table 3, Fig. 7). In the Pamlico system dif- ferences between areas were not large; however, middle area spot had significantly (P<0.05) lower weights per length, es- pecially in larger individuals (Fig. 7). In the Cape Fear sys- tem, spot in the upper area had significantly (P<0.05) larger weights per length than those from the other two areas. Mortality Spot mortality rates were based on individuals aged >85 days. All regression slopes describing the declin- ing numbers of spot with increasing ages were significantly different from zero (P<0.0001). Instantaneous mortality rates over this time period ranged from 0.037 to 0.066 (Fig. 8). Analysis of covariance indicated that within each 394 Fishery Bulletin 101(2) 3 PAMLICO , 2- // ] . X'Middl* ai 2 0- CD CO 4- 1 ■ CAPE FEAR / 3 Upp«r / 2- 1- / y / y 0- -^^^^^""^ 15 25 35 45 Standard lengtti (mm) Figure 7 Spot weight-length relationships for the Pamlico and Cape Fear, NC, estuaries, March-June 1987. The line for the area that exhibited a significant difference was labeled (i.e. Pamlico Middle area. dashed line, was different and Cape Fear Upper area, solid line was different); the remaining line in each panel represented the relationship for all other areas combined. system, upper area spot displayed significantly (P<0.05) lower instantaneous mortality rates than did fish from the lower estuaries, especially in Cape Fear Lower Cape Fear spot exhibited a statistically similar (P>0.05) instan- taneous mortality rate to those in the lower and upper Pamlico creeks. Daily mortality rates during the present study were 4.97 %/d in the upper Pamlico, 6.39 %/d in the lower Pamlico, 3.63 %IA in the upper Cape Fear, and 6.01 '7r/d in the lower Cape Fear Micropogonias undulatus Distribution Atlantic croaker distributions in both sys- tems were skewed toward upstream, oligohaline creeks (Fig. 9-11). In the Pamlico system almost no Atlantic croaker were collected in the lower or middle areas (Fig. 9). Atlantic croaker recruitment in the Pamlico lagged behind the ("ape Fear in abundance and timing (Fig. 9), and peak densities occurred throughout the Pamlico near the end of the sampling. Patterns of Atlantic croaker recruitment were like those of spot in the upper and middle Cape Fear. Like spot, most of the Atlantic croaker year class had recruited to these ar- eas by mid-April, although small Atlantic croaker (<20 mm) continued to colonize these creeks through mid-late May (Fig. 10). Also, peak abundance was reached in the middle and upper Cape Fear during the same weeks as those for spot (2nd and 3rd, respectively) (Fig. 9). Except for the larger CPUE in mid-March, Atlantic croaker recruitment in the lower Cape Fear was similar to that of most Pamlico system creeks (Fig. 9). Atlantic croaker entering PNAs from mid-March through late April in both systems appeared to bypass lower and middle area nursery creeks (unlike spot) to a greater extent than fish recruiting after April (Fig. 9). Late recruitment of small Atlantic croaker was especially apparent in the mid-lower Pamlico (Fig. 11). Size distribution Atlantic croaker initially collected in the Cape Fear creeks through the first week of April spanned a size range of 11-23 mm (Fig. 10, 15.4 mm SL mean), and mean sizes were not significantly different (paired /-test, P>0.05) between areas in the first or second sampling week. After this time, Atlantic croaker from the middle Ross: Relative value of different estuanne nursery areas for juvenile manne fishes in North Carolina 395 PAMLICO PAMLICO 10- Upper "-v^ Lower 8- 6- 4- "^^^^%^--~ i ^^\^ ^*^ 2- LoggN=12.40-0.051(Age) ""-~- ^^^-^^ "" r2 = 0.86,n=55 ""~--,^ ^^^ LoggN=14.44-0.066(AgeP-^ ^\^^ "-^ r2 = 0.87,n=50 ""^^ ^^-.^^ E 0- e J C "^^^ - ^"~-^. ^ 10- ^ CAPE FEAR CAPE FEAR 8- "~-^^^ Upper ^^^ ^•-^^ Lower 6- 4- ""^^^S^ ~~\$C\- 2- LoggN=9,79-0.037(Age) ""--^^ LogeN = 13.34-0.062(AgeP-. 0- r2 = 0.60,n=42 r2 = 0.76,n=35 60 80 100 120 140 160 180 200 220 60 80 100 120 140 160 180 200 220 Age (days) Figure 8 Spot mortality rate curves (solid lines) based on otolith daily ages for upper and lower areas of the Pamlico and Cape Fear, NC, estuaries. Dotted lines represent 95 percent confidence intervals. Cape Fear area were usually significantly shorter (P<0.05) than those from other areas, and Atlantic croaker from the lower Cape Fear were significantly longer (P<0.05) than those of other areas. Atlantic croaker collected from the upper Pamlico creeks were significantly larger {P<0.05) on all sample dates than those occupying any area of the Cape Fear during the same weeks. Within the Pamlico, upper and mid-lower mean SLs were the same (P>0.05) except in the last week when mid- lower mean SL was significantly larger (P<0.05). Growth Atlantic croaker ( 12-35 mm SL, n =383 ) ages esti- mated from otoliths ranged from 62 to 234 days, and all age-SL relationships (Fig. 12) were significant (P<0.0001). Residuals of these regressions exhibited no pattern: there- fore, the growth models appeared to be appropriate. The instantaneous daily growth rates within each system were not significantly different (analysis of covariance, P<0.05) between upper and lower areas (Fig. 12). Between systems, upper Pamlico Atlantic croaker grew more slowly than those from the upper Cape Fear (P<0.05). Overall age- length relationships for upper and lower Cape Fear com- bined and for the upper and lower Pamlico Atlantic croaker combined were the following: for Cape Fear — log, gSL = 0.915 -I- 0.0027{age ), r2=0.87, n=229: for Pamlico— logj^SL = 0.970 -I- 0.0024(age), r2=0.87, n=158. The above combined equations for each system were used to calculate age-specific absolute and relative Atlan- tic croaker growth rates (Table 4). Early absolute Atlantic croaker growth rates in the Cape Fear system increased most rapidly in ages <105 days, averaging 0.085 mm/d (Table 4). After this age. Cape Fear growth rates increased at a steady, slow rate, reaching 0. 19 mm/d by age 215 days. Relative Atlantic croaker growth rates in the Cape Fear were constant over the whole age range at about 0.63 %/d SL. The larger Pamlico system Atlantic croaker exhibited similar absolute growth rates to Cape Fear fish and these increased rapidly from 0.077 mm/d between ages 60 and 65 d to 0.106 mm/d between ages 120 and 125 d to 0.175 mm/d between ages 210 and 215 d (Table 4). Relative growth rates were less than those from the Cape Fear and were constant around 0.56 %/d SL. Weight-length relationships for Atlantic croaker were highly significant in all areas (P<0.G001 ) (Table 3). In both systems fish from upper area creeks exhibited significantly larger (P<0.05) weights per length than those from other areas, particularly at the larger sizes (Fig. 13). Slopes of middle and lower Cape Fear weight-length relationships were not significantly different from each other (P>0.05). Mortality Catch curves used to estimate Atlantic croaker mortality rates were calculated by using ages >125 days. All regression slopes were significantly different from zero (P<0.0001), although the relationship was more variable for the mid-lower Pamlico area because of the small sample size. Instantaneous mortality rates for Atlantic croaker in the nursery creeks ranged from 0.008 to 0.038 (Fig. 14). Atlantic croaker in the upper and mid-lower Pamlico 396 Fishery Bulletin 101(2) PAMLICO SOUND upper Figure 9 Mean catch per unit of effort for Atlantic croaker in general areas of the Cape Fear and Pamlico Sound, NC, estuaries by sampling week, March- June 1987. Vertical bars represent plus and minus one standard error of the mean. The .v-axis labels (from left to right) stand for March, April. April, April, May, May, June. areas had similar instantaneous mortality rates (analysis of covariance, P>0.05). Upper Cape Fear Atlantic croaker exhibited significantly lower mortality rates than those in the lower Cape Fear (P<0.0,5). All Atlantic croaker mortal- ity rates in the Cape Fear were significantly higher than those in the Pamlico. Daily mortality rates for Atlantic croaker in the upper and lower Cape Fear were 2.96 %/d and 3.73 %/d, respectively and in the upper and mid-lower Pamlico were 0.90 and 0.80 ^r/d, respectively. Discussion Primary nursery area habitats in two different estuaries were not equally valuable for spot and Atlantic croaker. Considered together, growth, mortality, and distribution data indicated that upstream oligohaline creeks provided the best environment, followed closely by downstream polyhaline areas. In all regards, the middle reaches of the estuary appeared to be less valuable (or at least less used). These consistent results for both species in the two separate estuarine systems lend support to their general applicability. Other studies were marginally useful in evaluating these results because of their lack of synoptic comparisons across a wide variety of habitats and because of limitations in or lack of growth and mortality data for estuarine juveniles of these species. The main evidence that oligohaline habitats provided better environments than polyhaline areas was that spot (both systems) and Atlantic croaker (in Cape Fear) exhib- ited significantly lower mortality in the freshwater PNAs. Miller et al. ( 1985) reported lower mortality for these spe- cies in mesohalino areas compared to high salinity areas of Pamlico Sound. In the Cape Fear River, Weinstein and Ross: Relative value of different estuanne nursery areas for juvenile marine fisfies in North Carolina 397 30 20 10- Li 50 20- 10- ^ 30 Sl 20 >^ ^ 10- 3 n- 0 0) I 20- £ 10 20- 10 0 20-1 1 . t. 1. 18 Mar n.30 lApr na6 15 Apr n=6l 29 Apr n^ 190 l3May n=573 27May n=3a09 -1 — r— r- 10 Jun -1"^ CAPE FEAR ESTUARY MIDDLE 1 i < 1 «i r-»r- X 4- n=689 i^ ^* 1 T 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Standard length (mm) Figure 10 Atlantic croaker length frequencies in general areas of the Cape Fear, NC, estuary by sampling week, March-June 1987. Solid dots represent mean SL. Walters (1981) found consistently high mortality for spot in polyhaline creeks during two years, but variable mortal- ity between years (one year higher, one year lower) in low salinity regions. Mortality rates reported in the present study may not be affected by fish density (Ross, 1992), and it is unlikely that starvation (Currin et al., 1984) played a major mortality role. Predation may cause most of the PNA natural mortality; it was previously proposed that preda- tion rates were lowest in oligohaline habitats because these areas contained relatively fewer predators (Weinstein and Walters, 1981; Currin et al., 1984; Miller et al., 1985). This hypothesis continues to lack direct, convincing evidence. Predators in oligohaline habitats (e.g. southern flounder, catfishes, gar, striped bass, etc.) may, in fact, be just as numerous near the upriver nurseries (author's pers. obs.; Patrick and Moser, 2001; Moser"*) as marine predators are around polyhaline creeks. Also, because water levels in the upriver creeks, especially in the Pamlico, do not vary as much as in polyhaline areas, predators may have more op- portunity to use these creeks (Currin et al., 1984). * Moser, M. L. 1998. Personal commun. NW Fisheries Science Center, NMFS, 2725 Montlake Blvd., Seattle, WA 98112 . One alternative explanation for lower mortality esti- mates in upriver PNAs is that mortality could be related, perhaps indirectly, to ambient salinity. Although freshwa- ter conditions probably do not increase mortalities of these fishes (Moser and Hettler, 1989), there may be negative effects of high salinity on survival that have not been in- vestigated. Moser and Hettler (1989) reported that spot exhibited the highest respiration rates in high salinity conditions, which suggest increased stress. Another potential explanation is that fishes may leave high salinity areas more rapidly than freshwater areas. Although 1 attempted to minimize effects of emigration on mortality estimates by limiting the analyses to the period before mid-June, the mortality rates 1 calculated could have contained an unknown effect of emigration. Other studies (Weinstein, 1983; Weinstein and O'Neil, 1986; Miller and Able, 2002; author's pers. obs. ) supported my assumption that emigration of spot and Atlantic croaker from PNAs was negligible at least through June. Such early habitat fidelity seems to be a common trait among juvenile fishes (Rountree and Able, 1992; Ross and Lancaster, 2002). Many individuals of OWS juvenile fishes leave PNAs by July (Ross, 1988; NC Division of Marine Fisheries^); therefore, 398 Fishery Bulletin 101 (2) PAMLICO UPPER [ [[ 18Mar n= 3 ^-U MID-LOWER *^,* n=78 J^-M ~i 1 1 1 1 r -+- n=70 n = 47 0 10 20 30 40 50 60 70 Standard length (mm) Figure 11 Atlantic croaker length frequencies in general areas of the Pamlico Sound, NC, estuary by sampling week, March- June 1987. Solid dots represent mean SL. previous mortality estimates are likely confounded by emi- gration because measures of declining fish numbers were extended longer into the nursery season (through August, Weinstein and Walters, 1981; through October, Currin et al., 1984; through July, Miller et al., 1985). Growth in weight (weight-length relationships) also in- dicated advantages of oligohaline habitats for these fishes. Higher weights per length have been equated with greater fitness (Friedland et al, 1988; Bolger and Connolly, 1989). Improved fitness was suggested by a consistent trend for individuals of both spot and Atlantic croaker in both sys- tems to be heavier per length in the oligohaline creeks. Laboratory experiments on spot (Moser, 1987) resulted in heavier fish per length in freshwater, and the weight differ- ence was attributed to a higher feeding rate in freshwater, rather than water absorption because of osmotic imbal- ance. Spot from oligohahne areas of the James River, VA, were heavier per length compared to those from several other estuaries (McCambridge and Alden, 1984), but the role of salinity in these differences was unclear. Peterson et al. (1999) indicated that reduced salinity itself caused higher growth rates (in weight) for Atlantic croaker in oli- gohaline conditions. Growth (in length) rates and size distributions indi- cated that PNA habitats at extreme ends of estuaries were equally valuable to both species (with one exception). The exception — depressed spot growth rates in the upper Pam- lico area — did not appear to be correlated with lower sa- linities or temperatures because spot from other areas with low salinity and similar or lower temperatures exhibited higher growth rates. The most obvious difference between upper Pamlico creeks and all other areas was the extremely long (often >100 km) estuarine migration required to reach them. Potential costs involved in such migrations should be examined as should the degree to which the lower spot growth rates persisted into later life. General lack of growth rate variation between oligohaline and polyhaline habitats suggested that salinity (and probably tidal influence) did not affect growth to a degree detectable in the present study. This conclusion is supported by previous studies (Moser and Gerry, 1989; Moser and Hettler, 1989; Miller et al., 2000) despite a general prediction that fish growth rates should be higher in brackish waters (Boeuf and Payan, 2001). The lack of evidence for negative effects of fish density on growth (Ross, 1992) indicated that resources in oligohaline or polyhaline PNAs may not limit these fishes. Currin et al. ( 1984) also suggested that food resources did not limit spot production in middle areas of Pamlico Sound. Lack of spatial variation in early estuarine growth rates was also found in the few relevant studies available. Wein- Ross: Relative value of different estuarine nursery areas for juvenile marine fisfies in North Carolina 399 40 PAMLICO PAMIICO Upper / Lower / / / / / / / , / / / / / / / ./ / / / / / 30- y • X y ^ X ^ 20- yy/ ^■y--'' ' X y yy ^y >' ,^-' Log,pSL=0.979+0.002(Ag«) Loa,QSL=0.98U0.002(A9e) r2 = 0.85,n=121 r2 = 0.91,n = 35 o> 10- c ' — r — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — r — 1 — 1 — r — i — i — i — i — r— i — r— i i i i i i i -i— i t i t r i i i i • ■ . i -5? 1 "D C2 ■a c CAPE FEAR ,' Upper / CAPE FEAR ^ ^0- Lower ^ / / / / / / / / / / / / / / / / 30- / / / ■ // f / ' / / / / / / / / / / ' / ' 20- /y .'#^ y / • ^ y y ■'y>-" ■■::^" -^^' Loa,QSl = 0.907+0.003(Age) ^^''Log,(jSL=0.936+0.003(Age) 10- '" r'' = 0.87,n=127 '' r2-0.87,n = 100 50 70 90 110 130 150 170 190 210 230 50 70 90 110 130 150 170 190 210 230 Age (days) Figure 12 Atlantic croaker growth rate curves (solid lines) based on otolith daily ages for upper and lower areas of the Cape Fear and Pamlico Sound, NC. estuaries. Dotted lines represent 95 percent confidence intervals. stein and Walters (1981) and O'Neil and Weinstein (1987) reported no consistent differences in spot growth rates be- tween oligohaline and polyhaline creeks in the Cape Fear River and York River, VA, estuaries, respectively. Miller et al. (1985) indicated that spot and Atlantic croaker growth rates were probably not different between Pamlico Sound mesohaline and polyhaline areas. Similarly, Beckman and Dean (1984) found no significant differences in spot growth rates among localities in a small, polyhaline South Caro- lina estuary. Necaise (2000) failed to find growth differ- ences among juvenile summer flounder caged (and fed ad libitum) over a wide range of abiotic habitats in southern North Carolina. Guindon and Miller (1995), however, did find growth rate differences among caged (not fed) south- ern flounder across abiotically similar oligohaline habi- tats in the Pamlico River. Differences in fish growth rates among estuarine habitats (e.g. Sogard, 1992) indicate that there are different species-specific responses to habitats, responses related to zoogeography, or responses related to habitat structure or food availability. My data and most of the above studies, covering different years and a variety of estuaries, suggest that variation in growth rates, especially for spot, between PNAs is generally lacking or at least dif- ficult to detect. Such results are consistent with the view that these fishes are hardy, omnivorous, opportunistic colonizers of an undersaturated environment. Increasing evidence suggests that oligohaline or fresh- water habitats in the southeastern United States are im- portant nurseries for the OWS juvenile fishes (Rogers et al, 1984; Rozas and Hackney, 1984;Moser and Gerry, 1989; Moser and Hettler, 1989; Peterson and Ross, 1991). In fact, they may be the most valuable habitats, particularly for maximizing survival of some species. The nursery creeks I sampled supported similar growth rates for two species; however, fitness may be most improved upriver, where both growth (in weight) and survival are optimized. Anderson (1988) predicted that juvenile temperate fishes gener- ally choose to maximize growth over reducing mortality. 400 Fishery Bulletin 101(2) Table 4 Predicted the upper age-specific mean standard lengths (SL), absolute (mni/d), and relative (%/d SL) growth rates for Atlantic croaker from and lower Cape Fear combined (CFR) and the upper and mid-lower Pamlico combined (PAM). Age (days) CFR PAM Mean SL Absolute growth rate Relative growth rate Mean SL Absolute growth rate Relative growth rate 60 11.94 13.61 65 12.32 0.08 0.63 14.00 0.08 0.57 70 12.71 14.39 75 13.11 0.08 0.62 14.79 0.08 0.56 80 13.52 15.21 85 13.95 0.09 0.63 15.63 0.08 0.55 90 14.39 16.07 95 14.84 0.09 0.63 16.52 0.09 0.56 100 15.31 16.98 105 15.79 0.10 0.63 17.45 0.10 0.56 110 16.29 17.95 115 16.81 0.01 0.64 18.45 0.10 0.56 120 17.34 18.97 125 17.89 0.11 0.63 19.50 0.11 0.56 130 18.45 20.04 135 19.03 0.12 0.63 20.61 0.11 0.57 140 19.63 21.18 145 20.25 0.12 0.64 21.78 0.12 0.56 150 20.89 22.39 155 21.55 0.13 0.63 23.01 0.13 0.56 160 22.23 23.66 165 22.94 0.14 0.63 24.32 0.13 0.56 170 23.66 25.00 175 24.41 0.15 0.63 27.70 0.14 0.56 180 25.18 26.42 185 25.97 0.16 0.63 27.16 0.15 0.56 190 26.79 27.93 195 27.64 0.17 0.63 28.71 0.16 0.56 200 28.51 29.51 205 29.41 0.18 0.63 30.34 0.17 0.56 210 30.34 31.19 215 31.30 0.19 0.63 32.06 0.18 0.56 although the two are intimately related (Werner and Gil- liam, 1984). Selecting for optimized growth, however, ap- pears not to be an issue for these two estuarine generalists. If upstream PNAs are better nurseries (i.e. provide better conditions for survival and perhaps growth), delayed PNA recruitment (longer estuarine migrations), especially for Atlantic croaker, may maximize ultimate fitness (Miller et al., 1985; Shapiro, 1987). Factors affecting transport of young to upstream areas may, therefore, be an important determinant of population fitness. Unexpected patterns of recruitment into middle region creeks suggested that their function or recruitment po- tential as fish nursery areas may difTer significantly from other regions. Even though these creeks were physically similar to creeks on either end of the estuarine transects, lower abundances of spot and Atlantic croaker in middle areas suggested that they either avoided (bypassed) middle areas or endured higher initial mortalities there. Higher initial mortality in middle regions seems unlikely because catches were generally low throughout the sampling pe- riod. Relatively poor habitat quality could explain the low densities of fishes in these creeks. This hypothesis was sup- ported by the fact that most fishes settling in middle re- gions of both systems exhibited significantly smaller mean lengths and were lighter per length. The same pattern was observed for Atlantic menhaden in these systems (Ross, 1992). Szedlmeyer (1991) also found lower abundances and species richness in middle reaches of a Florida estuary and suggested that either less diverse habitat or greater salinity variation (or both) influenced this result. Ross and Kpperly (1985) found stations close to the periphery of Pamlico Sound (including the middle area of this study) Ross: Relative value of different estuanne nursery areas for juvenile manne fishes in Nortfi Carolina 401 PAMLICO 2 Uppsr / 1- y^ Lower Weight (g) o ^.^^'''^ CAPE FEAR .. /Mid Upper / / 2- /^ J/ 1- Jt'/ Lower 0- -.^..^.^^^'^''^ 10 20 30 40 50 Standard length (mm) Figure 13 Atlantic croaker weight-length relationships by area in the Pamlico and Cape Fear, NC, estuaries, March-June 1987. to be the most productive, but their study lacked stations near the inlets and in freshwater areas. Weinstein et al.'s (1980) uppermost Cape Fear stations were the same as my middle area and generally produced lower densities of spot than polyhaline areas, but lacking upriver stations, the meaning of this in the present context is inconclusive. These fishes seem to opt either for rapid settlement in poly- haline environments or delayed settlement in oligohaline areas — mesohaline settlement being less "preferred." The conclusion that PNAs were not equally valuable and the observation that spot were not most abundant in the best habitats, indicated that variation in estuarine distribution could control or at least regulate (fine tune) year-class strength. If movement to general regions of the estuary is largely passive (Pietrafesa et al., 1986b; Pi- etrafesa and Janowitz, 1988), then my results predict that year-class strength of these species would be decreased when transport conditions force the majority of the recruits toward middle or lower region PNAs. Alternatively, year class strength would be enhanced by conditions favoring greater upstream transport, assuming carrying capacities of the habitats were not exceeded. Ross (1992) proposed that these systems were recruitment limited, that post- settlement mortality was less important in controlling year-class strength than early life history events prior to settlement. If true, factors affecting variation in estuarine distribution may indirectly adjust year-class strength, not control it. Additional data on mortality rate variation in relation to density during the estuarine and oceanic early life history is required to validate this hypothesis. Acknowledgments I thank John M. Miller, G, T. Barthalmus, L. B. Crowder, and L. J. Pietrafesa for their support during this study. I thank K. H. Pollock for statistical advise. Field sampling required the efforts of many people. The NC Division of Marine Fisheries (Washington and Wilmington offices) played a large role in sampling, and I especially thank Fred Rohde, John Schoolfield, Otto Rutten, Morris Allison, Greg Judy, Lele Tison, and Jess Hawkins of that organiza- tion. B. M. "Mac" Currin was important throughout the study, and I thank him for his contributions in the field, laboratory, and in reviewing manuscripts. John S. Burke also provided help in the field. I thank the Beaufort Labo- 402 Fishery Bulletin 101(2) I ° ^ 10- o _I 8 PAMLICO Upper Lo9gN=5.48-0.009(Age) 2] r2 = 0.73,n = 31 -I — I — I — I — 1 — I — I — I — I — I — I — I — 1 — I — r- CAPE FEAR Upper LogaN = 11.44-0.030(Age r2 = 0.90,nO6 log^N=3.53-0.008(Age) r2 = 0.28,n = 27 PAMLICO Lower T — 1 — I — r — 1 — r — I — 1 — I — 1 — I — I — I — r — i — i — i — i — i — r CAPE FEAR Lower LoggN=12.61-0.038(Age r2=0.94,n = 30 T — I—I — I — I — I — I — I — 1 — I — I — I — 1—1 — I — 1 — I — 1 — I — 1—1 — I — 1 — 1 — I— - — I — I — I — I — I — 1 — I — 1 — I — I — I — 1—1 — 1 — I — I — I — I — I — I — I — I — I — I — r- 70 90 110 130 150 170 190 210 230250 270 290 70 90 110 130 150 170 190 210 230250 270 290 Age (days) Figure 14 Atlantic croaker mortality rate curves (solid lines) based on otolith daily ages for upper and lower areas of the Pamlico and Cape Fear, NC, estuaries. Dotted lines represent 95 percent confidence intervals. ratory (National Marine Fisheries Service) and the Biol- ogy Laboratory of Carolina Power and Light Company for providing space. David Colby made valuable contributions to this research. I appreciate Jeff Isely's advice and help in analyzing fish otoliths. I thank Ernie Aschenbach for sorting samples and mounting otoliths. And lastly I thank Mary L. Moser for help and support during all stages of this work from field sampling to reading numerous manu- script drafts. Literature cited Anderson, J. T. 1988. A review of size dependent survival during pre-recruit stages of fishes in relation to recruitment. J. Northwest Atl. Fish. Sci. 8:55-66. Baldevarona, R. B. 1987. Effects of feeding and stocking density on growth and survival of spot, LeioHtDtnus xanthurus. Ph.D. diss., 117 p. Univ. South Carolina, Columbia, SC. Beck, M. W., K. L. Heck Jr. K. W. Able, D. L. Childers, D. B. Eggleston, B. M. Gillanders, B.. Halpern, C. G. Hays, K. Hoshino, T. J. Minello, R. J. Orth, P F Sheridan, M. P. Weinstein. 2001. The identification, con.servation. and management of estuarine and marine nurseries for fish and invertebrates. Bioscience 51:6.'i.'i-641. Beckman, D. W., and J. M. Dean. 1984. The age and growth of young-of-the-year spot, Leios- tomus xanthurus Lacepode, in South Carolina. Estuaries 7:487-496. Boeuf, G., and P. Payan. 2001. How should salinity influence fish growth? Comp. Biochem. Physiol, (part C) 130:411-423. Boesch, D. F, and R. E. Turner 1984. Dependence of fishery species on salt marshes: the role of food and refuge. Estuaries 7:460- 468. Bolger, T, and R L. Connolly. 1989. The selection of suitable indices for the measurement and analysis offish condition. J. Fish. Biol. 34:171- 182. Campana, S. E., J. A. Gagn, and J. Munro. 1987. Otolith microstructure of larval herring iClupea harengus): image or reality? Can. J. Fish. Aquat. Sci. 44; 1922-1929. Cowan, J. H., Jr. 1988. Age and growth of Atlantic croaker, Micropogonias undulatus, larvae collected in the coastal waters of the northern Gulf of Mexico as determined by increments in saccular otoliths. Bull. Mar Sci. 42:349-357. Crecco, V, T. Savoy, and L. Gunn. 1983. Daily mortality rates of larval and juvenile American shad (A/o.sa sapidissima) in the Connecticut River with changes in year-class strength. Can. J. Fish. Aquat. Sci. 40:1719-1728. Currin, B. M., J. P. Reed, and J. M. Miller. 1984. Growth, production, food consumption and mortality of juvenile spot and croaker: a comparison of tidal and non- tidal nursery areas. Estuaries 7:451-459. Essig, R. J., and C. F. Cole. 1986. Methods of estimating larval fish mortality from daily increments in otoliths. Trans. Am. Fish. Soc. 115:34-40. Friedland, K. D., G. C. Carman, A. J. Bejda, A. L. Studholme, and B. 011a. 1988. Interannual variation in diet and condition in juvenile Ross: Relative value of different estuanne nursery areas for luvenile manne fishies in Nortfi Carolina 403 bluefish during estuanne residency. Trans. Am. Fish. Soc. 117:474-479. Giese, G. L., H. B. Wilder, and G. G. Parker Jr. 1979. Hydrology of major estuaries and sounds of North Carolina. U.S. Geological Survey Water Resources Invest. 79-46, 175 p. Guindon, K. Y., and J. M. Miller 1995. Growth potential of juvenile southern flounder, Parali- chthys lethostigma, in low salinity nursery areas of Pamlico Sound, North Carolina, USA. Netherlands J. Sea Res. 34: 89-100. Hobbie, J. E. 1970. Hydrography of the Pamlico River Estuary, N.C. Water Resources Res. Inst. Rep. 39, 69 p. Water Resources Research Institute, Raleigh, NC. McCambridge, J. T., Jr, and R. W. Alden III. 1984. Growth of juvenile spot, Leiostomus xanthurus Lace- pede, in the nursery region of the James River, Virginia. Estuaries 7:478-486. Mclvor, C. C, and W E. Odum. 1988. Food, predation risk, and microhabitat selection in a marsh fish assemblage. Ecol. 69: 1341-1351. Mercer, L. P. 1987a. Fishery management plan for Atlantic croaker {Mi- cropogonias undualtus). Atlantic States Mar Fish. Comm. Fish. Management Rep. 10, 90 p. Atlantic States Marine Fisheries Commission, Washington, DC. 1987b. Fishery management plan for spot (.Leiostomus xanthurus). Atlantic States Mar. Fish. Comm. Fish. Man- agement Rep. 11, 81 p. Atlantic States Marine Fisheries Commission, Washington, DC. Miller, J. M., L. B. Crowder, and M. L. Moser 1985. Migration and utilization of estuarine nurseries by juvenile fishes: an evolutionary perspective. In Migration: mechanisms and adaptive significance (M. A. Rankin, ed.), p. 338-352. Contrib. Mar Sci. (suppl. 27. Miller, J. M., W. H. Neill, K. A. Duchon, and S.W. Ross. 2000. Ecophysiological determinants of secondary produc- tion in salt marshes: a simulation study. In Concepts and controversies in tidal marsh ecology (M. P. Weinstein and D. A. Kreeger, eds.), p. 315-331. Kluwer Academic Press, Dordrecht, NL. Miller, M. J. and K. W. Able. 2002. Movements and growth of tagged young-of-the-year Atlantic croaker (Micropogonias undulatus L.) in restored and reference marsh creeks in Delaware Bay, USA. J. Exp. MarBiol. Ecol. 267:15-33. Miltner, R. J., S. W. Ross, and M. H. Posey 1995. Influence of food and predation on the depth distri- bution of juvenile spot [Leiostomus xanthurus) in tidal nurseries. Can. J. Fish. Aquat. Sci. 52:971-982. Moser, M. L. 1987. Effects of salinity fluctuation on juvenile estuarine fish. Ph.D. diss., 150 p. North Carolina State Univ., Raleigh, NC. Moser, M. L., and L. R. Gerry. 1989. Differential effects of salinity changes on two estua- rine fishes, Leiostomus xanthurus and Micropogonias undulatus. Estuaries 12:35-41. Moser, M. L., and W. F. Hettler. 1989. Routine metabolism of juvenile spot, Leiostomus xan- thurus (Lacepede), as a function of temperature, salinity and weight. J. Fish Biol. 35:703-707. Necaise, A. M. 2000. Habitat evaluation as measured through the growth of juvenile red drum, Sciaenops ocetlatus, and summer (lounder,Paralichthysdentatus. M.S. thesis, 49 p. North Carolina State Univ. Raleigh, NC. Nelson, D..M., M..E. Monaco, E..A. Iriandi, L..R. Settle, and L. Coston-Clements. 1991. Distribution and abundance of fishes and inverte- brates in southeast estuaries. ELMR (Estuarine Living Marine Resources) Rep. 9, 167 p. NOAA/NOS Strategic Environmental Assessment Division, Rockville, MD. Neter, J., W. Wasserman, and M. H. Kutner 1985. Applied linear statistical models. Regression, analy- sis of variance, and experimental design, 1127 p. Irwin. Homewood, IL. O'Neil, S. P., and M. P. Weinstein. 1987. Feeding habitats of spot, Leiostomus xanthurus, in polyhaline versus meso-oligohaline tidal creeks and shoals. Fish. Bull. 85:785-796. Patrick, W. S., and M. L. Moser 2001. Potential competition between hybrid striped bass (Morone saxatilis x M. americana) and striped bass (M. saxatilis) in the Cape Fear River estuary, North Carolina. Estuaries 24:425-429. Peters, D. S., J. C. DeVane Jr., M. T. Boyd, L. C. Clements, and A. B. Powell. 1978. Preliminary observations on feeding, growth, and energy budget of larval spot {Leiostomus xanthurus). In Ann. Rep. Southeast Fish. Cent., Beaufort Lab. to U.S. Dep. Energy, p. 377-379. Beaufort Laboratory, National Marine Fisheries Service, Beaufort, NC. Peterson, M. S., and S. T. Ross. 1991. Dynamics of littoral fishes and decapods along a coastal river-estuarine gradient. Est. Coast. Shelf Sci. 33: 467^83. Peterson, M. S., B. H. Comyns, C. F. Rakocinski, and G. L. Fulling. 1999. Does salinity affect somatic growth in early juvenile Atlantic croaker, Micropogonias undulatus (L.)? J. Exp. Man Biol. Ecol. 238:199-207. Pietrafesa, P. J., and G. S. Janowitz. 1988. Physical oceanographic processes affecting larval transport around and through North Carolina inlets. Am. Fish. Soc. Symp. 3:34-50. Pietrafesa, L. J., G. S. Janowitz, T. Chao, R. H. Wiesberg, F. Askari and E. Noble. 1986a. The physical oceanography of Pamlico Sound. Univ North Carolina Sea Grant Publ. UNC-WP-86-5, 125 p. Univ. North Carolina Sea Grant Program, Raleigh, NC. Pietrafesa, L. J., G. S. Janowitz, J. M. Miller, E. B. Noble, S. W. Ross, and S. P Epperly 1986b. Abiotic factors influencing the spatial and tempo- ral variability of juvenile fish in Pamlico Sound, North Carolina. In Estuarine variability (D. A. Wolfe, ed.), p. 341-353. Academic Press, New York, NY. Ricker,W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Can. 191, 382 p. Rogers, S. G., T E. Targett, and S. B. Van Sant. 1984. Fish-nursery use in Georgia salt-marsh estuaries: the influence of springtime freshwater conditions. Trans. Am. Fish. Soc. 113:595-606. Ross. S. W. 1988. Age, growth and mortality of Atlantic croaker in North Carolina, with comments on population dynamics. Trans. Am. Fish. Soc. 117:461-473. 1992. Comparisons of population dynamics of juvenile spot {Leiostomus .xanthurus), Atlantic croaker {Micropogonias 404 Fishery Bulletin 101(2) undulatus), and Atlantic menhaden {Brevoortia tyrannus) among diverse North Carolina estuarine nursery areas. Ph.D. diss.. 144 p. North Carolina State Univ., Raleigh, NO. Ross, S. W., and S. P. Epperly. 1985. Utilization of shallow estuarine nursery areas by fishes in Pamlico Sound and adjacent tributaries. North Carolina. In Fish community ecology in estuaries and coastal lagoons (A. Yanez-Arancibia, ed.), Ch. 10, 207-232. UNAM (Universidad National Autonoma de Mexico) Press, Mexico. Ross, S. W., and J. E . Lancaster 2002. Movements and site fidelity of two juvenile fish spe- cies using surf zone nursery habitats along the southeastern North Carolina coast. Environ. Biol. Fishes 63:161-172. Rountree, R. A., and K. W. Able. 1992. Foraging habitats, growth, and temporal patterns of salt-marsh creek habitat use by young-of-the-year summer flounder in New Jersey. Trans. Am. Fish. Soc. 121:765-776. Rozas, L. R and C. T. Hackney 1984. Use of oligohaline marshes by fishes and macrofaunal crustaceans in North Carolina. Estuaries 7: 213-224. SAS Institute, Inc. 1988. SAS/STAT user's guide, release 6.03 edition. SAS Inst., Inc. Cary, NC. Shapiro, D. Y. 1987. Inferring larval recruitment strategies from the dis- tributional ecology of settled individuals of a coral reef fish. Bull. Mar. Sci. 41:289-295. Siegfried, R. C, II, and M. R Weinstein. 1989. Validation of daily increment deposition in the otoliths of spoHLeiostomus xanthiiruii). Estuaries 12:180-185. Sogard, S. M. 1992. Variability in growth rates of juvenile fishes in differ- ent estuarine habitats. Mar Ecol. Prog. Ser. 85:35-53. Szedlmayer, S. T. 1991. Distribution and abundance of nearshore fishes in the Anclote River estuary, west-central Florida. Northeast Gulf Sci. 12:75-82. Thresher, R. E. 1985. Distribution, abundance, and reproductive success in the coral reeffishAcanthochromispolyacanthus. Ecol. 66: 1139-1150. Warlen, S. M. 1980. Age and growth of larvae and spawning time of Atlan- tic croaker in North Carolina. Proc. Ann. Conf Southeast Assoc. Fish. Wildl. Agencies 34:204-214. Warlen, S. M., and A. J. Chester. 1985. Age, growth, and distribution of larval spot, Leiosto- mus xanthurus, off North Carolina. Fish. Bull. 83:587- 599. Weinstein, M. P. 1979. Shallow marsh habitats as primary nurseries for fishes and shellfish. Cape Fear River, North Carolina. Fish. Bull. 77:339-357. 1982. Commentary: a need for more experimental work in estuarine fisheries ecology. Northeast Gulf Sci. 5:59-64. 1983. Population dynamics of an estuarine-dependent fish, the spot (Leiostomus xanthurus), along a tidal creek-sea- grass meadow coenocline. Can. J. Fish. Aquat. Sci. 40: 1633-1638. Weinstein, M. P, and S. R O'Neil. 1986. Exchange of marked juvenile spots between adjacent tidal creeks in the York river estuary, Virginia. Trans. Am. Fish. Soc. 115:93-97. Weinstein, M. P., and M. P. Walters. 1981. Growth, survival and production in young-of-the-year populations of Leiostomus xanthurus Lacepede residing in tidal creeks. Estuaries 4:185-197. Weinstein, M. P., S. L. Weiss, and M. F. Walters. 1980. Multiple determinants of community structure in shallow marsh habitats. Cape Fear River estuary. North Carolina, USA. Mar Biol. 58:227-243. Welch, J. M., and B. B. Parker. 1979. Circulation and hydrodynamics of the lower Cape Fear River, North Carolina. U.S. Dep. Comnier., NOAA Tech. Rep. NOS 80, 108 p. Werner, E. E., and J. F Gilliam. 1984. The ontogenetic niche and species interactions in size- structured populations. Ann. Rev. Ecol. Syst. 15:393-425. Zar, J. H. 1984. Biostatistical analysis, 2nd ed., 718 p. Prentice Hall, Inc., Englewood Cliffs, NJ. 405 Abstract— Age and growth estimates for the winter skate (Leucoraja ocel- lata) were estimated from vertebral band counts on 209 fish ranging in size from 145 to 940 mm total length (TL). An index of average percent error (lAPE) of 5.8% suggests that our aging method represents a precise approach to the age assessment of L. ocellata. Marginal increments were significantly different between months (Kruskal- Wallis P<0.001) and a distinct trend of increasing monthly increment growth began in July. Estimates of von Berta- lanffy growth parameters suggest that females attain a slightly larger asymp- totic TL (L„=1374 mm) than males (L_=1218 mm) and grow more slowly (*=0.059 and 0.074, respectively). The oldest ages obtained for the winter skate were 19 years for males and 18 years for females, which corresponded to total lengths of 932 mm and 940 mm, respectively. The results indicate that the winter skate exhibits the charac- teristics that have made other elasmo- branch populations highly susceptible to exploitation by commercial fisheries. Age and growth estimates of the winter skate (Leucoraja ocellata) In the western Gulf of Maine James A. Sulikowski Michael D. Morin Seung H. Suk W. Huntting Howell Zoology Department, Spaulding Hall University of New Hampshire 46 College Road Durham, New Hampshire 03824 E mail address (for J A, Sulikowski) |sulikow@hotmail com Manuscript accepted 17 December 2002. Manuscript received 31 December 2002 at NMFS Scientific Publications Office. Fish. Bull. 101:405-413 (2003). Little is known about the biology of many elasmobranchs, including impor- tant parameters such as validated age, growth, age at maturity, reproductive cycles and annual fecundity (Frisk et al., 2001). Difficulty in obtaining samples, the large size of specimens, their high mobility, and minor commercial value are just a few of the problems that make such studies complicated and in some respects impractical (Cailliet et al., 1983; Cailliet et al., 1986). The recent intensification in commercial fishing of elasmobranchs (Cailliet et al., 1983; Brown and Gruber, 1988; Kusher et al., 1992; Dulvey et al., 2000) has made the collection of their life history informa- tion essential to the realistic manage- ment of their populations (Cailliet et al., 1983; Ryland and Ajayi, 1984; Dulvey et al., 2000). Historically, batoids have been of minimal commercial value (Otwell and Lanier, 1979; Sosebee, 1998); hence the majority of research on elasmobranchs has focused on com- mercially valuable sharks (e.g. Holden, 1977; Natanson et al., 1995; Walmsley- Hart et al., 1999). According to the characteristics outlined by Winemiller and Rose (1992) and the comparative analyses of Frisk et al. (2001), skates, like other elasmobranchs, fall into the category of equilibrium strategists and as such reach sexual maturity at a late age, have a low fecundity, and are rela- tively long-lived. These characteristics, coupled with fisheries that select for the removal of large individuals (especially those over 100 cm total length), make these particular fish highly susceptible to overfishing (Hoenig and Gruber, 1990; Dulvey et al., 2000; Frisk et al., 2001). Traditionally, skates caught by ground fishing operations were dis- carded (Martin and Zorzi, 1993; Jun- quera and Paz, 1998; Sosebee, 1998). New and expanding markets for skate wings have made retention of these fish commercially more lucrative in recent years (Sosebee, 1998; New England Fishery Management Council^. Skate harvests in the U.S. portion of the western North Atlantic are currently unregulated. Moreover, biological infor- mation on skate life histories is almost nonexistent (Frisk, 2000). This combi- nation of factors is believed to have led to a depletion of common skates (Raja batis) in the Irish sea (Brander, 1981). The winter skate (Leucoraja ocellata) is a large species (total length over 100 cm) of skate of the family Rajidae (Big- elow and Schroeder, 1953; Robins and Ray, 1986; New England Fishery Man- agement Council'). It is endemic to the inshore waters of the western Atlantic, from the Newfoundland Banks and the southern Gulf of St. Lawrence in Can- ada to North Carolina in the United States (Bigelow and Schroeder, 1953). Despite this wide range, little direct biological data is available for this spe- cies (Simon and Frank, 1996; Casey and ' New England Fishery Management Coun- cil. 2001. 2000 Stock assessment and fishery evaluation (SAFE) report for the northeast skate complex, 179 p. New England Fishery Management Council, 50 Water Street, Mill 2 Newburyport, MA 01950. 406 Fishery Bulletin 101(2) Myers, 1998, Frisk, 2000). Recent assessment studies in the northeast U.S. (Northeast Fisheries Science Center^), suggest that the biomass of the winter skate may be below threshold levels mandated by the Sustainable Fisheries Act (SFA). To add insight into the life history of this species and the status of the stock (Simpfendorfer, 1993; Frisk et al., 2001 ), we estimated age and growth rates of L. ocellata by interpreting annular counts and marginal increments on vertebral centra from specimens collected in the western Gulf of Maine. Materials and methods Sampling A total of 304 winter skates were captured by otter trawl between November 1999 and May 2001 at locations that ranged from 1.6 to 32 km off the coast of New Hampshire. Approximate depths at these locations ranged between 9 and 107 m. Skates were maintained alive on board the vessel until transport to the University of New Hampshire's Coastal Marine Laboratory (CML). There, individual fish were euthanized (0.05g/L bath of MS222). We measured total length (TL in mm) as a straight line distance from the tip of the rostrum to the end of the tail, and disc width (DW in mm) as a straight line distance between the tips of the widest portion of pectoral fins. Total wet weight (kg) was also recorded. In order to differentiate between the small, immature specimens of little skates (Leucoraja erinacea, a congener species also found in the Gulf of Maine) and winter skates, rows of teeth in the upper jaw were counted. Skates whose number of teeth ranged between 72 and 110 per row were identified as L. ocellata and skates whose number of teeth ranged between 38 and 64 per row were identified as L. erinacea (Bigelow and Schroeder, 1953). To reduce any uncertainty in species identification, skates having between 38 and 71 teeth per row were not used in this study. Preparation of vertebral samples Vertebral samples, taken from above the abdominal cavity, were removed from 132 females and 98 males, labeled, and stored frozen. After defrosting, three centra from each specimen were freed from the vertebral column, stripped of excess tissue and air dried. Large centra were cut sagit- tally, while held within a vise, with a DremeF"^' tool fitted with a mini-saw attachment. Smaller centra were sanded with a DremeF'^' tool to replicate a sagittal cut. Processed vertebrae were mounted horizontally on glass microscope slides and ground with successively finer grits (#180, #400, #600), of wet-dry sandpaper. Each vertebra was then remounted and the other side ground to produce a thin (300 micrometer) "hourglass" section. Counts of annuli Vertebral sections were viewed through a compound micro- scope (25-40x) with reflected light (Fig. 1). A growth ring (annulus) was defined as an opaque and translucent band pair that traversed the intermedialia and that clearly extended into the corpus calcareum (Casey et al., 1985; Brown and Gruber, 1988). The birth mark (age zero) was defined as the first distinct mark distal to the focus that coincided with a change in the angle of the corpus calca- reum (Casey et al, 1985; Wintner and Cliff, 1996). Three nonconsecutive counts of annuli were made for the three vertebral sections from each specimen without prior knowledge of the length of the skate or previous counts. If the variability between readings was more than two years, that particular specimen was eliminated from further analyses. Count reproducibility was estimated by using the index of average percent error ( lAPE ) described by Beamish and Fournier (1981): M/^£=1/A'^(|//?^(|a-,, -X,|/X,)jxl00, where N = the number of skates aged; R = the number of readings; X = the ith age determination of thejth fish; and X = the average calculated for thejth fish. An upper limit for the lAPE was arbitrarily set at 15'7( for each vertebra. Vertebrae with statistically acceptable lAPE indexes were used for estimation of asymptotic growth rates (Brown and Gruber, 1988; Cailliet and Tanaka, 1990). The average of the mean counts for all three centra defined the age estimate for each specimen (Casey et al., 1985; Wintner and Cliff, 1996). A von Bertalanffy growth function (VBGF) was fitted to the data with the following equation (von Bertalanffy, 1938): where L, K to total length at time t (age in years); theoretical asymptotic length; Brody growth constant; and theoretical age at zero length. ^ Northeast Fisheries Science Center. 1999. 30th northeast regional stock assessment workshop, 477 p. Northeast Fisher- ies Science Center, 166 Water Street Woods Hole, MA 02543- 1026. Growth in length data were analyzed by using FISHPARM, a computer program for parameter estimation of nonlinear models with Marquardt's ( 1963) algorithm for least-square estimation of nonlinear parameters (Prager et al., 1987). Marginal increment analyses The annual periodicity of band pair formation was inves- tigated by using marginal increment analyses (MIA). Because the annuli in older specimens were closer together, marginal increments were calculated from five specimens per month whose centra contained either four or five annuli. For MIA determination, the distance of the final opaque band and the penultimate opaque band from the centrum edge were measured with an ocular micrometer. Sulikowski et al.: Age and growth estimates of Leucorqa ocellata 407 ^m WM 4 "\ J w\ / . ' h_ -Jm J^ #: 1 OSS-section of a vertebral centn. = birth mark; Arrows represent u r ale caught in July and estimated to Longitudinal ci be 5 years. BM w Figure 1 im from a 422-mm-TL m age in years. The marginal increment was calculated as the ratio of the distance between the last and penultimate bands (Brans- tetter and Musick, 1994; Cailliet, 1990; Simpfendorfer, 1993; Simpfendorfer, 2000). Mean average increments by month of capture were plotted to identify trends in band formation by using a Kruskal-Wallis one-way analysis of variance on ranks. (Simpfendorfer, 1993; 2000). Results Morphological measurements A total of 230 specimens were used for this study. Males (;!=98) ranged between 147-932 mm TL, 82-601 mm DW, and 0.015-6.2 kg. Females (/i = 132) ranged between 145-940 mm TL, 82-635 mm DW, and 0.015-7.5 kg. A linear relationship existed between the total length, disk width, and mass relationships for male, female, and the sexes combined (all r- values were greater than 0.85). Two skates (one male: TL=147 mm, DW=82 mm, weight=0.015 kg; and one female TL= 145 mm,DW=82 mm, weight=0.015 kg) hatched from egg cases during May 2001 in the CML after gestating 18 months. One wild male specimen (age-0, TL=175 mm, DW=100 mm, weight=0.027 kg) was also cap- tured and incorporated into the results of this study. the centra) were easily distinguished from complete bands. Of the 230 processed vertebrae, 209 (91%) were readable. These 209 vertebrae (males=88;females=121) had annular count estimates that agreed within two years, resulting in an TAPE of 5.8%. Mean total length and disk width at age for male, female, and sexes combined are given in Table 1. The relationship between TL and centrum diameter was linear {r~=Q.92\ P<0.05; Fig. 2) and there were no signifi- cant differences (ANCOVA, P<0.05) between males and females. Because no significant difference existed for TL and centrum diameter between the sexes, the data were combined (Fig. 2). Marginal increments were averaged from five speci- mens for each month, except June when skates belonging to the 4 and 5 year age classes were unavailable. Marginal increments were significantly different between months (Kruskal-Wallis P<0.001) and a distinct trend of increas- ing monthly increment growth began in July (Fig. 3). Maxi- mum marginal increment measurement occurred in May. Minimum marginal increment measurement occurred in July Two recently hatched males (one from the laboratory (147 mm TL) and one from field collections (175 mm TL)) had opaque zones on the distal edge of their vertebral cen- tra. Reviewing this information, we suspect that a single opaque band may be formed annually on the vertebral centra during June-July in the winter skate. Vertebral analyses No difficulty was encountered in estimating the age of L. ocellata. False bands (bands that do not completely encircle Age and growth estimates We assumed that opaque-translucent band pairs were formed annually, and we fitted von Bertalanffy growth 408 Fishery Bulletin 101 (2) 1000 - Inlerccpi -0.63 • f^"""^ * Slope 0.01 , Jt»V*^ 800 - 209 - • f%^ • r 0.92 • •"TC^^« '• e" IXT-** ' E. 600 - • 1 -iP^' c *4 *'^* • IS 400- jtf'^V^ o 5^r* 1- ^^^9 .^""""^f ^ ^""^ %^ 200 - 2 4 6 8 10 Centrum diameter (mm) Figure 2 The relationship of total length (mm) to centrum diameter (mm) for combined sexes of winter skate. Table 1 Average total length, TL, and d ISC width, DW, at age for winter skates (L ocellafa) by sex and combined sexes. Mean ±1 SEM; sample sizes (no. offish in samp e) are given in parentheses. Age Male TL Female TL Sexes combined Male DW Female DW Sexes combined 0 161 (2) ±14 145(1) 156 (3) ±10 93 ±7 81 89 ±7 1 228(1) — 228(1) 139 — 139 2 264 (5) ±14 268 (4) ±21 266 (9) ±11 153 ±5 158 ±8 155 ±5 3 340 (4) ±20 317 (9) ±9 324 (13) ±9 198 ±6 188 ±6 191 ±6 4 379 (12) ±8 392 (25) ±8 388 (37) ±6 223 ±8 233 ±5 230 ±4 5 435 (4) ±19 429 (16) ±13 430 (20) ±6 264 ±11 259 ±13 260 ±6 6 536 (5) ±13 501 (7) ±16 516(12)±12 338 ±8 310 ±15 322 ±11 7 609(1) 551 (12) ±6 556 (13) ±7 392 342 ±6 346 ±6 8 651(1) 565 (11) ±13 570(12) ±13 401 352 ±10 356 ±10 9 658 (9) ±24 632 (9) ±20 645 (18) ±15 420 ±18 403 ±16 411 ±11 10 690 (12) ±20 704 (8) ±18 696 (20) ±14 441 ±22 447 ±17 444 ±11 11 735 (10) ±17 761 (5) ±22 744 (15) ±14 479 ±24 498 ±20 485 ±14 12 743 (5) ±24 763 (5) ±19 753 (10) ±15 488 ±12 501 ±15 494 ±8 13 830 (3) ±7 772 (3) ±16 801 (6) ±15 495 ±6 506 ±8 500 ±5 14 838 (4) ±10 803 (3) ±23 821 (7) ±13 530 ±9 527 ±24 529 ±10 15 841 (4) ±12 — 841 (4) ±12 541 ±19 — 541 ±13 16 860 (4) ±4 842(1)±0 857 (5) ±5 565 ±16 542 560 ±10 17 921(1) — 921(1) 579 — 579 18 — 940 (2) ±0 940 (2) ±0 — 623 ±13 623 ±13 19 932(1) — 932(1) 601 — 601 curves ( VBGC ) to total length-at-age data ( Fig. 4 ). The VBGC provided a good fit with a low .standard error for males, females, and both sexes combined (Table 2). The /„ values (-1.4 to -1.6) compared favorably with gestation rates for the two skates hatched in captivity (1.5 years) (Table 2). The von Bertalanffy growth parameters for males, females, and the sexes combined were similar but k values were higher for males and sexes combined, than for females. Sulikowski et al : Age and growth estimates of Leucora/a ocellata 409 1 0- 09- i Average marginal increment o o o o o 1 Y" / r-^ 1 03- 1 Jan Feb Mar Apr May Jun Jul Aug Sept Oct Hov Dec Month Figure 3 Mean monthly marginal increments of opaque bands for L. ocellata from the Gulf of Maine. Marginal increments were calculated from five specimens per month whose centra contained either 4 or 5 annuli. Error bars represent ±1 SEM. Discussion The relationship between TL and centrum diameter was linear and significant, indicating that the centra grew pro- portionally to skate length for all size classes, and thus this structure was useful for age analyses (Kusher et al., 1992). The 5.8% lAPE index suggests that our aging method rep- resents a precise approach to the age assessment of L. ocel- lata. Minimal width of the marginal increment for winter skates captured in May supports the hypothesis of annual band formation in this species. Moreover, these results compare favorably to growth cycles in marginal incre- ments for other skates found in temperate waters whose vertebral bands are formed annually (Holden and Vince, 1973; Waring, 1984; Natanson, 1993). Von Bertalanffy parameters, as determined by our study, suggest that females attain a slightly larger asymptotic TL^ (1374 mm) than males (1218 mm) and grow more slowly (/f=0.059 and 0.074, respectively). This trend follows a com- mon pattern in batiods. Holden ( 1977), Waring ( 1984), Ry- land and Ajayi ( 1984), Brander ( 1981 ), and Walmsley-Hart et al. (1999) found similar tendencies in several species of skates, and Martin and Cailliet (1988) found comparable results in the bat ray (Myliohatis californica). Our estimates of L„ exceeded those of the largest speci- mens in our field collections (940 mm for females and 932 mm for males). Nevertheless, data from extensive trawl surveys in the western Gulf of Maine and the Mid-Atlantic offshore region spring and autumn bottom trawl surveys from 1967 to 2000 (Northeast Fisheries Science Center-) indicated that mean TL did not exceed 1000 mm. Thus, we suspect that our von Bertalanffy equation produces an accurate estimation of L,^ for winter skate. Walmsley-Hart Table 2 Calculated von Bertalanffy parameters for male, female. and combined sexes of L. ocellata. r~ is the coefficient of determination. Parameter Male Female Combined sexes LjmmTL) 1218 1374 1314 k (/year ) 0.074 0.059 0.064 /,, (year! -1.418 -1.609 -1.531 r- 0.946 0.939 0.946 SE 0.01 0.01 0.001 n 88 121 209 et al. (1999) overestimated L^ fori?. puUopunctata and sug- gested that small sample size and rareness of large indi- viduals were most likely responsible. Because fishing gear was not biased towards a specific marketable skate size and because all size classes of L. ocellata were represented, it is quite possible that the rareness of large individuals led to the augmented L.^ in combined and individual sexes in our study. Possibly, a larger sample size of winter skates would produce significant and divergent results with re- gard to von Bertalanffy parameters. However, the close fit of the data to the VBGC for L. ocellata indicates the VBGC is an appropriate model for this species. Preliminary estimates of age and growth parameters are available for winter skate in Canadian waters ( eastern Sco- tian Shelf) from Simon and Frank ( 1996), who reported the results of a study conducted at St. Mary's University by R. Nearing. Combined sexes of winter skates (/2=242) with TL 410 Fishery Bulletin 101(2) 1000 - A 800 - ,,v>H''^^ ▼ f L'^ f >^^ -^^ » ^ 600 - JX » . 400 - 200 ; , y^ I - 1 r 1 r 1 1 1 1 1 1 £ 0 2 4 6 8 10 12 14 16 18 20 c 1 1000-1 B 800- 600- i-P 400- ^^ 200- ( y^ 0 - 0 2 4 6 8 10 12 14 16 18 20 Age (years) Figures 4 Von BertalanfTy growth curves generated from vertebral data for (A) male, (B) female, and (C) combined sexes of winter skate (L. oceltata) from the western Gulf of Maine. Individual VBGC parameters are given in Table .3. ranging from 120 to 1060 mm and ages from 0 to 16 years provided von Bertalanffy parameters of L. = 114.1 cm, k - 0.14405, and /„ = 0.00315. However, these data should be viewed with caution because no lAPE values nor valida- tion of the annual nature exist for these estimates, and it is likely that the older specimens had been under-aged by four or more years (Simon '). /C values (an estimation of how quickly an animal grows to LJ were similar for both sexes of winter skate. These growth rates are commensurate with other skate species of ^ Simon, J. 2001. Personal commun. Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. similar size, but slower than skate species of smaller size (Table 3). The oldest ages obtained for the winter skate were 19 and 18 years for males and females, respectively. These data are in agreement with the assumption that larger batoids, such as L. ocellata and R. pullopunctata (Walmsley-Hart et al., 1999) are longer lived and grow more slowly than smaller species, such as R. erinacea, which has been aged to 8 years with a k value of 0.352 (Johnson, 1979; Waring, 1984). Accurate stock assessment data for skates is difficult to collect in the northeast United States because species are rarely differentiated in landings information (New England Fishery Management Council'). Because of this lack of dif- ferentiation of species in landings, fluctuations in stock size Sulikowski et al.: Age and growth estimates of Leucora/a ocellata 411 1200 ■ c 1000 • sngth (mm) 0 i J Jrf ' J^^g i O O Female - 600 ■ S o 1- Q B'^ » Male 400 ■ ll" 200 : ^^ 1 ' 0 2 4 6,8 10 12 14 16 18 Age (years) Figures 4 (continued) 20 Table 3 Comparisor of von Bertalanffy growth parameters for several skate species. Scientific name Sex L„ (mm) k t, (years) Max age (yr) Source Raja rhina 90- 1047 (TL) 0.17 -0.16 13 Zeiner and Wolf 1993 Raja microoceltata 90- 1370 (TL) 0.086 -3.009 9 Ryland and Ajayi, 1984 Raja montagui 90- 978 (TL) 0.152 -1.719 7 Ryland and Ajayi, 1984 Raja erinacea 90- 527 (TL) 0.352 -0.449 8 Waring, 1984 Raja wallacei 90- 422 (DW) 0.26 -0.17 15 Walmsley-Hart et al., 1999 Raja clauata 90- 1050 (TL) 0.215 0.045 10 Brander and Palmer 1985 Raja pullopunctata o- 771 (DW) 0.05 -2.20 18 Walmsley-Hart et al., 1999 Raja pullopunctata 9 1327 (DW) 0.08 -1.95 14 Walmsley-Hart et al., 1999 Leucoraja ocellata 90- 1314 (TL) 0.064 -1.531 19 This study will be difficult to detect and successful implementation of fisheries management plans will remain problematic. Our study provides some basic age and growth parameters for the winter skate and it supports the hypothesis that L. ocel- lata, like other elasmobranchs, require conservative man- agement because they grow slowly and are susceptible to overexploitation ( Brander, 1981; Kusher et al., 1992; Zeiner and Wolf, 1993; Frisk et al., 2001). Acknowledgments We thank Captain Joe Jurek of the FV Mystique Lady for the collection of skates. We also thank Noel Carlson for maintenance of the fish at the U.N.H. Coastal Marine Laboratory and Charles Walker for use of his equipment. This project was supported by a University of New Hamp- shire Hubbard Endowment Fund and the U.N.H. Center for Marine Biology. Literature cited Beamish, R. J., and D. A. Fournier 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Bigelow, H. B., and W. C. Schroeder 1953. Fishes of the Gulf of Maine. Fish. Bull. 53:63-65. Brander, K. 1981. Disappearance of common skate Rata batis from Irish Sea. Nature 290 (5801):48-49. Brander, K.. and D. Palmer 1985. Growth rate of Raja clavata in the Northeast Irish Sea. J. Cons. Ciem.42(2):125-128. Branstetter, S., and J. A. Musick. 1984. Age and growth estimates for the sand tiger in the north- western Atlantic ocean. Trans. Am. Fish. Soc. 123:242- 254. Brown, C. B., and S. H. Gruber 1988. Age assessment of the lemon shark, Negaprion bre- virostris, using tetracycline validated vertebral centra. Copeia 3:747^753. 412 Fishery Bulletin 101(2) Cailliet, G. M. 1990. Elasmobranch age determination and verification; an updated review. In Elasmobranchs as living resources: advances in the biology, ecology, systematics and the status of the fisheries (H. L. Pratt Jr, S. H. Gruber, and T. Tanuichi eds), p. 157-165. U.S. Dep. Commer., NOAA Technical Report, NMFS 90. Cailliet, G. M., L. K. Martin, D. Kusher. P. Wolf, and B. A. Weldon. 1983. Techniques for enhancing vertebral bands in age esti- mation of California elasmobranchs. U.S. Dep. Commer., NOAA Tech. Report NMFS 8:157-165. Cailliet, G. M., R. L. Radtke and B. A. Weldon. 1986. Elasmobranch age determination and verification. In Indo-Pacific fish biology: proceedings of the second interna- tional conference on Indo-Pacific fishes (T. Uyeno, R. Aral, T. Taniuchi and K. Matsuura, eds.), p. 345-360. Icthyol. Soc. Jpn., Tokyo. Cailliet, G. M., and S. Tanaka. 1990. Recommendations for research needed to better understand the age and growth of elasmobranches. In Elasmobranchs as living resources: advances in the biol- ogy, ecology, systematics and the status of the fisheries (H. L. Pratt Jr., S. H. Gruber and T. Tanuichi, eds), p. 505-508. U.S. Dep. Commer., NOAA Technical Report NMFS 90. Casey, J. G., H. L. Pratt, and C. E. Stillwell. 1985. Age and growth of the sandbar shark (Carcharhinus plumbeus ) from the western North Atlantic. Can. J. Fish. Aquat. Sci. 42{5):963-975. Casey, J. M., and R. A. Myers. 1998. Near extinction of a large widely distributed fish. Science 28:690-692. Dulvey N. K., J. D. MetCalfe, J. Glanville, M. G. Pawson, and J. D. Reynolds. 2000. Fishery stability, local extinctions, and shifts in com- munity structure in skates. Cons. Biol. 14( 1 ):283-293 Frisk, M. G. 2000. Estimation and analysis of biological parameters in elasmobranch fishes and the population dynamics of the little skate Raja erinacea, winter skate R. ocellata and Barndoor skate R. laevis. Masters thesis, 167 p. Center for Environmental Science, Univ. Maryland, College Park, MD. Frisk, M. G., T. J. Miller, and M. J. Fogarty 2001. Estimation and analysis of biological parameters in elasmobranch fishes: a comparative life history study. Can. J. Fish. Aquat. Sci. 58(5);969-981. Hoenig, J. M., and S. H. Gruber. 1990. Life history patterns in the elasmobranchs: implica- tions for fisheries management. In Elasmobranchs as living resources: advances in the biology, ecology, system- atics and the status of the fisheries (H. L. Pratt Jr, S. H. Gruber and T. Tanuichi eds), p. 1-16. U.S. Dep. Commer., NOAA Technical Report, NMFS 90. Holden, M. J., and M. R. Vince. 1973. Age validation studies on the centra of Raja clauata using tetratcycline. J. Cons. Int. Explor Mer 35:13-17. Holden, M. J. 1977. Elasmobranchs. In Fish population dynamics (J. A. GuUand, od.i, p. 187-214, J. Wiley and Sons, London. Johnson. G. F. 1979. The biology of the little skate. Raja erinacea, in Block Island Sound, Rhode Island. Masters thesis, 119 p. Univ. Rhode Island. Kingston, RI. Junquera, S., and X. Paz. 1998. Non-traditional resources: skate fishery and survey results in Division 3 NO. Sci. Counc. Res. Doc. NAFO, no. 98/26, 6 p. Northwest Atlantic Fisheries Organization, Dartmouth, Nova Scotia [Canada]. Kusher, D. I., S. E. Smith, and G. M. Cailliet. 1992. Validated age and growth of the leopard shark, T^iakis semifasciata, with comments on reproduction. Environ. Biol. Fish. 35:187-203. Martin L. K., and G. M. Cailliet. 1988. Age and growth determination of the bat ray, Mylioba- tis californica, in central California. Copeia 3:762-763. Martin, L., and G. D. Zorzi. 1993. Status and review of the California skate fishery. In Conservation biology of elasmobranchs (S. Branstetter, ed.), p. 39-52. U.S. Dep. Commer., NOAA Tech. Report NMFS 115. Marquardt. D. W. 1963. An algorithm for the least squares estimation of non- linear parameters. J. Soc. Ind. Appl. Math. 2:431-441. Natanson, L. J. 1993. Effect of temperature on band deposition in the little skate. Raja erinacea. Copeia 19931 1 ):199-206. Natanson, L. J., G. C. Casey, and N. E. Kohler 1995. Age and growth estimates for the dusky shark, Car- charhinus obscurus. in the western North Atlantic Ocean. Fish. Bull. 93(1):116-126. Otwell, W S., and T. C. Lanier. 1979. Utilization of North Carolina skates and rays. Compl. Rep. N.C. Div. Mar Fish. February, 49 p. North Carolina Department of Natural Resources and Community Devel- opment, Morehead City, NC. Prager, M. H., S. B. Saila, and C. W. Recksiek. 1987. Fishparm: a microcomputer program for parameter estimation of nonlinear models in fishery science. Tech. Rep. (87-10):l-37. Old Dominion University Norfolk VA. Robins, C. R., and G. C. Ray 1986. A field guide to Atlantic coast fishes of North America, 354 p. Houghton Mifflin Company, Boston, MA. Ryland, J. S., and T. O Ajayi. 1984. Growth and population dynamics of three Raja species (Batoidei) in Carmarthen Bay, British Isles. J. Cons. Int. Explor Mer 41:111-120. Simon, J. E., and K. T. Frank. 1996. Assessment ofthe Division 4VsW Skate Fishery. DFO Atl. Fish. Res. Doc. 96/105, 6 p. Department of Fisheries and Oceans, Dartmouth. Canada. Simpfendorfer, C. A. 1993. Age and growth of the Australian sharpnose shark. Rhizoprionodon taylori. from north Queensland. Australia. Environ. Biol. Fi.sh. 36(3):233-241. 2000. Age and growth of the whiskery shark. Furgaleus macki. from southwestern Australia. Environ. Biol. Fish. 58:335-343. Sosebee, K. 1998. Skates. Status of fishery resources off the northeast- ern United States. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NE. 115:114-115. von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquires of growth laws II). Human Biology 10:181-183. Walmsley-Hari, S. A., W. H. H. Sauer, and C. D. Buxton. 1999. The biology ofthe skates Raja wallacei and R. puUo- punctata (Batoidea: Rajidae) on the Agulhas Bank, South Africa. S. Afr. J. Mar. Sci. 21:165-179. Waring, G. T. 1984. Age, gi-owth and mortality ofthe little skate off the northeast coast of the United States. Trans. Am. Fish, Soc. 113:314-321. Sulikowski et al : Age and growth estimates of Leucora/a ocellata 413 Winemiller, K. O., and K. A. Rose. 1992. Patterns of life history diversification in North Ameri- can fishes: imphcation for population regulation. Can. J. Fish. Aquat. Sci. 49:2196-2218. Wintner, S. P. and G. Cliff. 1996. Age and growth determination of the blacktip shark, Carcharhinus limbatus, from the east coast of South Africa. Fish. Bull. 94:135-144. Zeiner, S. J., and P G. Wolf 1993. Growth characteristics and estimates of age at matu- rity of two species of skates iRaja binoculata and Raja rhina) from Monterey Bay, California. In Conservation biology of elasmobranches (S. Branstetter, ed.) p. 87-90. U.S. Dep. Commer, NOAA Technical Report NMFS 115. 414 Abstract— Large (>458 mm) striped bass {Morone saxatilis) are dominant predators in Chesapeake Bay. In recent years, the Chesapeake Bay stock of striped bass has increased dramatically, raising concerns about their predatory impact and their forage requirements. In response to these concerns and the need for more recent ecological stud- ies, this investigation was conducted to characterize feeding habits of large striped bass in Chesapeake Bay. Stom- ach contents from 1225 striped bass from 458 to 1151 mm TL were exam- ined in the spring and fall of 1997 and 1998. Striped bass consumed 52 differ- ent species of vertebrates and inverte- brates; however, only a few species of clupeoid and sciaenid fishes dominated diets across both the seasons and size ranges of striped bass examined. Of finfish species, menhaden \Brevoortia tyrannus) was the dominant prey in most areas and gizzard shad (Doro- soma cepedianum) replaced menhaden in importance in lower salinity waters. Spot [Leiostomus xanthurus) and other sciaenid fishes and anadromous herrings t.Alosa spp.) also contibuted large percentages of striped bass diet. Although pelagic schooling fishes formed the majority of the diet, benthic fishes contributed a higher percentage to the diet than in previous studies of striped bass diet composition. Diet composition of large striped bass iMorone saxatilis) in Chesapeake Bay* John F. Walter III Herbert M. Austin Virginia Institute of Marine Science School of Marine Science The College of William and Mary PO Box 1346, Gloucester Point, Virginia 23062 E mail address (for J, F Walter): ifwalterfiivims edu Manuscript accepted 22 October 2002. Manuscript received 9 January 2003 at NMFS .Scientific Publications Office. Fish. Bull. 101:414-423 (2003). Along the Atlantic coast of North Am- erica, the striped bass is one of the most important commercial and recreational fishes (Richards and Rago, 1999). In the face of intense overfishing, the Atlantic Coast population of striped bass experienced drastic declines in the 1970s (Field, 1997; Richards and Rago, 1999). During these periods of intense harvesting, smaller fish dominated the stock composition and the fishery (Koo, 1970). With the relaxation of fishing pressure and the implementation of regulations designed to protect older age classes, populations rebounded to the point where, currently, large, older fish comprise a high percentage of the population (Richards and Rago, 1999). The increased abundance of large striped bass has raised concerns over both the predatory impact and prey needs of this large population of seasonally abundant species in Chesa- peake Bay. Within Chesapeake Bay, historically a center of striped bass abundance and one of the largest sources of juvenile production for the Atlantic coast (Mer- riman, 1941; Berggren and Lieberman, 1978; Kohlenstein, 1981), striped bass are seasonally abundant upper trophic level predators. Chesapeake Bay striped bass are partitioned into a resident, primarily male or juvenile, group of fish found year-round and a migratory group consisting of older, larger (>711 mm total length) and of- ten primarily female fish found in the spring and fall (Chapman, 1987). The Atlantic States Marine Fisheries Com- mission manages fish greater than 711 mm (28 inches) total length as migra- tory (ASMFC) because the majority of these fish leave Chesapeake Bay and migrate throughout the Atlantic coast. Striped bass within Chesapeake Bay migrate during the spring when mature fish ascend tidal freshwater tributaries to spawn (Chapoton and Sykes, 1961; Dorazio et al., 1994). After spawning, these fish leave Chesapeake Bay and migrate northward along the Atlantic coast, returning to Chesapeake Bay in large numbers during the fall. With a major peak in March-April and a minor peak in October-November, the histori- cal landings data reflect the migratory behavior and seasonal abundance of larger fish (Koo, 1970). Diet studies represent the first step in determining the magnitude and di- rection of trophic interactions and are essential data for the management of both predators and prey (Livingston, 1985). For the management of multi- species fisheries, detailed information on fish food habits is required in order to account for the temporal, spatial, and ontogenetical nature of trophic interactions (Walters et al., 1999; Hol- lowed et al., 2000; Wliipple et al, 2000). Although the feeding habits of resident juvenile and early adult striped bass have received considerable study in Chesapeake Bay (Hollis, 1952; Markle * Contribution 2507 of the Virginia Insti- tute of Marine Science, School of Marine Science, The College of William and Mary, Gloucester Point, VA 23062. • ASMFC (Atlantic States Marine Fisheries Commission I. 2000. Public informa- tion document for Amendment 6 to the Interstate Fishery Management Plan for striped bass, 17'p. ASMFC, 1444 Eye Street NW. Washington, DC 20005. http: //www.jcaa.org/I^ID.htm (March 2001). Walter and Austin: Diet composition of Morone saxalilis in Chesapeake Bay 415 and Grant, 1970; Setzler et al., 1980; Boynton et al., 1981; Limburg et al., 1997; Hartman and Brandt, 1995a) and in other locations (Schaefer, 1970; Manooch, 1973; Rulifson and McKenna, 1987), no studies have in- cluded enough specimens larger than 600 mm total length to adequately characterize the diet of migratory fish. The absence of dietary informa- tion for these larger striped bass may have been due to the difficulty in sampling larger striped bass and also to the relative scarcity of large striped bass in Chesapeake Bay during times of severe overfishing (Koo, 1970). Nevertheless, the absence of diet data represents a gap in our knowledge of the trophic dynamics of large striped bass that form the major portion of the spawning stock, are prized fisheries targets and, through successful fish- eries management, have emerged as a significant seasonal predatory force within Chesapeake Bay. We specifi- cally address the diet composition of large (458-1151 mm) striped bass in Chesapeake Bay to determine the im- portant species in their diet during the spring and fall periods of abundance. James River '-.j.^ yi^' ^' ii^V-V ir, AflanBc Ocean '4- 100 7S'0O- 100 74'.S0- 200 Kilometers Figure 1 Map of Chesapeake Bay showing spatial distribution of striped bass samples from March 1997 to May 1998. Methods From March 1997 to May 1998, 1225 striped bass were collected from various localities in Chesapeake Bay, its Virginia tributaries, and the Chesapeake Bay mouth (Fig. 1). Fish were collected from recreational fishermen, charterboat captains, and seafood dealers, as well as from scientific monitoring programs in the spring (48.5%) and fall (51.5%), corresponding to seasonal migration patterns and fishing seasons. Fish ranged in size from 458 to 1151 mm TL (mean 653.7mm) and were 0.91-17.6 kg in weight (mean 3.69 kg). Hook-and-line gear, gill nets, fyke nets, and otter trawls were used to capture fish. Fish captured in pound nets were excluded from this analysis because of complications introduced by the confinement of the fish in pound nets. Fish captured by hook and line were recorded as such in order that the bait and chum used with this gear could be excluded from the diet analyses. Total length (±1.0 mm), sex, and weight (±0.001 kg wet weight) were recorded for each fish, as well as location, date, and method of capture. Stomachs were removed by cutting the alimentary canal anterior to the stomach and posterior to the pylorus and the contents were frozen until processed. In some cases, stomachs of fish donated by charterboat captains and recreational fishermen were removed by the fisherman. Fish stomachs were thawed and emptied, and their contents were blotted dry and weighed. Contents were sorted and identified to the lowest possible taxon, weighed, counted, and measured. Diet composition was analyzed by using three measures described in Hyslop (1980): percent frequency of occurrence, percent weight, and percent number These values were combined to give an index of relative importance (Pinkas et al., 1971). The index of relative importance for a particular prey category i (IRI,) is expressed as IRl^ = (%N+ %W) X %F, where %N = the percentage of a prey species by number; %W - the percentage of a prey species by weight; and %FO = the percent frequency of occurrence of a prey species. IRI values were calculated as percent IRI values (Cortes, 1997 ). In calculating IRI values, we excluded several items appearing in the stomachs, such as chum (ground men- haden), bait, trash and plant material because they were deemed to be non-naturally occurring food items. Several prey species were combined either because of difficulties in identification of partially digested prey to species or because of ecological or taxonomic similarity. Both bay anchovy iAnchoa mitchilU) and striped anchovy (Anchoa 416 Fishery Bulletin 101(2) hepsetus) were combined into a single-prey category. In addition, gizzard and threadfin shad (Dorosoma cepedia- num and D. petenense), and blueback and alewife herring (Alosa aestivalis and A. pseudoharengus) were treated as single-prey categories. Unidentified prey consisted pri- marily of unidentified fish remains and were recorded as such. Striped bass were categorized by fish length and month of capture. Fish were partitioned into two size classes cor- responding to mixed resident and migratory fish (458-710 mm total length) and coastal migrant fish (711-1255 mm total length) based on the Atlantic States Marine Fisher- ies Service classification of fish 711 mm and above as fully recruited to the coastal migratory stock. For spatial analysis of feeding habits, each fish was placed into one of two salinity regimes; tidal freshwater (0-5 ppt) or meso- haline waters (6-28 ppt). Tidal freshwater- waters include the upper reaches of the James, York, Rappahannock, and Potomac rivers. Mesohaline waters include the open waters of Chesapeake Bay and the lower reaches of most rivers. No fish were collected in the fall from tidal freshwater For both monthly and spatial analyses, diet was quantified by weight only. To measure intensity of feeding, a stomach fullness index (SFI) was calculated according to Hureau ( 1969); „^, Stomach coiuen! wt'iiilu Sri = X 10. Fish weight SFI values were calculated for all fish regardless of the presence or absence of stomach contents. A regression of striped bass total length versus prey total length was fitted by least-squares linear regression of the untransformed values. Prey lengths were reconstructed from partially digested backbones by using regressions of backbone length on total length obtained from samples col- lected in 1998 by the authors and those given by Hartman and Brandt (1995a). Results Of the 1225 striped bass examined, 688 (56%) contained items in the stomachs (Table 1). Thirty-four different spe- cies offish and 18 species of invertebrates were observed in the diet. Overall, clupeid fishes dominated the diet and men- haden, in particular, accounted for 44';'( of the weight and occurred in 18% of all stomachs (Table 2). Menhaden ranged in length from 103 to 360 mm total length. A '/r IRI value of 58.3 for menhaden was higher than that for all other species combined. Anchovies were numerically the most abundant (229^ ) of all prey items and were equal to spot (Leiostomus xanthiirus) in '7( IRI, both sharing a value of 12.3. Other prey in order of decreasing '^r IRI were gizzard shad (genus Doro- soma ) with a ''/i IRI of 6.7, and blue crab (C'a//(>iectes saptdus} with %IRI values of 3.4. Atlantic croaker (Micropogonius undulatus) and summer flounder iParaUchthys dentatus) had '''( IRl values of 1.1 and 1.0, respectively. All other prey categories had 'MRI values <1 and ap- peared relatively unimportant in the overall diet of striped bass, although some increased in relative importance at certain times and locations. Invertebrates were relatively minor constituents of the overall diet of large striped bass, providing only 4.4% of the total IRI. In contrast, clupeid fishes contributed 65% of the IRI and both sciaenid and engraulid fishes combined contributed over 25% of the total IRI. Clear seasonal and spatial patterns in diet corresponded with the migratory behavior of large striped bass. Striped bass in both sizes classes, 458-710 mm and 711 mm and above, migrated into tidal freshwater to spawn in the months of March, April, and May. Striped bass fed in the tidal freshwater region, although at a reduced intensity as evidenced by the lower stomach fullness values and the lower percentages of nonempty stomachs compared to those at other times and locations (Table 1). Gizzard shad, white perch (Morone americana), and anadromous herrings (Alosa pseudoharengus and Alosa aestivalis) were the main constituents of the diet of both sizes of striped bass in the tidal freshwater region (Table 3, Fig. 2). During spring, striped bass also pass through the meso- haline waters of Chesapeake Bay prior to and after spawn- ing, during which time they feed fairly heavily as indicated by higher than average stomach fullness values and per- centages of nonempty stomachs (Table 1). Approximately 83% of the striped bass sampled from mesohaline waters during this time had food in the stomachs indicating active feeding during the pre- and postspawning migration. Men- haden dominated the diets by weight of both size classes of striped bass from mesohaline waters in the spring. Striped bass of both size classes also consumed croaker, blue crab, and white perch (Table 3, Fig. 2); however, the size classes differed in that smaller fish consumed bay anchovy and juvenile spotted hake ( Urophycis regia ) and larger striped bass consumed anadromous herrings. Large striped bass are generally absent from Chesapeake Bay in significant numbers in the summer and return in the fall to mesohaline waters of Chesapeake Bay and its lower tributaries. The fall return is essentially a feeding migration and the high stomach fullness values and high percentages of nonempty stomachs (Table 1 ) indicate active feeding. Striped bass of both size classes fed predominantly upon menhaden, which had percent weight values between 53% and .58 Ve (Fig. 3). Sciaenid fishes, including spot, At- lantic croaker, and weakfish iCynoscion regalis) combined provided between 23"^^ and '3l"i of the diet by weight for both size classes of fish. Notable differences occurred in the high percentage of summer flounder (Paralichthys dentatus) found in the diets of larger striped bass (\b% by weight) and in the high percentage of both butterfish (Peprilus triacanthus, 4%) and gizzard shad (11%) found in the diets of smaller fish (Fig. 3). The only invertebrates found in abundance in the diets during this time were blue crabs, which contributed 70"^; of the diet by weight for the smaller size class of striped bass in September (Table 3). The greatest number of species occurred in the diet in fall with forty-four different species of prey items observed, although many were isolated occurrences of rare prey and only a few species contributed to the overall diet at this time. Walter and Austin: Diet composition of Morone saxatilis in Chesapeake Bay 417 Table 1 Distribution of striped bass collections by month with location, capture method. percentage of nonempty (% full) stomachs, and | stomach f uUness index. % Stomach Standard Month Location Method Total full fullness index deviation Striped bass, 458-710 mm total 1 ength Feb Potomac River gill net 14 64.3'7f 1.13 1.73 Mar York, Rappahannock, James River gill net, fyke net 116 47.47r 0.36 0.79 Apr York, Rappahannock James River gill net, fyke net 159 25.2% 0.38 1.52 May Upper York River electroshock 28 71.4% 1.15 1.80 Jun Middle Bay gill net, hook and line 77 93.5% 4.85 3.87 Sep Middle Bay hook and line, gill net 74 27.0% 0.30 0.62 Oct Lower Bay hook and line, gill net 245 58.4% 1.06 1.93 Nov Lower Bay hook and line, gill net 114 74.6% 2.08 3.08 Dec Lower Bay hook and line, gill net, trawl 12 91.7% 1.48 1.24 Striped bass 711-1255 mm total length Mar York, Rappahannock James River gill net, fyke net 12 50.0% 0.31 0.69 Apr York, Rappahannock, James River gill net, fyke net 85 31.8% 0.60 1.44 Mav Upper York River electroshock 7 85.7% 0.82 1.74 Jun Middle Bay hook and line 66 81.8% 2.75 2.14 Sep Middle Bay hook and line, gill net 20 25.0% 0.21 0.24 Oct Lower Bay hook and line, gill net 45 42.2% 0.71 1.52 Nov Lower Bay hook and line, gill net 95 74.7% 1.69 2.80 Dec Lower Bay hook and line, gill net 56 80.4% 1.23 1.66 Total all all 1225 56.1% 1.00 2.03 Table 2 Stomach contents of striped bass from Chesapeake Bay, 1997-98 (n =688, total number of stomachs with quantified contents). % frequency %by Weight %by Prey Occurrences of occurrence Number number in grams mass %IR1 Class Osteichthyes Clupeidae Brevoortia tvrannus 132 20.63 319 18.11 14757.03 44.40 58.34 Alosa spp. 7 1.09 20 1.14 977.38 2.94 0.20 Dorosoma spp. 43 6.72 142 8.06 4623.73 13.91 6.68 Unknown clupeid 18 2.81 21 1.19 134.56 0.40 0.20 Moronidae Morone saxatilis 1 0.16 1 0.06 19.46 0.06 0.00 Morone americana 19 2.97 24 1.36 750.09 2.26 0.49 Sciaenidae Leiostomus xanthurus 86 13.44 179 10.16 3315.84 9.98 12.25 Bairdielta chrvsoura 13 2.03 17 0.97 244.61 0.74 0.16 Cynoscion regalis 15 2.34 19 1.08 835.62 2.51 0.38 Micropogonias undulatiis 20 3.13 21 1.19 2123.82 6.39 1.07 Unknown scieanid 14 2.19 21 1.19 61.41 0.18 0.14 continued A significant relationship between striped bass total length and prey total length (P<0.05, r'-=0.26) was ob- served which indicated that larger and older striped bass ate larger prey (Fig. 41. The fit of the regression was poor, indicating that, although larger striped bass did consume larger prey, they also consumed smaller prey. 418 Fishery Bulletin 101(2) Table 2 (continued) Prey Occurrences '7f frequency % by Weight % by of occurrence Number number in grams mass %m\ Engraulidae Anchoa spp. Other fish Paralichthys dentatus Membras martinica Menidia menidia Anguilla rostrata Symphurus plagiusa Pepritus triacanthus Urophycis regia Notropis spp. Trinectes maculatus Pomatomus saltatrix Eucinostomus argenteus Gobiosoma bosc Synodus foetens Strongylura marina Scophthalmus aquosus Mugil curema Sphoeroides maculatus Hypsoblennius hentzi Fundulus heteroclitus Unidentified fish remains Class Crustacea Callinectes sapidus Neomysis americana Squtlla empusa Ovalipes ocellatus Lironeca ovalis Callinectes spp. Penaeus setiferus Crangon septemspinosa Paleamonetes pugio Cancer irroratus Upogebia affinis Class Bivalvia Mytilus edulis Crossostrea virginica Class Gastropoda All gastropods Class Polychaeta All polychaetes Class Hydrozoa All hydroids Phylum Porifera All sponges Miscellaneous items Chum (ground menhaden) Bail I menhaden, spot, etc) 74 17 1 12 10 9 6 3 3 5 3 3 1 2 56 55 13 23 13 6 4 5 3 4 1 1 11.56 2.66 0.16 1.88 1.56 1.41 0.94 0.47 0.47 0.78 0.47 0.47 0.16 0.31 0.16 0.16 0.16 0.16 0.16 0.16 8.75 8.59 2.03 3.59 2.03 0.94 0.63 0.78 0.47 0.63 0.16 0.16 0.16 0.63 0.31 0.16 399 30 15 25 21 40 12 26 5 5 3 3 1 2 3 129 90 35 15 6 7 5 5 9 1 1 22.66 1.70 0.85 1.42 1.19 2.27 0.68 1.48 0.28 0.28 0.17 0.17 0.06 0.11 0.17 0.06 0.06 0.06 0.06 0.06 4.03 7.33 5.11 1.99 0.85 0.34 0.40 0.28 0.28 0.51 0.06 0.06 0.06 0.23 0.11 0.06 256.29 0.77 0.39 0.00 7.92 0.02 0.00 0.00 2.29 0.01 12.26 2256.59 6.79 1.02 26.17 0.08 0.01 27.00 0.08 0.13 544.48 1.64 0.20 111.59 0.34 0.17 385.88 1.16 0.08 400.00 1.20 0.06 8.45 0.03 0.01 23.56 0.07 0.01 184.21 0.55 0.02 39.92 0.12 0.01 0.10 0.00 0.00 68.54 0.21 0.00 67.96 0.20 0.00 14.42 0.04 0.00 36.08 0.11 0.00 4.80 0.01 0.00 4.15 0.01 0.00 3.39 0.01 0.00 128.37 0.39 1.75 439.81 1.32 3.36 11.09 0.03 0.47 174.26 0.52 0.41 103.68 0.31 0.11 0.54 0.00 0.01 28.83 0.09 0.01 13.00 0.04 0.01 1.34 0.00 0.01 2.24 0.01 0.01 7.73 0.02 0.00 0.59 0.00 0.00 2.00 « ** 2.00 * ** 159 28 0.00 0.01 0.00 0.00 continued Walter and Austin: Diet composition of Morone saxatilis in Chesapeake Bay 419 Tidal Freshwater, 458-7 10 mm (« =43) menhaden unk clupeid ,„, 2% ' '» Tidal Freshwater. 711 mm and above (n=39) white perch Mesohaline, 458-710 mm (H=70) gizzard shad 89% Mesohaline, 711 mm and above ("=12) river herring blue crab 40/^ 11% croaker / 19% / spotted hake \';////,';,';,z:';' white perch ^s'W4<~^~^^^~^^^~^~^~^^^-^^^>^^^ Figure 2 Percentage by weight of prey in the diets of striped bass captured in the spring (February-June). Note that only stomachs with contents other than bait were used in the construction of these figures. Table 2 (continued) % frequency % bv Weight 'Xf by Prey Occurrences of occurrence Number number in grams mass %IRI Miscellaneous items (cont.) Plant material 11 * * * « * ** Woody material 6 * * * « * ** Plastic trash 1 * * * * * ** Cigarette butts 2 * * * * * ** Stones, gravel 2 * * * * * ** Feathers 2 * + * * * ** * Not quantified. ** Not included in IRI ca culations. 420 Fishery Bulletin 101(2) tfl Hi 3 "35 o o O o o O in c^ •* O o o o O CO rH tC d d d d d in d d d d d d d d ■^ CM d _> 1-H CM M „ oj -i; d. o in in o o 00 en 1-H i> o o o o o t^ CO CO a o a d t^ d d d CO d d d d d d d d in CO d cfl c w fN CM CO 0) '=C C 0; k^ a. 0) c 4 1- r3 CO [/: OJ QJ N ;-. CO u — QJ o d oi q CN O d o d 1— ( CO d o o d o d o d o d O d o d CO CD CM rH rH d a S o CU > CO c M 00 02 1 u o X CO o d CO irj o o d o d eg in 00 q CO CO d CD d o d o d CO d 02 in LO CM I— ' o ■—I c^ •— t ^H -^ >^ CO oa o o CD o o o o q Tt; c^ o o o o o [r^ 00 00 a d r-i d d d d ,-i t>^ CO d d d d d iri CD d OJ M (N CM CO a a w -^ ^ to tt; u .2 i a o d o I— 1 CD I— 1 1—1 ■* ■* in o d o d l-H d o d o d CO CO d CM CM O d d c: :^ CO tH CO •-H « a 03 O -Q -a o a C3 'C « d. o o CTl t^ o o o o o o ir; 1-H t^ o o o o X O Q. d d CC CO d d d d d d CO C-^ CO d d d d ^ CO o t^ o c: ^j X tx o JIX .ti o C35 T)" o CD ^ »-H o o •— ' 1-H o •-* CM o o o ^ ^ Q; d CO t^ d d d d d d d lO d i—i CO d d d ^ a lO •"* C^ .— t CM c a* o bi OJ a o ^ s o d ?— 1 CO ■^ o o o o CO o in "* CM o O o CM O a m c-^ CO d d d oq 1-H d CO d CM d d d CM d i. CO ■^ CO C^i CM CJJ o CM 'x c3 o X J3 ft c c 0} ^ c _2 0; ■a CO CO ■3 ts o f-* o CT> !>J -* o 05 ^ o o o O (J) in t^ rH CO c JS o S g d (N d d 00 d 00 iri ■^ ■*-* d d d 1-^ CO ■^ CO CO s t^ CO in ^ 6 6 CO r^ CM Tf f- B a> o Ui !/l 1 i2 rH r-4 1 i^ rH I> CA CO s ? in o o 55 s CO 00 i-H ^H CO J3 £ CM CO 5- in 5 5 T-H rH .^5 "V -a c o a < c a -4-> O z a ■c CO a < CO s c 3 ■-3 a o > O ♦J s 55 M Discussion Our study addresses the diet of striped bass above 458 mm total length in Chesapeake Bay. In previous studies of striped bass diet (Hollis, 1952; Hartman and Brandt, 1995a) in Chesa- peake Bay and adjacent waters (Manooch, 1973), few fish above 458 mm were sampled. The comprehensive work by Hartman and Brandt ( 1995a) did not include fish above age 6. The current study focuses specifically on the diet of larger striped bass that previously were undersampled or were rare during peri- ods of severe overfishing (Koo, 1970). Throughout the two size ranges of striped bass sampled and in both seasons and loca- tions, schooling fishes dominated the diets in Chesapeake Bay. In particular, clupeid fishes (menhaden, gizzard shad) and the closely related anchovies exceeded all other prey spe- cies in frequency of occurrence, number, and biomass. Among other fishes, only spot rivaled the clupeids and anchovies in overall impor- tance; however, white perch, croaker, weakfish, and summer flounder contributed important percentages of the diet in certain seasons. Hol- lis (1952), Manooch (1973) and Hartman and Brandt ( 1995a ) and Overton ( 2002 ) also found that schooling clupeoid fishes formed the ma- jority of the diets of striped bass from Chesa- peake Bay and nearby Albemarle Sound. There was a shift in the relative importance of smaller schooling fishes (anchovies) in striped bass 458-710 mm to larger schooling fishes (menhaden, gizzard shad) in striped bass 711-1151 mm. Although there was a tendency for larger striped bass to consume larger prey, this relationship should more accurately be described as one where larger striped bass have a greater size range of prey to consume (Fig. 4). The largest striped bass consumed prey ranging from several millime- ters up to 400 mm in total length, correspond- ing to 40'^( of their total length and equaling the ratio of mean maximum forage length to striped bass length found by Manooch ( 1973). Similarly, smaller striped bass consumed prey that approached 40'~f of their total length; however, most prey consumed by all sizes of striped bass were smaller, young-of-the-year fishes — a finding corroborated by Overton (2002), who predicted an optimal prey size to be 21% of the striped bass length. The predominance of fish in adult striped bass diets attests to the piscivorous nature of larger striped bass and corroborates the find- ings of other studies (Hollis, 1952; Manooch, 1973; Overton, 2002). Hartman and Brandt (1995a) and Gardinier and Hoff (1982) ob- served an ontogenetic shift at 200 mm TL Walter and Austin: Diet composition of Morone saxatilis in Chesapeake Bay 421 Mesohaline, 458-710 mm (n=2.')9) croakerweakfisti butterfisri 2% 2% 4% Mesohaline, 711-1151 mm (H=14()) eel gizzard shad weakfish ,% 6% menhaden 58% flounder 15% Figure 3 Percent by weight of prey in the diet of striped bass captured in the fall (September-December). Note that only stomachs with contents other than bait were used in the construction of these figures. 450 -\ M)0 - Ql 150 from invertebrate to vertebrate prey in the diet of smaller striped bass. In the present study, we sampled size ranges above 458 mm and found no clear ontogenetic dietary shift between vertebrate and invertebrate prey. Invertebrates, primarily blue crab, constituted a minor percentage of the over- all diet and were significant in the diet only in May and September in mesohaline wa- ters of Chesapeake Bay. This is in contrast to the high percentages of invertebrates found in the diets of large striped bass in New England waters and likely represents latitudinal differences in the availability of fish prey (Nelson et al.^). The seasonal and spatial differences in the diet of striped bass correspond to the behavioral and seasonal migration pat- terns of the fish and reflect changes in the community composition at the location and time of capture. The major seasonal trend is spring feeding on gizzard shad, anad- romous herrings, and white perch, corre- sponding to spawning migrations of both striped bass and their prey into tidal freshwater Many spring samples came from upper river sites where gizzard shad and white perch are year-round residents and herrings are anadromous mi- grants (Murdy et al., 1997). This pattern of spring feeding P<0.05, 1^^0.26 Plot 250 500 750 1000 Striped bass total length (mm) Figure 4 of prey total length against total length for striped bass. 1250 on anadromous herrings and gizzard shads was also found by Trent and Hassler 1966) in the Roanoke River, NC. Striped bass captured in spring from the lower, more sa- line sections of the rivers exhibited high levels of feeding intensity and consumed primarily menhaden, sciaenids, anchovies, and blue (VIMS^) crabs. In the spring, Manooch 2 Nelson, G. A., B. C. Chase and J. Stockwell. 2002. Feeding habits of striped bass {Morone saxatilis) from coastal waters of Massachusetts, 29 p. Massachusetts Department of Marine Fisheries Annisquam River Marine Fisheries Field Station 30 Emerson Ave. Gloucester, MA 019.30. ' VIMS (Virginia Institute of Marine Science). 2002. Juvenile fish and blue crab trawl survey. VIMS, P. O. Box 1346 Glouces- ter Point. VA 23062. http://www.fisheries.vims.edu/vimstrawl data/. (March 20011 422 Fishery Bulletin 101(2) (1973) found menhaden and anadromous herrings to be predominant (Homer and Boynton'') foods in brackish wa- ters of Albemarle Sound and HoUis (1952) found menhaden as well as anchovies and blue crabs to be predominant food of striped bass in brackish waters of Chesapeake Bay. The predatory impact of migratory striped bass depends upon their residence time in these waters, as well as on striped bass population size and feeding rates. Carmichael et al. (1998) estimated that striped bass spend approximately one week in their upstream and one week in their down- stream transit of the Roanoke River There are no estimates of residence time in the open waters of Chesapeake Bay or Albemarle Sound; however, striped bass larger than 711 mm are captured in recreational fisheries in Chesapeake Bay into June, suggesting that they are present in Chesa- peake Bay from March through June. After leaving Chesapeake Bay and summering in New England waters, large striped bass return to the bay in fall (Dorazio et al., 1994) and fed primarily upon menhaden, spot, and anchovies. At this time, most fish were taken from open waters of Chesapeake Bay. In the lower bay during fall, large numbers of transient young-of-the-year (YOY) marine fishes (menhaden, spot, croaker, flounder, and weakfish) congregate in open waters of Chesapeake Bay prior to the fall out-migration, thus making them acces- sible prey for returning striped bass. Striped bass exhibited higher stomach fullness values and higher percentages of nonempty stomachs in November and December than in all other months, with the exception of June. This finding, in conjunction with observations of striped bass aggres- sively pursuing baitfishes in surface waters during the fall (Mollis, 1952, this study), indicates high feeding intensity. In bioenergetic simulations, striped bass growth potential and prey density peaked in October (Brandt and Kirsch, 1993). Because much of the annual growth (Hartman and Brandt, 1995a, 1995b) and gonadal development (Berlin- sky and Specker, 1991) occur in the fall, this period is of primary importance both for the accumulation of body mass for overwintering and for the initial development of gonadal products. Although pelagic fishes, notably anchovy and menhaden, provided the bulk of the diet for large striped bass, this study differs from the diet study of Hartman and Brandt (1995a) and the network analysis of Baird and Ulanowicz (1989) in that benthic fishes also contributed significantly to the diets. Baird and Ulanowicz ( 1989) estimated that striped bass ob- tained 91-100''^( of their diet from pelagic trophic pathways and Hartman and Brandt (1995a) estimated that 68-75% of the diet of age-2 to age-6 striped bass came from pelagic sources. These estimates contrast with the high percentages of benthic spot, croaker, summer flounder, and gizzard shad observed in this study and indicate that larger striped bass either prey to a greater extent upon benthic fishes or the overall diet has shifted towards benthic prey Menhaden and bay anchovy juvenile abundance indices have declined ' Homer. M, and W R. Boynton. 1978. Stomach analysis offish collected in the (lalvert (Miffs region, Chesapeake Bay — 1977. Rep. UMCEES 78-154 CBL, ,!6,i p. Chesapeake Biological Laboratory, Univ. Maryland, Solomons, MD. over the past 10 years (VIMS^) suggesting that a dietary shift towards benthic prey may have occurred since the collections of Hartman and Brandt (1995a) and the stud- ies cited in the Baird and Ulanowicz ( 1989) model. Without comprehensive and systematic annual diet sampling, it is difficult to separate dietary shifts from differences in the sizes of fish sampled or the sampling locations. Baird and Ulanowicz (1989) incorporated diet composition data from Hollis (1952), Gardinier and Hoff (1982), Manooch (1973), and Homer and Boynton'* that included very few striped bass larger than >600 mm and their model included no link- ages between striped bass and gizzard shad, spot, croaker, or summer flounder Furthermore, the absence of gizzard shad in the Baird and Ulanowicz ( 1989) model represents a missing pathway that might link benthic detritus directly to piscivore production as occurs in freshwater impound- ments where gizzard shad are the major prey of striped bass (Mathews et al., 1988) and play a pivotal role in the freshwater ecosystem (Stein et al., 1995). Acknowledgments This work represents part of a thesis presented to the Col- lege of William and Mary (School of Marine Science) by the first author. We would like to thank the first author's committee members, David Evans, Robert Diaz, John Hoenig, and Thomas Munroe, for reviewing the thesis and this manuscript. We would like to acknowledge the many seafood dealers and recreational and commercial fisher- man who provided fish samples. This research was funded by the Virginia Recreational Fishing Advisory Board and the Virginia Commercial Advisory Board (grant numbers RF-97-08 and CF-97-08). Literature cited Baird, D., and R. E. Ulanowicz. 1989. The seasonal dynamics of the Chesapeake Bay eco- system. Eeol. Mono. 59(4):329-364. Berggren, T. J., and J. T. Lieberman. 1978. Relative contribution of Hudson, Chesapeake and Roanoke striped bass, Morone saxatilis, stocks to the Atlan- tic Coast fishery Fish. Bull. 76:335-342. Berlinsky, D. L. .and J. L. Specker. 1991. Changes in gonadal hormones during oocyte develop- ment in the striped bass, Morone saxatilis. Fish Phys. Biochem. 9:51-62. Boynton, W. R., T. T. Polgar, and H. H. Zion. 1981. Importance of juvenile striped bass food habits in the Potomac estuary Trans. Am. Fish. Soc. 110:56-63. Brandt, S. B., and J. Kirsch. 1993. Spatially explicit models of striped bass growth potential in Chesapeake Bay. Trans. Am. Fish. Soc. 122: 845-869. Carmichael, J. T, S. L. Haeseker, and J. E. Hightower, 1998. Spawning migration of telemetered striped bass in the Roanoke River, North Carolina. Trans. Am. Fish. Soc. 127:286-297 Chapman, R. W. 1987. Changes in the population structure of male striped Walter and Austin: Diet composition of Morone saxatills in Chesapeake Bay 423 bass spawning in three areas of the Chesapeake Bay from 1984 to 1986. Fish. Bull. 85:167-170. Chapoton, R. B., and J. E. Sykes. 1961. Atlantic coast migration of large striped bass as evi- denced by fisheries and tagging. Trans. Am. Fish. Soc. 90: 13-20. Cortes, E. 1997. A critical review of methods of studying fish feed- ing based on analysis of stomach contents: an applica- tion to elasmobranch fishes. Can. J. Fish Aquat. Sci. 54: 726-738. Dorazio R. M., K. A. Hattala, C. B. McCollough, and J. E. Skjeveland. 1994. Tag recovery estimates of migration of striped bass from spawning areas of the Chesapeake Bay. Trans. Am. Fish. Soc. 123:950-963. Field, J. D. 1997. Atlantic striped bass management: Where did we go right? Fisheries 22:6-9. Gardinier, M., and T Hoff. 1982. Diet of striped bass in the Hudson River Estuary. N.Y. Fish Game J. 29:152-165. Hartman, K. and S. Brandt. 1995a. Trophic resource partitioning, diets and growth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124: 520-537. 1995b. Predatory demand and impact of striped bass, blue- fish and weakfish in the Chesapeake Bay: applications of bioenergetics models. Can. J. Fish. Aquat. Sci. 52: 1667-1687. Mollis, E. H. 1952. Variations in the feeding habits of the striped bass, Roccus saxatitis, in Chesapeake Bay. Bull. Bingh. Ocean. Coll. 14(1):111-131. Hollowed, A. B., N. Bax, R. Beamish, J. Collie, M. Fogarty, P. Livingston, J. Pope, and J. C. Rice. 2000. Application of multispecies models in assessment of impacts of commercial fishing. Symposium proceedings in the ecosystem effects of fishing. ICES J. Mar Sci. 57: 707-719. Hureau, J. C. 1969. Biologie comparee de quelques poissons anarctiques (Nototheniidae). Bull. Inst. Oceangr. Monaco 68:1-44. Hyslop. E. J. 1980. Stomach content analysis- a review of methods and their application. J. Fish Biol. 17:411-429. Kohlenstein, L. C. 1981. On the proportion of the Chesapeake Bay stock of striped bass that migrates in the coastal fishery. Trans. Am. Fish. Soc. 110:168-179. Koo, T S. Y. 1970. The striped bass fishery in the Atlantic states. Ches. Sci. 11:73-93. Livingston. P. A. 1985. An ecosystem model evaluation: the importance offish food habits data. Mar. Fish. Rev. 47:9-12. Limburg, K. E., M. L Pace, and D. Fischer. 1997. Consumption, selectivity and use of zooplankton by larval striped bass and white perch in a seasonally pulsed estuary. Trans. Am. Fish. Soc. 126:607-621. Manooch, C. S., III. 1973. Food habits of yearling and adult striped bass, Morone saxatilis. from Albemarle Sound, North Carolina. Ches. Sci. 14:73-86. Markle, D., and G. Grant. 1970. The summer food habits of young-of-the-year striped bass in three Virginia rivers. Ches. Sci. 11:50-54. Mathews, W. J., L. G. Hill, D. R. Edds, J. J. Hoover, and T. G. Heger. 1988. Trophic ecology of striped bass, Morone saxatilis, in a freshwater reservoir (Lake Texoma, USA). J. Fish Biol.: 33:273-288. Merriman, D. 1941. Studies on the striped bass (Roccus saxatilis) of the Atlantic Coast. Fish. Bull. 50: 1-77. Murdy, E. O., R. Birdsong, and J. A. Musick. 1997. Fishes of the Chesapeake Bay, 324 p. Smithsonian Institute Press, Washington, DC. Overton, A. S. 2002. Striped bass predator-prey interactions in Chesa- peake Bay and along the Atlantic Coast. Ph.D. diss., 226 p. Univ. Maryland Eastern Shore, Princess Anne, MD. Pinkas, L. 1971. Food habits study. /« Food habits of albacore, bluefin tuna and bonita in California waters (L. Pinkas, M. S. Oliph- ant, and I. L. K. Iverson, eds.), p. 47-63. Calif Dep. Fish Game, Fish. Bull. 152. Richards, R. A., and P. J. Rago. 1999. A case history of effective fishery management: Chesapeake Bay striped bass. N. Am. J. Fish. Manage. 19:356-375. Rulifson, R. A., and S. A. McKenna. 1987. Food of striped bass in the upper Bay of Fundy, Canada. Trans. Am. Fish. Soc. 116:119-122. Schaefer. R. 1970. Feeding habits of striped bass from the surf waters of Long Island. N.Y. Fish Game J. 17:1-17. Setzler, E. M., W. R. Boynton, K. V. Wood, H. H Zion, L Lubbers, N. K. Mountford, P. Frere. L. Tucker, J. A. Mihursky. 1980. Synopsis of biological data on striped bass, Morone saxatilis (Walbaum). U.S. Dep. Commer., NOAA Techni- cal Report NMFS Circ. 433, 69 p. Stein, R. A., D. V. DeVries, and J. M. Dettmers. 1995. Food-web regulation by a planktivore: exploring the generality of the trophic cascade hypothesis. Can. J. Fish Aquat. Sci. 52: 2518-2526. Trent, L., and W Hassler 1966. Feeding habits of adult striped bass, Roccus saxati- lis, in relation to stages of sexual maturity. Ches. Sci. 7: 189-192. Walters, C, D. Pauly. and V. Christensen. 1999. Ecospace: prediction of mesoscale spatial patterns in trophic relationships of exploited ecosystems, with empha- sis on the impacts of marine protected areas. Ecosystems 2:539-554. Whipple, S. J., J. S. Link, L. P. Garrison, and M. J. Fogarty. 2000. Models of predation and fishing mortality in aquatic ecosystems. Fish Fisheries l(l):22-40. 424 Abstract— The tautog, Tautoga onitis (Linnaeus), ranges from Nova Scotia to South Carolina and has become a popular target for recreational and com- mercial fisheries. Although tautog are a multiple spawning species, reproductive potential, measured as annual fecun- dity, has not been estimated previously with methods (batch fecundity, spawn- ing frequency) necessary for a species with indeterminate annual fecundity. A total of 960 tautog were collected from the mouth of the Rappahannock River in the lower Chesapeake Bay to 45 km offshore of Virginia's coastline to inves- tigate tautog reproductive biology in the southern portion of the species range. Tautog did not exhibit a 1:1 sex ratio; 56'7f were females. Male tautog reached 50^1 maturity at 218 mm TL, females at 224 mm TL. Tautog spawned from 7 April 1995 to 15 June 1995. at locations from the York River to 45 km offshore. Batch fecundity estimates ranged from 2800 to 181,200 eggs per spawning for female tautog age 3-9, total length 259- 516 mm. Mean batch fecundity ±SEM for female tautog ages 4-6 was 54,243 ±2472 eggs and 106,256 ±3837 eggs for females ages 7-9. Spawning frequency was estimated at 1.2 days, resulting in 58 spawning days per female in 1995. Estimates of potential annual fecun- dity for tautog ages 3-9 ranged from 160,000 to 10,510,000 eggs. Reproductive seasonality, fecundity, and spawning frequency of tautog {Tautoga onitis) in the lower Chesapeake Bay and coastal waters of Virginia* Geoffrey G. White School of Marine Science Virginia Institute ol Manne Science College of William and fVlary PO. Box 1346 Gloucester Point, Virginia 23062 Present address: Atlantic States Marine Fistienes Commission 1444 Eye Street, NW, 6"^ Floor Washiington, DC. 20005 E-mail address: gwhiteigasmfc org Thomas A. Munroe National Marine Fisheries Service National Systematics Laboratory. NMFS/NOAA Smithsonian Institution Post Office Box 37012 NHB, WC 57, MRC-153 Washington, DC 20013-7012 Herbert M. Austin School of Manne Science Virginia Institute of Manne Science College of William and Mary PO Box 1346 Gloucester Point, Virginia 23062 Manuscript accepted 22 October 2002. Manuscript received 31 December at NMFS Scientific Publications Office. Fish. Bull.:424-424 (2003). The tautog, Tautoga onitis (Linnaeus), ranges from Nova Scotia (Bleakney, 1963; Scott and Scott, 1988) to South Carolina (Sedberry and Beatty, 1989; BeardenM, although it is most abun- dant between Cape Cod and New Jersey (Bigelow and Schroeder, 1953). In Virginia, tautog occur within the Chesapeake Bay from Gwynn's Island (mouth of Rappahannock River) and Sandy Point (Eastern Shore) southward to the mouth of the bay (Hildebrand and Schroeder, 1928), and in coastal Atlantic waters out to 65 km offshore (Richards and Castagna, 1970; Musick, 1972; Hostetter and Munroe, 1993). The major habitat requirement for this spe- cies is hard-bottom structure that fish can remain under, within, or alongside (011a et al., 1974). Adult tautog inhabit hard-bottom environments including natural reefs and rock outcroppings, as well as man-made structures such as jetties, bridge-tunnel networks, arti- ficial reefs, and shipwrecks. Near the southern terminus of the species range suitable hard-bottom habitat to support tautog populations becomes less abun- dant and may limit population size (Eklund and Targett, 1990; Hostetter and Munroe, 1993). Tautog are a long-lived, slow-growing species with a maximum recorded age of 34 years in Rhode Island (Cooper, 1967) Contribution 2505 of the Virginia Institute of Marine Science, Gloucester Point, VA 23062. Bearden, C. M. 1961. Common marine fishes of South Carolina. Bears Bluff Lab. Conlr, vol 34, 47 p. IDeposited at South Carolina Department of Natural Resources, Marine Resources Library, 217 Fort Johnson Road, PO. Box 12559, Charleston, SC 29422.1 White et al : Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 425 and 31 years in Virginia (White, 1996). Growth parameters offish between northern and southern regions (Hostetter and Munroe, 1993) are comparable, except that Virginia tautog have exhibited almost twice the growth increments in young-of-the-year and age 13+ fish. Likewise, growth relationships are similar for tautog from New York (Briggs, 1977) and Virginia (Hostetter and Munroe, 1993). Within preferred habitats, juvenile and adult tautog develop home sites (011a et al., 1979). Tagging studies in- dicate seasonal movements between inshore and offshore habitats, but minimal north-south movement (Cooper, 1966; Briggs, 1977; Lynch-; Bain and Lucy''). During win- ter, adult tautog located offshore of Virginia are active at temperatures above 6.1°C (Adams, 1993). Likewise, tautog at inshore locations within Chesapeake Bay (Arendt et al., 2001a, 2001b) remain active at water temperatures of 5°C or above. In northern parts of its range, adult tautog move inshore and spawn when water temperatures increase in the springtime (Chenoweth, 1963; Cooper, 1966; Stolgitis, 1970; 011a et al, 1974, 1979), although some portion of the population remains offshore year-round (011a and Samet, 1977; Hostetter and Munroe, 1993). Very little is known about the reproductive biology of tautog in Virginia. Tau- tog begin spawning when water temperatures reach about 11°C (Chenoweth, 1963; 011a et al., 1974, 1980; Eklund and Targett, 1990; Hostetter and Munroe, 1993); thus the spawning season begins later in the spring at higher lati- tudes. Spawning season extends from mid-April through June in Virginia (Hostetter and Munroe, 1993), mid-May through early August in Massachusetts (Stolgitis, 1970), and from late May to early June in Rhode Island (Che- noweth, 1963). Macroscopic gonad analyses and gonadoso- matic indices have indicated that male tautog mature by age 3 and females by age 3-4 throughout the species range (Chenoweth, 1963; Cooper, 1967; Stolgitis, 1970; Briggs, 1977; Hostetter and Munroe, 1993); however, sample sizes of young (age 2-3) fish were small in these studies, and earlier maturation has been noted (011a and Samet, 1977; Hostetter and Munroe, 1993). To date there has been no histological examination of the reproductive biology of this species in Virginia, or elsewhere. In laboratory aquaria, tautog have been observed to be a multiple spawning species, spawning as discrete pairs and as groups (011a and Samet, 1977; OUa et al., 1977). Although hermaphroditism is common among labrids (Warner and Robertson, 1978), tautog are thought to be strictly gonochoristic (Olla and Samet, 1977). However, two color patterns of males exist in samples from Virginia waters; approximately 85% of males exhibit strong dimor- ^ Lynch, T. R. 1996. Marine finfish investigation, sport fish population survey in Rhode Island marine waters: tautog stud- ies, 1987-1993. " Rhode Island Division of Fish and Wildlife performance report project: F-54-R-1, study I-I. reference docu- ment TT-396, 55 p. Rhode Island Division of Fish and Wildlife, Marine Fisheries Section, 3 Fort Wetherill Road, Jamestown, RI 02835. ^ Bain.C. M.and J. A. Lucy. 1996. Virginia Game Fish Tagging Program annual report, 1995. Virginia Mar Res. Rep. 96-2, 10 p. Virginia Marine Resources Commission, 2600 Washington Ave, Newport News, VA 23067. phism and 15% of smaller males (<550 mm) show external coloration similar to females (Hostetter and Monroe, 1993, this study). Further, histological analysis of 379 male tau- tog revealed no evidence of hermaphroditism (Munroe and White, unpubl. data) The reproductive potential of tautog, measured as annu- al fecundity, has not been addressed to date. To accurately estimate potential annual fecundity for multiple spawning species, batch fecundity must be multiplied by the number of spawnings per year, i.e. spawning frequency multiplied by spawning season length (Hunter and Macewicz, 1985). Chenoweth (1963) and Stolgitis (1970) estimated batch fecundity and length of spawning season but did not mea- sure spawning frequency. The only estimate of number of spawnings per female per year were those of Olla et al. (1977), who observed tautog spawning daily for 68-96 consecutive days in laboratory aquaria; however, results obtained in aquaria studies may not directly apply to natu- ral habitats and have not been used to estimate potential annual fecundity. Our study, the first to investigate cellular aspects of the reproductive biology of tautog in natural habitats of the lower Chesapeake Bay, was necessary 1) to describe this species as a determinate or indeterminate spawner; 2) to describe annual and spawning season ovarian cycles at the cellular level; 3) to assess age at maturity based on histological sections of gonad tissue; 4) to estimate batch fecundity for females from the southern portion of the spe- cies range; 5) to estimate spawning frequency; and 6) to estimate potential annual fecundity. Materials and methods A total of 960 tautog (>150 mm total length [TL]) were collected opportunistically between April 1994 and Sep- tember 1995 from commercial and recreational fishermen (ca. 909c ) as well as from research projects (ca. 10% ) at Vir- ginia Institute of Marine Science (VIMS). A minimum size of 150 mm TL was selected based on maturity information presented in Hostetter and Munroe ( 1993). Collection loca- tions ranged from Gwynn's Island at the mouth of the Rap- pahannock River to 45 km offshore of Virginia's coastline, at depths of 1-35 m (Fig. 1). Approximately one quarter of the fish were taken from within the Chesapeake Bay, one quarter from the Chesapeake Bay Bridge Tunnel (CBBT, depth=5-15 m), and one half from around the Chesapeake Bay Light Tower (24 km offshore, depth= 17-20 m). For each fish, total length (mm) and total weight (TW. g) were measured. Sex was assessed externally from sev- eral dimorphic characters previously described by Cooper (1967), Olla and Samet (1977), and Hostetter and Munroe (1993). Males were distinguished by their pronounced white chin, blunt forehead, solid black or gray coloration on the upper half of the body and white underneath, and a small white spot (about 15 mm diameter) mid-laterally, below the middle of the dorsal fin. Female tautog have a less pronounced chin, sloped forehead, and a mottled brown coloration. After determination of sex (external examina- tion), gonads were excised, staged macroscopically, and 426 Fishery Bulletin 101(2) Figure 1 Map of lower Chesapeake Bay and nearby coastal waters of Virginia. Closed circles indicate sites where tautog were collected; open circles indicate collection sites of tautog in spawning condition. CBBT = Chesapeake Bay Bridge Tunnel. CELT = Chesapeake Bay Light Tower. weighed to the nearest 0.01 g (GW). Maturity classification was assigned as outlined in Table 1, based on eight macro- scopic stages, modified from Lowerre-Barbieri et al. ( 1996) for multiple spawning species. One gonad was randomly chosen by coin toss for histological processing and placed in Davidson's fixative. For females staged macroscopically as spawning females, the remaining ovary was placed in 10% neutrally buffered formalin for batch fecundity counts. Whole unsectioned opercle bones are the accepted method to age tautog (Cooper, 1967; Simpson, 1989;Hostet- ter and Munroe, 1993). Opercle bones were removed and processed to examine age at maturity and age-related fecundity. Opercles were boiled for 1-3 minutes to remove flesh, scrubbed under warm flowing water, dried for two days, and read with transmitted light. Age of each fish was determined from two readings of both opercles (when possible). An annulus was defined as the transition from a translucent zone to an opaque zone. Annulus formation was previously validated by Hostetter and Munroe ( 1993). 1 April was used as a birth date to allow maximum growth within the biological year (April to March), and to avoid overlap with fish in the next year class. Gonads selected for histological processing were placed in Davidson's fixative for two days before transverse sec- tions of anterior, middle, and posterior ovarian tissue (or anterior and posterior sections of testes) were taken and placed in tissue cassettes. Variation between left and right gonads was accounted for by random selection of one go- nad for fixation. Tissue samples were then rinsed overnight with flowing tap water and placed in TO'^r EtOH. Standard histological processing (tissue embedded in paraffin, sec- tioned at 5-7 pm, and stained with Harris's hematoxylin and eosin-Y) (Luna, 1968) was performed for all samples. Male gonads were classified microscopically into two stages: sexually mature or immature. Female microscopic gonad stages were assigned based on the occurrence and relative abundance of seven oocyte developmental stages (Wallace and Selman, 1981; West, 1990; Hunter et al., 1992): primary growth, cortical alveoli, partially yolked, advanced yolked, germinal vesicle migration, germinal ves- icle breakdown, and hydrated oocytes. Final oocyte matura- tion (FOM) comprises germinal vesicle migration, germinal vesicle breakdown, and hydrated oocyte stages (Wallace and Selman, 1981; West, 1990). Fully developed ovaries were distinguished from partially spent/redeveloping ovaries by the presence of postovulatory follicles (POFs). Microscopic gonad stages are described in "Description of microscopic gonad stages," ("Results" section), summarized While et al : Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis All Table 1 Description of macroscopic and microscopic gonad stages (modified from Lowerre-Barbieri et al.,1996) for female tautog. Mac- roscopic criteria refer to whole fresh ovaries. Gonad stages 4, 5, and 3a comprise the inner spawning cycle. FOM = final oocyte maturation. GSI = gonadosomatic index. GVBD = germinal vesicle breakdown. POF = postovulatory follicle. MA = macrophage aggregate. Gonad stage Macroscopic criteria Microscopic criteria 1 Immature 2 Developing 3 Fully developed 4 Hydrated 5 Running ripe 3a Partially spent/ redeveloping Spent/ regressing 7 Resting Ovaries very small, tubular in shape, white to light pink in color; no oocytes visible (mean GSI=0.50) Ovaries small to medium, tubular shape, dark yel- low to light orange in color; yolked (opaque) oocytes begin to appear (mean GSI=2.25 ) Ovaries medium to large, appear slightly grainy, pale mustard in color; yolked oocytes are abundant (mean GSI=3.25 ) Ovaries large to very large, pink to orange in color; firm, yolked oocytes interspersed with large trans- parent (hydrated) oocytes (mean GSI=11.74 ) Ovaries large to very large; few transparent oocytes in ovarian tissue, transparent oocytes have been ovulated into expanded lumen, and are easily ex- truded when gonad is excised (mean GSI=10.12) Ovaries somewhat flaccid, large, slightly more pink than in hydrated stage; lumen has collapsed, occa- sionally a few remnant hydrated oocytes extruded from excised ovary (mean GSI=8.84); similar to stage 3. Ovaries flaccid, small to medium, red to purple; some tissue devoid of yolked oocytes at anterior end of ovary, yolked oocytes visible but less abun- dant (mean GSI=1. 37) Ovaries small, purple-opaque to maroon in color; few or no yolked (opaque) oocytes visible (mean GSI=1.50) Oogonia and primary growth oocytes present; high pro- portion of connective tissue, no atresia or MAs, ovarian membrane thinner than in resting stage. Primary growth, cortical alveoli, and some partially yolked oocytes present. Primary growth to advanced yolked oocytes present; no FOM stages, POFs, or remnant HOs. Primary growth to germinal vesicle migration (GVMl and hydrated oocytes present, hydrated oocytes are unovulated; 1-day POFs may be present. Primary growth through GVM, and ovulated hydrated oocytes and fresh POFs present; lumen usually seen as separation of ovigerous folds. Primary growth through GVBD oocytes present, no unovulated hydrated oocytes, few remnant ovulated hydrated oocytes; lumen collapsed, POFs abundant. Primary growth through advanced yolked oocytes pres- ent; major atresia of all stages except primary growth oocytes. Primary growth and cortical alveoli oocytes present, occasional atretic oocytes; MAs abundant, more oogonia tissue, less connective tissue, and thicker ovarian mem- brane than in immature stage. in Table 1, and shown in Figure 2, A-H. Percent agreement between macroscopic and microscopic female gonad stages was calculated to evaluate the accuracy of macroscopic staging (used in all previous studies of tautog reproduc- tive biology). Microscopic stages were assumed to be more accurate because histologic examination provides evidence of differences in cellular development. Chi-square analysis (n=489 fish) was used to test for significant deviations from an expected 1:1 sex ratio for all fish. Deviations from a 1:1 sex ratio among 100-mm length intervals were also analyzed by chi-square to determine if size or age had a significant effect on sex ratio. Length and age at maturity were analyzed for fish col- lected from April to rnid-June to reduce the possibility of classifying resting, mature fish as immature. Females were considered mature if classified into microscopic stages 2-7 (Table 1). Males were considered mature if spermatoc3ftes or spermatozoa were present in histological sections. Length at maturity was based on 110 females and 79 males (150-350 mm TL). A logistic regression curve was fitted to the data, to estimate length at 50% maturity (Lgg). Age at maturity was based on 135 females and 104 males (ages 1-6). To determine the annual spawning season, a gonadoso- matic index (GSl=(gonad weight I somatic weight) x 100) was calculated by using somatic weight [S'W=TW-GW) for each sex. A more precise estimate of tautog spawning season was determined from microscopic gonad stages. The spawning season was defined by the first and last day that female tautog. were collected with ovaries staged as either hydrated, running ripe, or partially spent/redeveloping. Spawning locations were detected by the presence of hy- 428 Fishery Bulletin 101(2) Figure 2, A-D. Histologic stages of tautog ovaries. (A) immature (stage 1). (B) developing (stage 2). (C) fully developed (stage 3). (Dl hydrated (stage 4). PG = primary growth oocyte. CA = cortical alveoli oocyte. PY = partially yolked oocyte. AY = advanced yolked oocyte. HO = hydrated oocytes. Magnification = 50x. drated and running ripe fish because those stages have high abundances of hydrated oocytes and fresh POFs, which indicate cither imminent or recent spawning activ- ity (Hunter and Macewicz, 1985). Oocyte development patterns (synchronous, group syn- chronous, or asynchronous) and type of annual fecundity (indeterminate or determinate) were assessed by oocyte size-frequency distributions (Hunter and Macewicz, 1985) and histology (Hunter and Macewicz, 1985; West, 1990). Six fish were selected for analysis of oocyte size-frequency distributions; three fish (TL=300, 400, 450 mm [±10 mm)) in April and another throe in June, representing gonad de- velopment early and late in the spawning season. For each fish, oocytes were hydraulically separated from the ovar- ian membrane and each other, collected in a 0.1-mm sieve, and preserved in 2% formalin following the method of Low- erre-Barbieri and Barbieri ( 1993). Preserved samples were stirred to reduce bias due to differential settling of different White et al.: Reproductive seasonality, fecundity, and spawning frequency of Toutoga onills 429 i fift^ fe^/^L.'^'. Figure 2, E-H. Histological stages of tautog ovaries (continued). (E) running ripe (stage 5), frame 1 shows obvious lumen as distinguishing character, frame 2 shows typical field of view without lumen. Hydrated oocytes have been ovulated (OHO) and now reside in the lumen (L). (F) partially spent/redeveloping (stage 3a). (G) spent (stage 6). (H) resting (stage 7). PG = primary growth oocyte. CA = cortical alveoli oocyte. AY = advanced yolked oocyte. GVM = germinal vesicle migra- tion oocyte. GVBD = germinal vesicle breakdown oocyte. POF = postovulatory follicle. AO = atretic oocyte. MA = macrophage aggregate. Magnification = 50x. stage oocytes, and a 5-mL aliquot was removed and placed in a gridded petri dish. Grids were selected for counting by using a random number table, and maximum diam- eters of the first 500 oocytes encountered were measured to the nearest 0.001 mm with a Biosonic Optical Pattern Recognition System.® Data were grouped in 0.05-mm size classes for presentation (0.1 mm group=0.075 to 0.124 mm oocytes). Batch fecundity was determined gravimetrically by using a modification of the hydrated oocyte method (Hunt- er etal., 1985). The method calls for both ovaries to be fixed in IC/f formalin, but we had a formalin wet weight for only 430 Fishery Bulletin 101(2) one ovarian lobe. Therefore, we conducted a calibration experiment to determine the percent change in ovarian weight between fresh- and formalin-fixed ovaries. On 25 April 1996, 18 female tautog were collected, fresh ovarian weight was measured to the nearest 0.01 g, and both ova- ries were placed in 10% neutrally buffered formalin. For- malin-fixed wet weight was measured to the nearest 0.01 g six times over 30 days to determine when weight stabilized after fixation. Percent change in weight was calculated for each specimen, and regressed against fresh weight of the ovary, thus percent change in weight between formalin- fixed wet weight and fresh ovary weight was calculated with the negative exponential relationship: Percent weight change = 21.452 e'" oie^cwi where GW = fresh gonad weight. =0.671; Calibrated (formalin fixed) gonad weight (CGW) was cal- culated as CGW = percent weight change x GW. Then, batch fecundity was estimated by using the formula Y = iy/x)CGW, where Y= batch fecundity; y = number of hydrated oocjrtes in the tissue sample; X = formalin wet weight of tissue sample; and CGW = calibrated formalin-fixed wet weight of ovaries. Assumptions of the hydrated oocyte method which must be met include 1) all eggs in the most advanced mode are spawned; 2) fecundity is directly proportional to ovary weight; and 3) no bias exists in the estimation of egg abundance within the most advanced mode, in the selection of mature females for analysis, or in the position within and between ovaries from which subsamples were taken (Hunter and Goldberg, 1980; Hunter et al., 1985). The use of hydrated oocytes, which are much larger than the next largest cell size class and are formed only when spawning is imminent, supported acceptance of these assumptions. Following the methods of Hunter et al. (1985), we selected ovaries from 29 females for batch fecundity analysis. These were the only females that had stage-4 (hydrated. Table 1) ovaries without postovulatory follicles as confirmed through histological analysis. If postovulatory follicles were found in histological sections, then that fish was excluded from fecundity analysis. To test for differential oocyte development between ante- rior, middle, and posterior sections of ovarian tissue, point counting analyses ( Weibel et. al., 1966) were performed on histological sections to determine the relative volume of seven coll types and POFs in the ovary. The relative vol- ume of each cell type was calculated by using the number of points within a grid (121 point.s/grid) overlying each cell type: V=FJF where V^, = relative volume of one cell type; P„ = number ofpoints overlying a specific cell type; and P,„i= number of points in grid. To ensure that fields of view were chosen randomly, each ovarian section was divided into 5x5 mm areas with an overlay grid. Three areas per section were chosen with a random number table to ensure that counting fields of view did not overlap. Within each 5x5 mm area, point counts were made through a gridded reticule ( 121 points) at 4x magnification. Average relative volume of each cell class was calculated from the three areas as P,/363, and compared between anterior, middle, and posterior ovarian sections with multiple analysis of variance (MAN OVA) (Minitab, 1995). Response variables (8) were the average relative volume of each cell class. Differences between fish (10) were removed by blocking on fish. After no posi- tional effects were detected ( Wilk's test value 0.25, F=1.35, df= 16,22, P=0.25), it was concluded that oocyte develop- ment was evenly distributed throughout the ovaries of tautog. Blocking by fish proved beneficial and effective in removing any artifact caused by differences in kill time between fish, and in increasing the quality of the test by increasing sample size. Nonsignificant positional effects in ovarian development allowed estimation of batch fecundity from only the middle ovarian section. All hydrated oocytes were counted from three subsamples of approximately 0.3 g from the middle of the formalin-fixed ovary. Simple linear regressions were used to describe relation- ships between batch fecundity and TL, TW, and age. Rela- tive fecundity was calculated as batch fecundity divided by GW, and regressed against TL, TW. and age. Diel spawning periodicity estimates for tautog at the mouth of the Chesapeake Bay indicated that spawning occurs during daylight hours but that spawning windows shift with ebb tidal currents (White, unpubl. data). To esti- mate spawning frequency by the hydrated oocyte method (DeMartini and Fountain, 1981; Hunter and Macewicz, 1985), samples with known kill times must be collected just prior to, and during, the spawning window. Most samples were collected at dockside; thus kill time for individual fish was unknown, and the hydrated oocyte method could not be performed. Therefore, spawning frequency was estimated by the POF method (Hunter and Goldberg, 1980; Hunter and Macewicz, 1985) by using descriptions of fresh (day 0, 0-12 h) and degenerating (day 1, 12-24 h) POFs of tautog (White, unpubl. data). Fresh POFs in tautog ovaries can be identified as a clearly defined, loosely folded ribbons of the- cal and granulosa cells that contain visible luniina. similar to "0 day" POFs in anchovies ( Hunter and Macewicz, 1985 ). One-day-old tautog POFs have deteriorated such that in- dividual cell walls are no longer apparent in thecal and granulosa cells, and the structure appears less organized and has a small or indistinguishable lumen similar to that of 24-48 h anchovy POFs (Hunter and Macewicz, 1985). A full description of POF degeneration in tautog ovaries is presented elsewhere (White, unpubl. manscr). Annual fecundity was estimated as the number of spawnings per female multiplied by batch fecundity for White et aL: Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 431 Table 2 Percent agreement between microscopic and macroscopic ovarian stages assigned to tautog (n=484) captured from June 1994 through Septen iber 1995. See Table 1 for gonad stage descriptions. Data are expressed as number of ovaries staged. Microscopic Macroscop ic gonad stage stage 1 2 3 4 5 3a 6 7 1 34 — — — — 3 2 3 15 — — — — — 40 3 4 5 3 18 11 3 31 38 1 8 2 — 5 1 3a 1 — 4 55 3 11 1 — 6 1 — — — — 1 6 — 7 12 3 — — — — 38 132 Agreement 637f 42'y, 69% 24% 67% 79% 13% 75';; each fish. Number of spawnings per female was calculated by dividing the number of days in the spawning season by estimated annual spawning frequency. The relationship between mean annual fecundity per 50-mm length inter- val and total length was analyzed with both linear and exponential regression. Results Description of microscopic gonad stages Tautog ovarian development was described by eight micro- scopic gonad stages (Table 1) characteristic of multiple spawning species. Each stage can be differentiated by a unique suite of histological characteristics. Immature ovaries (Fig. 2A) are characterized by the presence of only oogonia and primary growth oocytes within a thin ovar- ian membrane and a relatively high volume of connective tissue. Developing stage ovaries (Fig. 2B ) are characterized by the presence of primary growth, cortical alveoli, and partially yolked oocytes. The fully developed ovary (Fig. 2C ) is characterized by the presence of primary growth to advanced yolked ooc3rtes and the absence of oocytes in final oocyte maturation (FOM) classes, POFs, or remnant HOs. Hydrated ovaries (Fig. 2D) are distinguished by the promi- nence of hydrated oocytes still inside the ovarian follicles and may also contain degenerating POFs from an earlier spawning, but they noticeably lack oocytes in the germinal vesicle breakdown state (GVBD). The running ripe stage (Fig. 2E) is classified by the presence of an expanded ovar- ian lumen, ovulated hydrated oocytes free in the lumen (although hydrated oocytes are frequently washed out of the sample during the staining procedure), a large number of fresh POFs, and germinal vessicle migration (GVM) oocytes that tend to be the most advanced stage present within ovigerous folds. Partially spent/redeveloping ovaries (Fig. 2F) are classified by the lack of an ovarian lumen and presence of occasional remnant hydrated oocytes, primary growth to GVBD oocytes, and an abundance of POFs. The spent stage (Fig. 2G) is characterized by resorption of yolked oocytes (atresia), and sometimes the presence of macrophage aggregates (MAs), which are groups of cells containing the pigments lipofuscin, ceroid, and melanin (Wolke, 1992). These cells appear to be a collection of scavenging cells that remove cellular debris and foreign substances by phagocytosis when stimulated by excessive degenerating tissue (Wolke, 1992). In tautog ovaries, MAs are assumed to be associated with the resorption of yolked oocytes after the spawning season. Resting stage ovaries (Fig. 2H) contain primary growth and cortical alveoli oocytes, and atresia may be present. Resting ovaries can be distinguished from immature stage ovaries by a thickened ovarian membrane, relatively more oogonia than connec- tive tissue and the presence of MAs. The reliability of macroscopic staging to predict actual reproductive stage as detected by microscopic analysis was examined for 484 females. We considered any level of agree- ment above 80% to be acceptable. The ability of macroscopic staging to predict actual microscopic stage varied consider- ably, from 13Vr to 79% agreement for individual stages (Table 2), indicating that macroscopic staging was generally unreli- able for estimating actual reproductive stage for many ovar- ian stages. Percent agreement between macroscopic and mi- croscopic staging was only 51% overall. The best agreement was for the fully developed, partially spent/redeveloping, and resting ovarian stages (agreement 69%, 79%, 75%, respectively). Intermediate values were obtained for im- mature, and running ripe stages, (agreement 63% and 67%, respectively). The poorest agreement occurred in assigning developing and spent stages (42%, 13%, respectively). Sex ratios Of the 938 tautog sexed, 522 (56%) were females and 416 were males. Overall, sex ratios varied significantly from an expected 1:1 ratio, with more females than males (1.25:1, ^2=11.98, P<0.01). At lengths <400 mm, females were 432 Fishery Bulletin 101(2) A Male tautog (n=79) (10 - B Female tautog (n=100) • •^*— MO ^ ^r fn - y= 100/(l+(38987e'""'""'"''"-') r = 0,88 « f 70 7 60 / 50 t ' 40 / 30 ■ / 20 - I / 10 ■ ^y H — ' — ^ ' 1 0 - ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' 1 50 150 200 250 Total length (mm) 350 Figure 3 Length at maturity for (A) male and (B) female tautog collected during the 1995 spawning season in lower Chesapeake Bay and nearby coastal waters of Virginia. Curves represent the logistic regression line. more abundant than males, whereas the sex ratios were not significantly different from 1:1 at fish lengths >400 mm (Table 3). Length and age at maturity Tautog length at 509f maturity (^r,o' was 218 mm for males (/!=79) and 224 mm for females (« = 110, Fig. 3). All males and females were mature at 300 mm. No females less than 227 mm had hydrated oocytes or POFs that would have indicated spawning activity. Age at first maturity was defined as the age at which at least 50% of the fish are mature. Mature gonads were present in Wc of males at age 1, 38% at age 2, and 93% at age 3. Zero percent of females were mature at age 2, 78% at age 3, and >97% at age 4. Spawning season and location GSI values indicated that tautog spawned from April through June and that peak values occurred in April for the 1995 spawning season (Fig. 4). The 1995 spawning season was more precisely defined as 7 April-15 June based on the presence of females in spawning condition (i.e. staged histologically as hydrated, running ripe). At the beginning of the spawning .season, females progressed into spawning condition over approximately two weeks. The end of the spawning season was determined conservatively, based Wh Ite et al : Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 433 lu o -Female {n=476) ■Male In =^79) Jun Month Figure 4 Gonadosomatic index based on somatic weight (total weight-gonad weight) for tautog col- lected in lower Chesapeake Bay during 1994-95. ?i=sample size; error bars indicate sample standard deviation. Table 3 Sex ratio of tautog by 100-mm length intervals at P=0.01. NS = nonsignificant. and chi-square va ues of tests for a 1:1 ratio. *Significance atP= =0.05. **Significance Total length (mm) No. of males No of femal es No expected (50%) % Females Chi-square 101-00 12 24 18 67 4.00* 201-300 57 88 73 61 6.63* 301-400 147 216 182 60 13.12** 401-00 140 125 133 47 0.85 NS 501-600 38 53 46 58 2.47 NS 601-700 11 14 13 56 0.36 NS Total 405 520 on the last day female tautog were collected in spawning condition (15 June) instead of the first collection date of a spent female (27 June). Tautog were collected (Fig. 1) in spawning condition within the Chesapeake Bay (York River, Buckroe Beach), at the mouth of the Chesapeake Bay (Chesapeake Bay Bridge Tunnel, Cape Henry wrecks, Anglo- African wreck), and at offshore locations (Chesapeake Bay Light Tower, one site 45 km offshore) and there was no apparent trend in spawning season by location. Ovarian developmental pattern and type of fecundity Ovarian stages defined for tautog (Table 1) are typical of multiple spawning species. Tautog hydrate and spawn only a small fraction of the yolked oocytes in the ovary for any one spawning event. Macroscopically, hydrated ovaries appear speckled because of the intermittent occur- rence of large, clear hydrated oocytes among the dominant numbers of opaque, yolked oocytes. Further, the lumen of running ripe ovaries was full of ovulated hydrated oocytes, yet there was still a large volume of tissue with maturing yolked oocytes. The occurrence of spawning stage ovaries over a protracted period also suggested a multiple spawn- ing pattern. Tautog were collected in spawning condition (hydrated and running ripe stages) and the partially spent/ redeveloping stage throughout the April-.June spawning period, but no spent or resting fish were collected until late June, suggesting that individual females were spawning repeatedly during the spawning season. 434 Fishery Bulletin 101(2) 1(K) 80 60 40 20 0 Total length = 300 mm D April 1995 ■ June 1995 -J4--t- Total length = 400 mm D April 1995 ■ June 1995 Oocyte diameter (mm) Figure 5 Oocyte size frequency distributions (0.05-mm intervals) in tautog ovaries. Six fish are represented, two in each length class: (A) 300 mm, (B) 400 mm, (C) 450 mm. Collection dates were 26 April 1995 and 2 June 1995. Fish were staged macroscopically as hydrated or running ripe, and oocytes were hydraulically separated and the diameter of 500 oocytes was measured to the nearest 0.001 mm. Oocyte size-frequency distributions measured at the begin- ning and end of the spawning season were also used to classi- fy tautog annual fecundity as determinate or indeterminate. Primary growth and cortical alveoli oocytes were continu- ously yolked, matured, and spawned throughout the spawn- ing season, evidenced by 1 ) lack of hiatus between advanced yolked oocytes and less mature oocytes, and 2) abundance of yolked oocytes (size range 0.30-0.55 mm) not decreasing over the spawning season (Fig. 5). This type of development defines tautog as having indeterminate fecundity. Two patterns of oocyte development are common among multiple spawning fishes: group synchronous and asyn- chronous oocyte development. Tautog oocyte size-frequency distributions (Fig. 5) show no distinct gaps in develop- White et al ; Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 435 ment, or modes of oocjrtes, except for hydrated oocytes (0.95 to 1.25 mm) — a finding that indicates asynchronous oocyte development. The presence of primary growth through advanced yolked oocytes, as well as oocytes in FOM and POFs in histologic sections of fully developed and partially spent ovaries (Fig. 2, C and F), is evidence that tautog exhibit a multiple spawn- ing pattern. Ovarian cycle ( 1 ) Immature 1) Sexual Maturity f7^ (7) Rest ing (2) Developing (6) Spent/ Regressing Egg maturation in tautog is a complex process comprising both seasonal and annual components. This complex pat- tern of multiple spawning exhibited by an inner spawning cycle (made up of hydration, ovulation, spawning, and redevelopment) within the annual ovarian developmental cycle, is sum- marized for tautog in Virginia in Figure 6. In the spring, fully developed ovaries contain primary growth to advanced yolked oocytes, but lack POFs. Fish enter the spawning cycle by hydration and ovulation of the first batch of oocytes. After the first spawning event, partially spent^redeveloping ova- ries contain fresh POFs (indicating recent spawning during the previous 24 hours) and another batch of oocytes in FOM in preparation for the next spawning event. Thereafter, the inner spawning cycle is repeated throughout the spawning season. Histological examination of hydrated ovaries during the spawning season revealed the co-occurrence of hydrated oocytes (indicating an imminent spawn) and degenerating POFs, suggesting that some tautog are capable of repeating the inner spawning cycle on a daily basis. At the end of the spawning season, ovaries progress to the spent-regressing stage, where, through the process of oocyte atresia, the remaining stock of yolked oocytes are resorbed before the ovary enters the resting stage. Batch fecundity Batch fecundity was determined for 29 female tautog rang- ing in total length from 260 to 520 mm, total weight 475 to 3,500 g, and ages 3-9 (Fig. 7). Although there was a high degree of variation in batch fecundity between individual fish, significant relationships were found between batch fecundity and fish length, weight, and age. Batch fecundity was more closely related to total length and total weight than to age. Batch fecundity (BF) increased significantly with total length (ANOVA, n=29, F=16.92, P<0.0005, power=0.97), following the regression equation (Fig. 7A) f'^:^ (3a) Partially spent/ Redeveloping (3) Fully Developed FOM Ovulation (5) Running Ripe (4) Hydrated ^g Figure 6 Annual and spawning season cycles of ovarian development in tautog. FOM = final oocyte maturation. BF = 56,066 Ln(TW) - 322,091 =0.501. Batch fecundity also increased significantly with age (ANOVA, n=29, F=10.22, P<0.004, power=0.88), following the regression equation ( Fig. 7C ) BF = 15,73 1(AG£) - 20,599 [r2=0.271. BF = 425.76(rL)- 84,534 |r2=0.39]. Batch fecundity increased significantly with total weight (ANOVA, n=29, F=16.80, P<0.0005, power=0.99), the regression equation being (Fig. 7B) Tautog relative fecundity (BF/GW) did not increase sig- nificantly with fish length (ANOVA, n=29, F=1.98, P=0.17) or age (ANOVA, n=29, F=1.72, P=0.20), but there was a significant increase in relative fecundity with total fish weight (ANOVA, n=29, P=4.46, P=0.044). Spawning frequency Histological examination of 169 tautog collected from 7 April 1995 to 15 June 1995 revealed some variation in the abundance of three reproductive states that were indica- tive of imminent or recent spawning. Forty-four percent of female tautog had HOs, 32'y^ had fresh POFs without any HOs, and 84'2 of females collected had 1-day-old POFs (Table 4). Tautog spawning frequency was calculated as 1.2 days based on the percentage of fish with 1-day-old POFs following the methods of Hunter and Goldberg (1980). Number of spawnings per female tautog in 1995 was calculated as the spawning season (70 days) divided by spawning frequency (1.2 days/spawning), yielding 58 spawnings per female. Individual tautog in natural habitats were capable of spawning daily after entering the spawning season. Evi- dence of daily spawning was provided by the rapid ovarian development observed in histological sections: 1) 70 fish with HOs and degenerating POFs: 2) 90 fish with both fresh and degenerating POFs: and 3) partially spent/ 436 Fishery Bulletin 101(2) ^uo.uou :^ L • 150.000 HKi.om 1 • • • • ^-<^ 50.000 • •• •• H — ' — • ' — . — 1 — . — • — . — . — (- • • BF = 425 76(TLl- 84.534 r' = 0.39 0 - _j — , — , — , — 1 — , — , — , — . — j — , — ^ — ' — ' — 1 250 300 350 400 450 Total length (mm) 550 ..'i)(l.illK.I - F B • 150.000 • • •« 100.000 • • • 50.(KI0 4 0^ • • 1 ' ' ' ■-1-^ • • • BF = = 56.066Ln(TW)- 322.091 r' = 0,50 () - 1 — ^ ' ' 1 J — , — , — . — i — . — , — , — , — 1 — ^ — , — ^_ ^-1-^ ' ' ' 1 0 50(.) ia)0 I5(K) :0(H) 2500 .3000 3500 400(1 Total weight (g) 00.(J00 ^ [C 1 • 50.000 - (W.OOO • • • • • 50.000 • — \ • • \ • • 1 • s 1- • BF = 15.73l(Agel- 20.599 r' = 0 27 0 J — 1 1 1 — 1 23456789 10 Age (years) Figure 7 Tautog batch fecundity (no. of eggs) versus (A) total length (mm), (B) total weight (g) and (C) age (years) for 29 tautog collected in lower Chesapeake Bay and nearby coastal waters of Virginia in 1995. (Complete data for indi- vidual fish are available in White, 1996). redeveloping females with both GVBD oocytes and fresh POFs (Fig. 2F). Potential annual fecundity Annual fecundity was calculated as 58 spawnings/female multiplied by batch fecundity. Annual fecundity varied from 160,000 eggs (259 mm, age-3 fish), to 10,510,000 eggs (511 mm, age-9 fish). Mean annual fecundity increased significantly (ANOVA, ;;=5, F=16.69, P=0.015) with fish size. A linear regression of mean annual fecundity on fish length (50 mm size classes, Fig. 8), was described by the following relationship: Mean AF = 23,480(TL) - (5x106) [r2 = 0.811. Discussion Macroscopic and microscopic gonad staging Macroscopic ovarian staging of multiple spawning fishes can be difficult because subtle differences at the cellular level may not be detectable macroscopically (Parrish et al., 1986). However, macroscopic analysis does provide a rapid estimate of maturity and results in a general descrip- tion of spawning seasons at a reduced cost compared to White et al.: Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 437 0,00E+00 300 350 400 450 Total lengtfi (mm) 500 550 Figure 8 Linear relationship between mean annual fecundity (MAF) and body length in 50- mm length classes for tautog. Sample sizes are noted above each data point. Error bars indicate standard error of the mean. Table 4 Reproductive state of female tautog collected during 1995 spawning season. HO = = hydrated oocyte. POF = = postovulatory follicle. No. with fresh No. with No. mature, Total no. of Date (1995) No. with HOs POFs, no HOs 1-day POFs not spawning mature females 7 April 11 0 11 1 13 8 April 11 4 12 13 29 15 April 1 0 1 0 1 22 April 9 2 10 4 14 26 April 12 23 47 1 49 9 May 15 16 33 2 35 24 May 0 0 1 0 1 31 May 5 6 11 0 11 1 June 10 33 14 0 14 2 June 1 0 1 0 1 15 June 0 0 1 0 1 Total 75 54 142 21 169 Total % 44.4 32.0 84.0 — — time-consuming histological methods. West (1990) noted that there have been few attempts to assess the accuracy of macroscopic gonad staging with histological analysis. Most scientists attempting to assess reproductive stage of female tautog will most likely use macroscopic criteria. Given that all eight microscopic stages cannot be identi- fied in a macroscopic context, a revised macroscopic gonad staging was developed (Table 5) with validated agreement against microscopic analysis. The validated stages (Table 5 ) generally agree with previous studies of tautog reproduc- tive biology (Chenoweth, 1963; Stolgitis, 1970; Hostetter and Munroe, 1993). However, even with this revised, sim- plified staging criteria, we caution others that agreement between the new criteria and microscopic staging is still low for some ovarian stages, rendering this method less reliable than microscopic analysis. Despite these limitations, macroscopic staging errors were usually offby only one developmental stage (Table 2 ). Errors in macroscopic staging were most likely due to the rapid de- velopment of ovarian tissue required to sustain daily spawn- ing events. The low percent agreement (SI** overall I between macroscopic and microscopic classifications of tautog ovar- 438 Fishery Bulletin 101(2) Table 5 Revised macroscopic gonad stages for future research on tautog. These revised macroscopic stages are based on the gonad stages seen in Table 1. Gonad stage (Table 1) Revised gonad stage Macroscopic appearance Immature Immature Ovaries very small, tubular in shape, white to light pink in color, no oocytes visible. (Same as "immature stage" from Table 1.) Developing Developing Ovaries medium to large with slightly grainy appearance, pale mustard and in color, yolked (opaque) oocytes present, no hydrated (transparent) Fully developed oocytes visible through ovarian membrane. (This stage is a combination of "developing" and "fully developed" stages from Table 1.) Hydrated Spawning Ovaries large to very large, pink to orange in color, may be dotted with and transparent oocytes, yolked oocytes interspersed with large transparent Partially spent/ (hydrated) oocytes, occasionally a few remnant hydrated oocytes. (This redeveloping stage is a combination of "hydrated" and "partially spent/redeveloping" stages from Table 1.) Running ripe Running ripe Ovaries large to very large, pink to orange in color, hydrated oocytes have been ovulated, expand lumen of ovary and are easily extruded from excised ovary; few hydrated oocytes in ovarian tissue. (Same as "running ripe" stage from Table 1 . ) Spent and Resting Spent Ovaries flaccid, small to medium, red to purple in color, few yolked (opaque) oocytes visible, some or all of ovary having no oocytes visible. (This stage is a combination of "spent" and "resting" stages from Table 1.) ian stages is similar to that of studies on Lutjanus vittus, which found the accuracy of macroscopic staging for ripe gonads to be only 61% (CSIRO data, cited in West, 1990). Macroscopic staging of tautog ovaries functioned to de- scribe the annual gonad cycle, yet it did not separate the fully developed (stage-3) and partially spent (stage-3a) ovaries. Macroscopic analysis does not yield proof of POFs, atretic oocytes, and macrophage aggregates — cellular structures that help distinguish fully developed, partially spent/redeveloping, spent, and resting females. Thus, mac- roscopic analysis could not provide evidence of multiple spawning in tautog. Histological techniques used in this study were necessary to accurately describe the annual cycle and the inner (multiple spawning) spawning cycle of ovarian development for tautog. Histological staging also permitted identification of fully hydrated ovaries that could then be used for batch fecundity estimation. Further, histology slides were used for point-counting analyses to test for positional differences in development between an- terior, middle, and posterior regions within the ovary. Sex ratios Sex ratios vary greatly among published studies on tautog life history. This variability may be due to true differences in the composition of local populations, or it may be an arti- fact of sampling strategies rooted in collection seasons or gear biases. In our study, sex ratios were skewed towards females for fish under 400 mm and did not differ signifi- cantly from a l.l ratio for tautog greater than 400 mm. Collections were primarily made by book-and-line angling throughout the year, although sample sizes were low between July and September. Hostetter and Munroe ( 1993) found no significant difference in sex ratios for fish less than 200 mm, but significantly more males than females for fish between 201-500 mm in Virginia. Their sampling occurred over a period of seven years, and fish were collected primar- ily with fish traps and hook and line. Eklund and Targett (1990) found a female-to-male sex ratio of 0.86:1 in the trap fishery between April and December 1987. Chenoweth (1963) collected more females than males with an otter trawl at three stations in Narragansett Bay, RI, between May and September 1961. Factors that affect sex ratios of tautog from fishery-dependent collections are still unknown and provide an opportunity for further research into the sex ratios and reproductive success of this species. Length and age at maturity Published reports of tautog length and age at maturity (from studies with macroscopic techniques and GSI) are similar for the entire species range. Estimates of tautog lengths and ages at maturity in our study were similar to results reported by Hostetter and Munroe (1993) for tautog captured off Virginia. Hostetter and Munroe ( 1993) reported that both sexes show evidence of gonadal matu- ration at age 3 in Virginia. Likewise, age and length at maturity for tautog collected in Massachusetts (Stolgitis, 1970) are also similar; 40% of age-2 (149-175 mm) males and 87% of age-3 (171-239 mm) males were mature, and females attained 71% maturity at age 3 (187-206 mm) and 100% maturity at age 4. In Rhode Island, Cooper (1966) White el al.: Reproductive seasonality, fecundity, and spawning frequency of Tautoga onttis 439 found that males matured at 200 mm (age 3) and females at 190 mm (age 3). For a small number of fish sampled in northern areas, Hostetter and Munroe (1993) suggested that precocious development may be occurring in tautog as a response to fishing pressure. The smallest females collected in spawning condition have been 227 mm in Virginia (this study), 261 mm in Massachusetts (Stolgitis, 1970), 216 mm (Chenoweth, 1963) and 180 mm (Hostetter and Munroe, 1993), in Rhode Island. A definitive answer on precocious development is not possible at this time because data on small fish are limited in all studies. Detailed histological analysis should be per- formed on tautog from 100 to 250 mm TL to discern maturity schedules for specimens in this size range. Spawning season and location Tautog spawn over at least a two-month period throughout the species range, and the initiation of spawning activity occurs later in the spring to early summer in more north- ern regions (Chenoweth, 1963; Stolgitis, 1970; Briggs, 1977; Hostetter and Munroe, 1993). In our study, spawn- ing occurred from 7 April through 15 June 1995 (70 days), similar to the time interval reported by Hostetter and Munroe ( 1993) for tautog in Virginia. In New York waters, tautog have been recorded to spawn for four months (early May through early September; Austin, 1973). In Rhode Island, tautog spawn from early June through late July (Chenoweth, 1963), and spawning seasons as long as three months (mid-May through early August) have been reported for fish in Massachusetts (Stolgitis, 1970). Abun- dance of tautog eggs in plankton collections also shows that the spawning season occurs progressively later in more northern regions (Sogard et al., 1992). The earlier spawn- ing season in Virginia has been attributed to differences in water temperature (Hostetter and Munroe, 1993). Increas- ing water temperature during springtime is a major cue to initiate spawning, but termination of spawning activity has not been related to environmental cues. However, Austin (1973) suggested that the effective spawning season may be shorter than the season of egg release for this species, based on a decrease in larval abundance as water tempera- ture exceeded 21.0°C in Long Island Sound. Tautog were collected in spawning condition within the Chesapeake Bay and as far as 56 km offshore in this study and by Hostetter and Munroe (1993). Eklund and Targett (1990) sampled tautog in spawning condition 22-37 km off the coast of Maryland and Virginia. Field observations of daily movements showed that tautog exhibit fidelity to a home site which they return to each night (011a et al., 1974, 1975), suggesting that tautog remain at one location throughout the spawning season. Arendt et al. (2001b) found that tautog tended to move between sites during the winter and early spring as the spawning sea- son began and remained at a single site throughout the summer. Tagging studies indicate that discrete spawning groups exist at sites in Narragansett Bay (Cooper, 1966); however, movements between sites were not quantified. Sufficient data are not available to determine if tautog ex- hibit spawning-site fidelity throughout the spawning sea- son, or if multiple spawning sites are used within general inshore and offshore classifications. It is generally believed that most tautog migrate inshore in the spring to spawn (Cooper, 1966) and some portion of the population remains offshore year round (Eklund and Targett, 1990; Hostetter and Munroe, 1993). Although we have documented adult spawning activity at both inshore and offshore locations, spawning success in these areas, as well as larval drift and recruitment patterns, are unknown at this time. Spawning pattern and type of fecundity Histological analysis of ovarian tissue supports the clas- sification of tautog as a multiple spawning species with a complex reproductive cycle. The complexity of ovarian maturation (Fig. 6) in this species has not been recognized in previous studies on its reproductive biology. The typical cycle of female development for multiple spawning species is defined by eight microscopic gonad stages (Lowerre-Bar- bieri et al., 1996) which include an annual cycle (5 stages) and an inner spawning cycle (3 stages). Although tautog have been observed to be multiple spawners in laboratory aquaria (Olla and Samet, 1977; Olla et al, 1977), we define oocyte development and type of fecundity using recently improved methods (Lowerre-Barbieri and Barbieri, 1993) and histological techniques on fish taken from natural envi- ronments; therefore, they are directly comparable to other studies of reproductive biology without artifacts associated with aquarium conditions. Analysis of oocyte size-frequency distributions and histological sections of ovarian tissue indicates that tautog have asynchronous oocyte develop- ment and indeterminate annual fecundity. Therefore, counting the number of oocytes in the ovary prior to the spawning season is inadequate to measure potential annual fecundity because new batches of eggs continuously mature from primary growth oocytes through hydrated oocytes and are released during spawning events (Hunter et al., 1985). Chenoweth ( 1963) analyzed oocyte size-frequency distribu- tions of three tautog collected over the course of the spawn- ing season in Rhode Island and noted that the number of mature yolked oocytes did not decline through the spawn- ing season. He suggested that not all yolked oocvtes were spawned and that some portion remained in the ovary and were resorbed after the spawning season. This observation is consistent with asjoichronous oocyte development. Fecundity This is the first study on tautog reproduction for which potential annual fecundity has been estimated by multiply- ing batch fecundity by spawning frequency. Batch fecundity was more closely related to total length and total weight than to age. This result makes sense when one considers the extreme variability in length at age exhibited by tautog (Cooper, 1967; Hostetter and Munroe, 1993). Batch fecun- dity ranged from 2800 eggs to 181,200 eggs in 29 females age 3-9 (Fig. 7). The oldest tautogs collected in this study were a 31-year-old male and a 17-year-old female. After reaching maturity, individual females may spawn up to 58 times a year for at least 14 years. 440 Fishery Bulletin 101(2) Table 6 Comparison of batch fecundity estimates for tautog from three different studies. Mean batch fecundity was calculated for two age groups, 4-6 and 7-9, because all females were mature by age 4 and over 90% of all tautog sampled in White (1996) were less than 10 years old. Mean batch fecundity ±SEM Study Age 4-6 n Age 7-9 n Chenoweth (1963)' Rhode Island Stolgitis ( 1970)' Massachusetts White (1996)' Virginia 49,967 ±1032 29 46,833 ±4500 6 54,243 ±2472 18 103,214 ±4005 14 117,478 ±2488 23 106,256 ±3837 10 ' Mean batch fecundity for age groups 4—6 and 7-9 was calculated from the raw data presented in the reference. Previous estimates of tautog fecundity by Chenoweth (1963) and Stolgitis (1970) were not annual fecundity es- timates (Table 6). They counted mature, transparent eggs in the ovary, currently referred to as hydrated oocytes, but they did not distinguish tautog as having indetermi- nate annual fecundity and had no measure of spawning frequency. By counting only the hydrated oocytes, these investigators actually estimated batch fecundity. However, it is interesting to note the similarity of batch fecundity estimates (Table 6) over the period of 30 years between studies and wide geographic areas, i.e. from Chesapeake Bay to Narragansett Bay (550 km). Spawning frequency had not been previously calculated for tautog with methods developed by Hunter and Macewicz ( 1985). Although the hydrated oocyte method is less expen- sive, it requires collection of females just prior to spawning. With the hook-and-line collection method, it is difficult to collect sufficient samples in a short period of time. Therefore spawning frequency was estimated in our study by using the POF method to read histologic preparations of ovarian tissue. A female spawning every 1.2 days over the 70-day spawning window would spawn on an estimated 58 days in 1995. Under artificial conditions, Olla et al. ( 1977) observed tautog spawning on 68-96 consecutive days in laboratory aquaria. Therefore, an estimate of 58 spawning days in nat- ural habitats is not unrealistic. Chenoweth (1963) raised, but could not answer, the question of whether individual tautog spawn throughout the entire spawning season. The spawning-frequency estimate presented here, and observa- tions of tautog spawning on 68-96 consecutive days in labo- ratory aquaria (Olla et al, 1977), indicate that tautog are capable of spawning daily throughout the spawning season in natural habitats under appropriate environmental condi- tions (temperature, day length, etc.). Estimates of potential annual fecundity for Virginia tautog age 3-9 ranged from 160,000 to 10,510,000 eggs. However, net annual fecundity may be lower because of remnant hydrated oocytes, atresia, nutritional status of adult females, or environmental conditions (McEvoy and McEvoy, 1992). Based on our samples (females age 3-9), a linear regression provided the most predictive power (r2=0.81) to estimate mean annual fecundity (Fig. 8). Although female tautog live to be \7+ years old, it is estimated that 90% of tautog in Virginia waters are age 10 or younger (Hostetter and Munroe, 1993). Therefore, as the data range in this study is similar to the age structure of the resource, we suggest that the regression equation (Mean AF=23,480(TL) - (SxlQi^)) is the most appropriate formula for use by fishery managers for estimating annual fecundity of tautog in the southern portion of its range. We have made a theoretical comparison of potential an- nual fecundity for tautog (ages 4-9) between the northern and southern areas by combining results from several stud- ies in the northern range of tautog. Although commonly cited as representing annual fecundity estimates, the methods of Chenoweth (1963) and Stolgitis (1970) clearly show that their results are batch fecundity estimates. For our comparison, we selected the lowest value for age-4 and the highest for age-9 tautog as the sampled range of batch fecundity estimates. We averaged northern data from Chenoweth (1963: age 4, 265 mm TL, 6000 BF and age 9, 401 mm TL, 224,000 BF) and Stolgitis (1970: age 4, 261 mm TL, 7000 BF and age 9, 486 mm TL, 260,000 BF) to create a batch fecundity range of 6500-242,000 eggs. This range was multiplied by the 68-day "spawning season" observed in laboratory aquaria by Olla et al. ( 1977) to cal- culate a range for potential annual fecundity of 442,000 to 16,456,000 eggs per female in northern areas. Our samples from the southern range (age 4: 275 mm TL, 5000 BF and age 9: 511 mm TL, 181,200 BF) multiplied by 58 spawn- ing events in 1995 results in potential annual fecundity of 290,000 to 10,510,000 eggs per female. Differences in these estimates of potential annual fecundity are primarily due to the number of spawnings per year and are questionable because the spawning frequency estimate based on aquari- um studies may not apply for naturally spawning fish. This comparison indicates that we still lack adequate informa- tion on the spawning frequency and annual fecundity for fish from the northern part of the species range. Although batch fecundity estimates appear similar between southern and Northern portions of the tautog's range, previous batch fecundity estimates from northern populations are 30 years old. Reported spawning seasons between areas also vary in length from two to four months, which could greatly affect potential annual fecundity esti- mates with this method. Estimates of annual fecundity in White et al : Reproductive seasonality, fecundity, and spawning frequency of Tautoga onitis 441 the northern regions of the species range should be pur- sued to determine if tautog annual fecundity varies with latitude. Evidence of different growth rates (Cooper, 1965, 1966, 1967; Stolgitis, 1970; Hostetter and Munroe,1993; White, 1996), seasonality of occurrence in coastal waters, and winter activity cycles between tautog in southern versus northern regions (011a et al., 1974; Hostetter and Munroe, 1993; Arendt et al., 2001a, 2001b) strongly point to considering latitudinal effects when analyzing and com- paring any biological features of this species. Even if batch fecundity and spawning frequency remain relatively con- stant over latitude, size structure of the stock will dictate estimates of total egg production: thus continued research is necessary to monitor size structure and abundance of tautog resources throughout the species range. Additional data on larger, older females is necessary to evaluate the relative contribution of older females to population fecun- dity and egg production. Because many aspects of tautog life history affect recruitment, further investigation is re- quired on egg dispersal, egg mortality, larval drift, larval mortality, hatching success, first feeding success, pre- and postsettlement mortality, juvenile mortality, recruitment, stock structure, and spawning stock biomass (ASMFC"*). Historically, tautog have supported a predominantly (90%) recreational fishery throughout their range (ASMFC^). Over the past 15 years, this popular food and sport fish has increased substantially in value as a com- mercially targeted species. As popularity and fishing effort increased, landings peaked in 1993 but have declined more recently, prompting the Atlantic States Marine Fisheries Commission (ASMFC"*) to pass a coastwide management plan for tautog in April 1996. Tautog annual fecundity is a key piece of data neces- sary for egg production models and estimates of spawning stock biomass. and there are no reliable estimates of tautog spawning stock biomass to date (ASMFC''). In April 1998, the ASMFC imposed a 14-inch (350-mm) minimum size limit, effective for tautog caught from Massachusetts to Virginia. The benefits of instituting a size limit for tautog are well supported by data from this study. A minimum size limit allows tautog in the southern regions of the species' distribution to have at least one spawning season, and most likely two, thereby affording the opportunity for each female to contribute on average 3.22 million eggs (calcu- lated from the linear regression equation. Fig. 8) towards the annual population fecundity. Acknowledgments The authors gratefully acknowledge suggestions and criti- cisms of three anonymous reviewers that greatly improved the manuscript. We thank all of those who made this research possible. This study represents part of a M.S. thesis (by G. G. Wliite) School of Marine Science, Virginia " ASMFC (Atlantic States Marine Fisheries Commission). 1996. Fishery management plan for tautog. Fish. Manage. Rep. 25, 56 p. Atlantic States Marine Fisheries Commission, 1444 Eye Street, NW, 6"> Floor, Washington, DC 20005. Institute of Marine Science (VIMS), College of William and Mary. VIMS volunteers who assisted with our sampling efforts included D. Estes, R. Holmquist, D. Seaver, and M. Wagner, members of the Juvenile Finfish Trawl Survey, and T. Holden, who collected specimens. We appreciate the interest and cooperation of recreational and commercial fishermen, especially "Old Joe," Clark and Chester Stultz, who provided fish, For help processing samples, we thank J. Brust, W. Coles, C. Cooksey, S. Gaichas, J. Harding, and M. Wagner (taug circles). We thank C. Bonzek and R. Harris for computing assistance, D. Evans, R. Diaz, and L. Gar- rison for statistical help, and J. Harding for critical edito- rial reviews of an earlier draft of the manuscript. We also thank R Blake for training in histology procedures, and W. Vogelbein for aid with interpretation of histology sections. Finally G. White thanks his parents, who had the foresight to get him "'hooked on" life in, on, or under water while very young. This project was funded by grant numbers RF-94-5 and RF-95-3 from the Virginia Marine Resources Commis- sion, Recreational Saltwater License Fees. Literature cited Adams, A. J. 1993. Dynamics of fish assemblages associated with an offshore artificial reef in the Southern Mid-Atlantic Bight. Unpubl. M.S. thesis, 97 p. College of William and Mary, Williamsburg, VA. Arendt, M. D., J. A. Lucy, and D. A. Evans. 2001a. Diel and seasonal activity patterns of adult tautog, Tautoga onitis, in lower Chesapeake Bay. inferred from ultrasonic telemetry. Environ. Biol. Fishes 62: .379-391. Arendt, M. D., J. A. Lucy, and T. A. Munroe. 2001b. Seasonal occurrence and site utilization patterns of adult tautog, Tautoga onitis, (Labridae), at manmade and natural structures in lower Chesapeake Bay. Fish. Bull. 99:519-527. Austin, H. M. 1973. Distribution and abundance of ichthyoplankton in the New York Bight during the fall in 1971. N.Y. Fish Game J. 23:58-72. Bigelow, H. B., and W. C. Schroeder. 1953. Fishes of the Gulf of Maine. U.S. Fish Wild!. Serv., Fish. Bull. .53:1-577. Bleakney, J. S. 1963. Notes on the distribution and reproduction of the fish Tautoga onitis in Nova Scotia. Can. Field-Nat. 77:64-65. Briggs, P. T 1977. Status of tautog populations at artificial reefs in New York waters and effect of fishing. N.Y. Fish Game J. 24: 154-167. Chenoweth, S. B. 1963. Spawning and fecundity of tautog, Tautoga onitis (L.). Unpubl. M.S. thesis, 60 p. Univ. Rhode Island, N. Kingston. RI. Cooper, R. A. 1965. Life history of the tautog, Tautoga onitis (Linnaeus), from Rhode Island. Unpubl. Ph.D. diss., 153 p. Univ. Rhode Island, N. Kingston. RI. 1966. Migration and population estimation of the tautog, Tautoga onitis (Linneaus), from Rhode Island. Trans. Am. Fish. Soc. 95:239-247. 442 Fishery Bulletin 101(2) 1967. Age and growth of the tautog, Tautoga onitis (Linnaeus), from Rhode Island. Trans. Am. Fish. Soc. 96:134-142. DeMartini, E. E., and R. K. Fountain. 1981. Ovarian cycling frequency and batch fecundity in the queenfish. Seriphus politus: attributes representative of serial spawning fishes. Fish. Bull. 79:547-60. Eklund, A. M., and T. E. Targett. 1990. Reproductive seasonality of fishes inhabiting hard bottom areas in the Middle Atlantic Bight. Copeia 1990: 1180-1184. Hildebrand, S. F., and W. C. Schroeder. 1928. Fishes of the Chesapeake Bay Bull. U.S. Bur Fish. 43 (part 1): 1-366. Hostetter, E. B., and T. A. Munroe. 1993. Age, growth, and reproduction of tautog Tautoga onitis (Labridae: Perciformes) from coastal waters of Virginia. Fish. Bull. 91:45-64. 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. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 79-94. U.S. Dep. Commer , NOAA Tech. Rep. NMFS 36. Hunter, J. R., N. C. H. Lo, and R. J. H. Leong. 1985. Batch fecundity in multiple spawning fishes. In An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 67-77. Dep. Commer, NOAA Tech. Rep. NMFS 36. Hunter J. R., B. J. Macewicz, N. C. H. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of female Dover sole Microstomus pacificus, with an evaluation of assump- tions and precision. Fish. Bull. 90:101-128. Lowerre-Barbieri, S. K., and L. R. Barbieri. 1993. A new method of oocyte separation and preservation for fish reproduction studies. Fish. Bull. 91: 165-170. Lowerre-Barbieri, S. K., M. E. Chittenden Jr, and L. R. Barbieri. 1996. The multiple spawning pattern of weakfish in the Chesapeake Bay and Middle Atlantic Bight. J. Fish Biol. 48:1139-1163. Luna, L. G. (ed.) 1968. Manual of histologic staining methods of the Armed Forces Institute of Pathology. American Registry of Path- ology, Z"^ ed., 258 p. McGraw-Hill Book Co., New York, NY. McEvoy, L. A., and J. McEvoy 1992. Multiple spawning in several commercial fish species and its consequences for fisheries management, cultivation and experimentation. J. Fish Biol. 41(suppl. B):125-136. Minitab. 1995. Minitab reference manual, release vlOXtra, 542 p. Minitab Inc., State College, PA. Musick, J. A. 1972. Fishes of Chesapeake Bay and the adjacent coastal plain. In A checklist of the biota of the lower Chesapeake BaylM.L.Wass.ed), p. 175-212. Virg. In.st. Mar Sci. Spec. Sci. Rep. 65. Olla, B. L., A. J. Bejda, and A D. Martin. 1974. Daily activity, movements, feeding, and seasonal occur- rence in the tautog, Tautoga onitis. Fish. Bull. 72:27-35. 1975. Activity, movements, and feeding behavior of the cunner, Tautogolabrus adspersus, and comparison of food habits with young tautog, Tautoga onitis, off Long Island, New York. Fish. Bull. 73:895-900. 1979. Seasonal dispersal and habitat selection of cunner, Tau- togolabrus adspersus, and young tautog, Tautoga onitis, in Fire Island Inlet, Long Island, New York. Fish. Bull. 77:255- 261. Olla, B. L., A. J. Bejda, and A. L. Studholme. 1977. Social behavior as related to environmental factors in the tautog, Tautoga onitis. In The behavior of marine organisms: social behavior and communication; navigation; and development of behavior, p. 47-99. Proc. Annu. North- east Reg. Meet. Animal Behav. Soc. Plenary Papers. Mar Sci. Res. Lab. Tech. Rep. 20, Memorial Univ, St. John's, Nfld. Olla, B. L., and C. Samet. 1977. Courtship and spawning behavior of the tautog, Tau- toga onitis (Pisces: Labridae), under laboratory conditions. Fish. Bull. 75:585-599. Olla, B. L., A. L. Studholme, A. J. Bejda, and C. Samet. 1980. Role of temperature in triggering migratory behav- ior of the adult tautog Tautoga onitis under laboratory conditions. Mar Biol. (Beri.) 59:23-30. Parrish, R. H., D. L. Mallicoate, and R. A. Klingbeil. 1986. Age dependent fecundity, number of spawnings per year, sex ratio, and maturation stages in northern anchovy, Engraulis mordax. Fish. Bull. 84:503-517. Richards, C. E., and M. Castagna. 1970. Marine fishes of Virginia's Eastern Shore (inlet, marsh, and seaside waters). Chesapeake Sci. 11:235-248. Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Can. Bull. Fish. Aquat. Sci. 219:1-731. Sedberry, G. R., and H. R. Beatty. 1989. A visual census of fishes on a jetty at Murrells Inlet, South Carolina. J. Elisha Mitchell Sci. Soc. 105:59-74. Simpson, D. G. 1989. Population dynamics of the tautog, Tautoga onitis, in Long Island Sound. Unpubl. M.S. thesis, 65 p. Southern Connecticut State Univ., New Haven, CT. Sogard, S. M., K. W. Able, and M. R Fahay 1992. Early life history of the tautog Tautoga onitis in the Mid-Atlantic Bight. Fish. Bull. 90:529-539. Stolgitis, J. A. 1970. Some aspects of the biology of the tautog, Tautoga onitis (Linnaeus), from the Weweantic River Estuary, Mas- sachusetts, 1966. Unpubl. M.S. thesis, 48 p. Univ. Mass., Amherst, MA. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21:325-343. Warner, R. R., and D. R. Robertson. 1978. Sexual patterns in the labroid fishes of the Western Caribbean, I: the wrasses (Labridae). Smithson. Contrib. Zool. 254:1-27. Weibel, E. R., G. S.Kistler, and W. F Scherie. 1966. Practical stereological methods for morphometric cytology. J. Cell Biol. 30:23-28. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar Freshwater Res. 41: 199-222. White, G. G. 1996. Reproductive biology of tautog. Tautoga onitis, in the lower Chesapeake Bay and coastal waters of Virginia. Unpubl. M.S. thesis, 100 p. College of William and Mary, Williamsburg, VA. Wolke, R. E. 1992. Piscine macrophage aggregates: a review. Ann. Rev, Fish Dis. 1992:91-108. 443 Investigation of congeneric hybridization in and stock structure of weakfish (Cynoscion regalis) inferred from analyses of nuclear and mitochondrial DNA loci* Jan F. Cordes John E. Graves School of Marine Science Virginia Institute of fv\anne Science College of William and Mary Gloucester Point, Virginia 23062 Present address (for J. F, Cordes): Department of Animal Science University of California Davis, California 95616 E-mail address (for J F Cordes): |fcordes@ucdavis edu The weakfish (Cynoscion regalis) is distributed along the east coast of the United States from Massachusetts to eastern Florida and is most abundant from New York to North Carolina (Big- elow and Schroeder, 1953). Histori- cally there has been some question as to the taxonomic relationship between weakfish and sand seatrout (Cynosc/on arenarius); some suggest they may be separate populations of a single species (Moshin, 1973; Weinstein and Yerger, 1976; Cowan, 1985; Ditty, 1989), and others treat them as separate species (Schlossman and Chittenden, 1981). Weakfish support substantial com- mercial and recreational fisheries throughout their range. Precipitous drops in total annual catches between 1980 and 1994 led to a temporary ban on commercial fishing in federal waters in 1995 (Anonymous, 1995), and there is concern that bycatch of juvenile weakfish by shrimp trawlers at the southern end of the species' range is adversely impacting abundance (Vaughan et al.'). As water temperatures warm in the spring, weakfish move north and inshore into estuaries to spawn. When inshore temperatures cool in the fall, juveniles move south to overwinter off the coast of North Carolina, and older fish are thought to migrate south and offshore (Wilk-). Traditional studies based on tag and recapture data (Nesbit, 1954), scale structure (Perlmutter et al., 1956), morphomet- ric data (Scoles, 1990), and various life history characters (Shepherd and Grimes, 1983, 1984) suggest two or more independent stocks of weakfish. These characters may be influenced by environmental differences, however (Shepherd and Grimes, 1983), and may not reflect genetically distinct (repro- ductively isolated) stocks. Genetic analyses of weakfish stock structure in the mid Atlantic Bight employing allozyme analysis (Crawford et al., 1989) and restriction fragment length polymorphism (RFLP) analysis of whole molecule mitochondrial (mt) DNA (Graves et al, 1992) were unable to reject the null hypothesis that weak- fish along the U.S. east coast comprise a single, genetically homogeneous stock. However, the power of both the analy- ses was limited by low overall genetic variation. Recent analyses of new molecular markers, including microsatellite DNA loci and nuclear gene intron regions, have revealed elevated levels of genetic variation in relation to tra- ditional methods, such as allozymes or RFLP analysis of mtDNA (Miller and Kapuscinski, 1996; Brunner et al., 1998). Although higher levels of genetic variation do not necessarily provide greater stock resolution (Seeb et al., 1998), microsatellite loci have revealed stock structure for some spe- cies, where more traditional molecular markers have not (Bentzen et al., 1996; Ruzzante et al., 1996; Patton et al.. 1997). Similarly, analyses of variable gene intron regions have revealed stock structure within several marine fishes (Palumbi and Baker, 1994; Moran et al., 1997; Leclerc et al., 1996; Chow and Takeyama, 2000). In this study we em- ployed analyses of nuclear and mtDNA markers to evaluate stock structure in weakfish along the east coast of the United States and to investigate pos- sible hybridization between weakfish and other Cynoscion species. Materials and methods Sample collections were restricted to young-of the-year (YOY) fish (less than 140 mm SL) that are reported to remain in their natal estuaries during the first several months of growth (Rowe and Epifanio, 1994). YOY were collected in the summers of 1996 and 1997 from five sites along the east coast of the United States (Fig. 1), maintained on ice after capture, transported to the laboratory, and frozen at -80°C. Muscle tissue was excised from each sample and either stored at -80°C or placed in DMSO buffer (25 mM EDTA, 20% DMSO, saturated NaCl) and stored at room temperature. Genomic DNA was isolated by following the protocol of Sambrook et al. (1989), as modified in Cordes (2000). Specific identification of individuals was determined by using a molecular key for 16 species of Chesapeake Bay * Contribution 2532 of the Virginia Institute of Marine Science. College of William and Mary, Gloucester Point, VA 23062. ' Vaughan, D. S.. R. J. Seagraves. and K. West. 1991. An assessment of the Atlantic weakfish stock, 1982-1988. Atl. States Mar. Fish. Comm. Spec. Rep. 21. 29 p. + tables. Atlantic States Marine Fisheries Commission, 1444 Eve St., NW 6"' Floor, Washington, D.C. 20005. - Wilk, S. J. 1976. The weakfish— a wide ranging species. Atl. States Mar. Fish. Comm, Mar. Resourc. Atl. Coast Fish. Leaflet 18, 4 p. Atlantic States Marine Fisheries Commission. 1444 Eye St., NW 6'*' Floor, Washington. D.C. 20005. Manuscript accepted 21 October 2002. Manuscript received 9 January 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:443-450 (2003). 444 Fisher/ Bulletin 101(2) 45" 40" 35" 30° Figure 1 Sampling locations for young-of-the-ycar (YOY) weakfish iCynoscion regalis) in the summers of 1996 and 1997. Sites are Peconic Bay, New York (NY); Delaware Bay, Delaware (DB); Chesapeake Bay, Virginia (CB); Pamlico Sound, North Carolina (NO; and Doboy Sound, Georgia (GA). sportfishes (including eight species of sciaenids) based on a 12S/16S mtDNA gene region digested with Rsa I (Cordes et al., 2001). Additional 12S/16S mtDNA/Rsa I patterns were generated for silver seatrout iCynsoscion nothus) and sand seatrout (C. arenarius) from the Gulf of Mexico, as well as for banded drum {Larinius fasciatus), gulf kingfish (Menticirrhus littoralis), and star drum iStellifer lanceola- tus) from the South Atlantic Bight following procedures in Cordes etal. (2001). The following microsatellite primers (Table 1 ) developed for red drum (Sciaenops ocellatus) and spotted seatrout {Cynoscion nehulosus) loci were used to amplify weakfish DNA: SOC050 and SOC044 (Turner et al., 19981, SOC014 (Chapman'), and CNE612 (Chapman et al., 1999). Ampli- fications of all microsatellite loci were carried out in 10 pL ^ Chapman, R. W. 1998. Unpubl. data. Marine Resources Re- search Institute, Department of Natural Resources, Charleston, SC 29422. reactions containing 8.30 pL sterile dHgO, 1.0 pL lOx PCR buffer with 15 niM MgCU, 0.20 pL 10 niM dNTP mixture, 0.05 pL forward primer ( 100 pm/pL) labeled with a fluores- cent dye (Licor), 0.20 pL reverse primer (100 pm/pL), 0.05 pL Taq I polymerase (5 U/pL), and 0.20 pL weakfish DNA. Samples were first denatured for 4 min at 95"C, followed by 32 cycles of PCR amplification performed under the follow- ing conditions: 1 min. at 94°C, 1 min. at 50°C, and 1 min. at 72°C. Reactions were given a final 7 min. extension at 72°C. PCR product alleles were separated electrophoreti- cally on a 6% Long Ranger™' polyacrylamide gel with a model 4000 automated DNA infrared sequencer from Li- Cor (Lincoln, NE). Universal actin gene primers developed by G. Warr and M.Wilson (cited in Reece et al., 1997) were used to identify and refine an approximately 800-bp actin intron region lo- cus (CRESIAl) in weakfish (Cordes, 2000). Weakfish PCR amplification products obtained with S7 ribosomal protein intron 2 (RP2) primers originally developed from swordfish Xiphius gladius (Chow and Hazama, 1998) were cloned and sequenced as described in Cordes (2000) and checked against sequences published in Genbank to confirm their identity. The original RP2 primers were then used without modification for amplification of all samples. Both the CRE- SIAl and RP2 amplifications were carried out under the same conditions outlined above for the microsatellite loci, with the exception that the annealing temperature was lowered to 45°C. CRESIAl and RP2 amplification products from a subset of each weakfish collection were screened for polymorphisms with a panel of restriction endonucle- ases and the resulting digestions were separated on 2.5% agarose gels with 1% NuSieve and 1.5% agarose in IxTBE buffer (Cordes, 2000). Gels were stained in IxTBE buffer containing 30 pL (5 mg/niL) ethidium bromide (EtBri, vi- sualized on a Spectroline model TR-302 transilluminator, and photographed with a Polaroid CU-5 land camera. Dra I was the only enzyme that revealed polymorphic restriction sites within CRESIAl, and only Hinf I revealed reliably scored polymorphisms in RP2. All YOY weakfish samples were subsequently screened for variation at CRESIAI/Z)ra IandRP2///in/'I. Microsatellite gel images and restriction enzyme diges- tion patterns for CRESIAl and RP2 were analyzed by using the software program RFLPScan Plus 3.0 (CSPI- Scanalytics, 1996). Statistical analyses for all loci were performed with the Arlequin 1.1 software program of Schneider et al. (1997). Nonparametric, exact-significance tests (exact 0 significance tests and exact probability tests) were used to evaluate sample genotype distributions for departures from Hardy-Weinberg expectations. Unbiased estimators of exact significance probabilities for the Hardy- Weinberg equilibrium tests were calculated by using the Markov chain algorithm of Guo and Thompson (1992) with a Markov chain length of 100,000 steps. Patterns of genetic diversity and divergence within and between populations were evaluated by using the analysis of mo- lecular variance (AMOVA) of Excoffier et al. (1992), which generates F-statistics analogous to the Svaluesof Weir and Cockerham (1984). Significance of /"'-statistics was evalu- ated with exact F permutation procedures (Excoffier et al., NOTE Cordes and Graves: Hyridization and stock structure in Cynosaon regalis 445 1992).Type-I error was controlled for all multiple testing with the sequential Bonferroni method of Rice (1989). Results and discussion Inclusion of nontarget species in weakfish samples Initial analysis of the SOC050 mi- crosatellite data revealed a sig- nificant departure of genotypic frequencies from expectations of Hardy-Weinberg equilibrium for the Georgia 1997 sample, even after correction for mul- tiple tests (a=0.005). Similarly, initial SOC050 AMOVA results indicated a significant within- population variance (P=0.031), and exact F permutation tests of population pairwise F^,^ values resulted in a number of near- significant corrected P values, all involving the Georgia 1997 sample. Inspection of the Georgia 1997 SOC050 alleles revealed a bimodal size distribution due to the presence of several unusually small alleles less than 187 bp in size. It was suspected that alleles in the smaller mode might be the result of misidentified individuals, hybridization, or introgression. Analysis of putative weakfish with small SOC0.50 alleles with the 12S/16S marker of Cordes et al. (2001) resulted in three distinct restriction digestion patterns. One pattern matched that reported for weakfish, and the other two did not match any of the 16 species surveyed by Cordes et al. ( 2001 ). To determine the identity of the unknown pat- terns, voucher samples of five ad- ditional sciaenid species (listed in the "Materials and methods" section above) were analyzed with the 12S/16S mitochondrial marker. The two unknown pat- terns matched those of silver seatrout (Cynoscion nothus) and sand seatrout (C arenarius) (Table 2). The SOC050 locus was subsequently amplified for all silver seatrout (/! = 13) and sand seatrout («=15) samples, produ- o So o c c _ XI JV a •c c ■& •c o o t" <= ^ *J to CO QJ a ^ , T3 -a C 3 CO u: is o x; U CO w o c cc !N CU Qi x 446 Fishery Bulletin 101(2) Table 2 Restriction digestion patterns of the 12S/16S mitochondrial DNA region for putative weakfish (Cynoscion regc lis) individuals in the Georgia 1997 sample, sand seatrout (C. arenarius) and silver seatrout (C. nothus) digested with the enzyme Rsa I. n = number of individuals exhibiting the adjacent pattern. Apparent total size differences may be gel artifacts due to unresolved bands <100 | bp in one or more of the species. Species n Restriction Fragment Sizes (bp) Total size (bp) Georgia 1997 Sample weakfish 3 461 300 200 167 1128 unknown A 7 413 300 200 167 1080 unknown B 5 461 300 256 167 1184 Known standards Cynoscion arenarius 15 461 300 256 167 1184 Cynoscion nothus 13 413 300 200 167 1080 cing allele sizes of 175-181 bp for silver seatrout and 175-193 bp for sand seatrout. Amplification of the silver seatrout and sand seatrout samples with the remaining three microsatellite and two intron loci did not provide fur- ther evidence of hybridization. SOC044 and CNE612 allele size ranges for both species fell within the range exhibited by the weakfish samples, and the SOC014 and both intron loci did not amplify in either the silver seatrout or sand seatrout samples. Individuals with unusually small SOC050 alleles from the Georgia 1997 sample fell into one of four general classes. Seven individuals had silver seatrout mtDNA and two small SOC050 alleles and were presumably pure silver seatrout. The inclusion of these individuals in the collection may not be surprising because both weakfish and silver seatrout are common in the South Atlantic Bight (Bigelow and Schroeder, 1953; Hildebrand, 1955) and are difficult to distinguish during their early life history stages. Although the latter species is known to inhabit deeper waters as adults (Ginsburg, 1931), both species are inshore summer spawners (Devries and Chittenden, 1982; Shepherd and Grimes, 1984). Three individuals possessed sand seatrout mtDNA and two small SOC050 alleles and were presumably pure sand seatrout. Some researchers have suggested that weak- fish and sand seatrout represent separate populations of a single species (Moshin, 1973; Weinstein and Yerger, 1976; Cowan, 1985; Ditty, 1989), and others treat them as separate species (Schlossman and Chittenden, 1981) with distributions confined to the western Atlantic (weakfish) and the Gulf of Mexico (sand seatrout). Paschall (1986) was unable to distinguish between the two species using allozyme electrophoresis. In contrast, results presented here are consistent with the existence of two distinct spe- cies, with weakfish and sand seatrout co-occurring off the east coast of the United States at least as far north as Doboy Sound, Georgia. This distribution pattern is consis- tent with the phylogeographic patterns of 19 freshwater, coastal, and marine species distributed along the U.S. East Coast and the Gulf of Mexico that exhibited geographically concordant forks in their intra- or interspecific mtDNA phylogenies (or in both phylogenies) (Avise, 1992). In the present situation, apparently distinct Gulf (sand seatrout) and Atlantic (weakfish) species may have reestablished contact in a hybrid zone (see below) through movement of the Gulf species into the Atlantic. Three individuals had weakfish mtDNA and a single small SOC050 allele and were presumably hybrids of weakfish and sand seatrout or silver seatrout (with female weakfish parentage). In addition, two individuals pos- sessed sand seatrout mtDNA and a single small SOC050 allele and were presumably hybrids of weakfish and sand seatrout with female sand seatrout parentage. These data suggest that hybridization occurs between weakfish and sand seatrout and that the genetic exchange is not gender restricted. Because of the overlap in microsatellite allele sizes seen between silver seatrout and sand seatrout, hy- bridization between weakfish and silver seatrout could not be excluded. The lack of suspected hybrids with silver seat- rout mtDNA, however, suggests that hybridization did not involve this species. The possibility exists that the putative hybrids are in fact weakfish with rare mtDNA haplotypes common to the three Cynoscion species studied here. This seems unlikely because only one 12S/16S mtDNA//?,sn I pattern was noted among 40 weakfish in the species identification study of Cordes et al. (2001). Furthermore, analysis of 20 weakfish taken from each of the four loca- tions outside of Georgia with the 12S/16S marker revealed no new mtDNA patterns. Also, the mtDNA haplotypes seen in sand seatrout and silver seatrout seem to vary in size and can not be clearly related to the weakfish haplotype by the addition or deletion of presumed restriction sites. This condition is more in keeping with mtDNA of different species, although the apparent size differences may be gel artifacts due to unresolved bands <100 bp in one or more of the species. Re-evaluation of the remaining 1996-97 SOC050 data re- vealed occasional occurrences of small alleles in individual fish m all but the New York samples (Table 3). Examina- tion of the 12S/16S mtDNA region of these individuals identified a single silver perch (Bairdiella chrysoiiro) in the Chesapeake Bay 1997 sample (silver perch mtDNA NOTE Cordes and Graves: Hyridization and stock structure in Cynosaon regalis 447 Table 3 Frequencies of unusual alle es in four geographical sam- pies of weakfish (Cynoscion regc ilis) taken in 1996 and 1997. The number of individuals with anoma lous alleles that were subsequently eliminated from the population structure analysis is given in parentheses after the sample | names. Sample Allele (bp) Frequency Georgia 1996 (4) 175 0.018 177 0.009 179 0.009 North Carolina 1996(1) 177 0.010 Chesapeake Bay 1996(1) 177 Delaware Bay 1996 (2) 179 0.011 181 0.011 Georgia 1997(15) 171 0.021 173 0.010 175 0.125 177 0.094 179 0.010 North Carohna 1997(1) 177 0.009 Chesapeake Bay 1997 (3) 171 0.009 177 0.009 179 0.009 Delaware Bay 1997 (5) 177 0.023 181 0.034 and two alleles 171 bp in size). All other individuals were putative hybrids with weakfish mtDNA and a single small SOC050 allele characteristic of silver and sand seatrout. As mentioned previously, subsequent analysis of 20 weakfish taken from each of the four locations outside of Georgia with the 12S/16S marker revealed only weakfish mtDNA. If the small SOC050 alleles found in the more northern populations are not simply rare weakfish alleles shared in common with the other two Cynoscion species, they may indicate that introgressive hybridization is responsible for the migration of the smaller alleles into northern waters (although the northward movement of hybrid fish out of the contact zone cannot be excluded). As a result of these findings, all individuals in the 1996 and 1997 collections exhibiting at least one small SOC050 allele less than 183 bp in length were eliminated from the population structure analyses. Stock structure analysis All four microsatellite loci were polymorphic in all sampled locations in both years. Allele frequency distributions for each locus are available from the authors upon request. Sample sizes in), number of alleles (A''), expected hetero- zygosities (gene diversities), and significance test results for Hardy-Weinberg equilibrium are provided in Table 4. Levels of variation differed greatly among the four mic- rosatellite loci. The number of alleles ranged from two (SOC014) to 37 (CNE612), and average expected heterozy- gosities ranged from 0.085 (SOC014) to 0.928 (CNE612). These values are consistent with heterozygosity ranges reported in other multilocus microsatellite studies on spe- cies including Atlantic cod (Bentzen et al., 1996), northern pike (Miller and Kapuscinski, 1996), pink and sockeye salmon (Seeb et al., 1998), and Arctic char (Brunner at al., 1998). In contrast, Crawford et al. (1989) and Graves et al. (19921 found very low levels of genetic variation in an analysis of weakfish populations using allozymes and mtDNA restriction fragment-length polymorphism (RFLP) analyses, respectively. None of the genotypic distributions for any of the four microsatellite loci at any of the collection locations in either year differed significantly from Hardy- Weinberg expectations after correcting for multiple tests (Table 4). Digestion of actin intron (CRESIAl) amplifications with the restriction endonuclease Rsa I revealed a single poly- morphic restriction site that produced two alleles. Expected heterozygosities ranged from 0.000 for the monomorphic Georgia 1997 sample to 0.096 for the Chesapeake Bay 1996 sample. Digestion of the RP2 amplifications with the restriction endonuclease Hinf I also resulted in two alleles. Expected heterozygosities ranged from 0.194 in the Delaware Bay 1997 sample to 0.370 in the Georgia 1997 sample. Levels of genetic variation within the two nuclear gene intron regions were low in relation to three of the four microsatellite loci and were more similar to those found in the polymorphic allozyme loci of Crawford et al. ( 1989). In another study where nuclear intron RFLP analysis was used, similar levels of heterozygosity in Pacific salmon were found (Moran et al., 1997), as in RFLP studies of anonymous single copy nuclear (ascn) DNA loci in Atlantic cod {Gadus morhua) (Pogson et al., 1995) and blue marlin {Makaira nigricans) (Buonaccorsi et al., 1999). In con- trast, higher heterozygosities (44-58%) were reported in an ascnDNA/RFLP analysis of striped bass {Morone saxa- tilis) by Leclerc et al. (1996). None of the sample genotype distributions for either locus differed significantly from Hardy-Weinberg expectations after correcting for multiple tests (Table 4). To test for population structure, microsatellite loci were analyzed individually and as a combined data set. AMOVA results did not reveal significant differences between sam- ple locations or years for any of the four loci or for the com- bined data (all P>0.05). Single-locus population pairwise Fgy values were relatively low, and mean Fj^.j. values ranged from 0.002 (SOC050 and CNE612) to 0.018'(SOC044). Exact F permutation tests were not significant for any of the four loci or the combined data set after correction for multiple testing. AMOVA results for both the actin and RP2 loci indicated no significant differences between sample locations or years (all P>0.05). Single-locus population pairwise F^^ values for the actin locus were consistently low (mean=0.005), ranging from Fgj < 0.000 for most of the comparisons to an Fg-p of 0.035 between Georgia 1996 and Georgia 1997 and between Chesapeake Bay 1996 and Georgia 1997. A single exact F permutation test, between Delaware 1996 448 Fishery Bulletin 101(2) Table 4 Sample sizes in), number of alleles (A^), expected heterozygosities ^H ) and P values for tests of Hardy-Weinberg equilibrium for four microsatellite loci the actin intron (CRESIAl), and the ribosomal protein 2 intron (RP2) gene regions GA = Georgia, NC = North Carolina, CB = Chesapeake Bay, DB = Del aware Bay, NTY = New York. NT = monomorph c sample not tested. GA 1996 NC 1996 CB 1996 DB 1996 NY 1996 GA 1997 NC 1997 CB 1997 DB 1997 NY 1997 SOC050 n 51 49 64 46 46 33 52 55 42 54 N 7 6 6 5 7 6 6 6 8 7 «exp 0.741 0.702 0.694 0.731 0.724 0.758 0.712 0.740 0.737 0.722 P' 0.067 0.542 0.507 0.130 0.959 0.566 0.577 0.721 0.349 0.174 SOC044 n 47 46 63 55 55 36 60 56 52 56 N 2 3 2 2 2 2 2 2 2 2 «e,p 0.362 0.434 0.374 0.251 0.416 0.407 0.302 0.350 0.203 0.419 P' 1.000 0.019 0.496 0.303 0.512 0.010 0.669 0.116 0.0.512 0.198 SOC014 n 43 48 64 52 54 39 56 55 52 57 N 2 2 2 2 2 2 2 2 2 2 ^exp 0.090 0.081 0.046 0.075 0.170 0.144 0.053 0.088 0.038 0.068 P' 1.000 1.000 1.000 1.000 1.000 1.000 0.027 1.000 1.000 1.000 CNE612 n 39 43 62 50 46 33 54 52 50 56 N 17 23 20 20 22 19 25 23 22 23 ^exp 0.916 0.943 0.928 0.912 0.916 0.935 0.934 0.934 0.923 0.936 P' 0.050 0.113 0.898 0.530 0.238 0.752 0.522 0.419 0.060 0.290 CRESIAl n 40 42 40 42 40 36 51 54 45 55 N 2 2 2 2 2 1 2 2 2 2 «exp 0.096 0.089 0.031 0.055 0.096 0.000 0.025 0.053 0.047 0.020 P' 0.076 0.091 1.000 0.036 0.078 NT 1.000 1.000 1.000 1.000 RP2 n 48 45 45 42 41 29 48 49 42 41 N 2 2 2 2 2 2 2 2 2 2 ^exp 0.237 0.200 0.217 0.230 0.253 0.373 0.237 0.201 0.194 0.253 P' 0.184 0.432 0.104 0.120 0.180 0.298 0.189 0.465 0.052 0.179 ' None of the samples differed significar tly from Hardy-Weinberg expectations after sequential Bonferroni corrections (a=0.005l. and Georgia 1997, was significant after correction for mul- tiple testing (a<0.001). Single-locus population pairwise Ffj.]. values for the RP2 locus were also low (mean=0.006), ranging from F^-^ < 0.000 for most of the comparisons to a high of 0.050 between Georgia 1997 and Delaware Bay 1997. None of the exact F permutation tests were signifi- cant after correction for multiple testing. From our results we were unable to reject the null hy- pothesis that weakfish comprise a single, genetically ho- mogeneous stock. These results are consistent with those based on allozymes (Crawford et al., 1989) and RFLP analysis of mtDNA (Graves et al., 1992) and illustrate the point that increased genetic variability in microsatellites in relation to more traditional markers will not always provide greater stock resolution (Seeb et al., 1998). The amount of genetic exchange necessary to prevent the ac- cumulation of significant genetic divergence between fish from different locations may be as little as a few individu- als per generation (Allendorf and Phelps, 1981). Weakfish tagging data indicate that low levels of exchange occur between geographically distant populations of weakfish (Bain et al., 1998). Estimates of natal homing in yearling weakfish, calculated by Thorrold et al. (2001) using geo- chemical signatures in the otoliths of the same weakfish used in the present study, indicated spawning-site fidelity ranging from SI'S to 81%, suggesting exchange rates suf- ficient to prohibit genetic divergence between locations. The inclusion of nontarget species in our weakfish sam- ples illustrates the advantages in using multiple marker systems. If only a single microsatellite locus had been used, or if the study had been restricted to nuclear intron markers alone, it is very likely that the sand seatrout and NOTE Cordes and Graves: Hyridization and stock structure in Cynosdon regalis 449 silver seatrout specimens would have gone unnoticed. This could easily have resulted in a type-II error. Likewise, if nongenetic markers such as otolith microchemistry had been used exclusively, the analyses ofThorrold et al. (2001) would have been based on a mixed-species sample. Instead, it was possible not only to recognize the individuals as anomalous but also to identify them to species and pro- vide evidence of hybridization between at least two of the Cyrioscion congeners. It is hoped that further refinement of the inter- and intraspecific molecular markers developed here and in other studies will eventually be helpful in fur- ther clarifying the taxonomic status, population structure, and possible hybridization within the genus Cynoscion. Acknowledgments We would like to thank all those who supplied us with weak- fish samples, including Louis Barbieri. Susan Lowerre- Barbieri, Christina Grahn, Patrick Geer, Mike Greene, and Simon Thorrold. We also appreciate the samples of banded drum, gulf kingfish, and star drum provided by Trey Knott and the silver seatrout and sand seatrout specimens pro- vided by Bill Karel. We gratefully acknowledge Kim Reece, John Gold, Linda Richardson, and Robert Chapman for gen- erously providing us with their published and unpublished primer sequences. Funding for this study was provided through the Virginia Marine Resources Commission. Literature cited Allendort; F. W., and S. R. Phelps. 1981. Use of allelic frequencies to describe population structure. Can. J. Fish. Aquat. Sci. 38(12):1507-1514. Anonymous. 1995. Overfished weakfish stock forces closure of federal waters. Fisheries 21:46-47. Avise, J. C. 1992. Molecular population structure and the biogeographic history of a regional fauna: a case history with lessons for conservation biology. Oikos 63:62-76. Bain C, J. Lucy, and M. Arendt. 1998. Virginia game fish tagging program annual report 1997. Virginia Marine Resource Report 98-3 (VSG-98- 01), 22 p. Virginia Institute of Marine Science. Virginia Sea Grant Marine Advisory Program, College of William and Mary, Gloucester Point, VA. Bentzen P., C. T. Taggart, D. E. Ruzzante, and D. Cook. 1996. Microsatellite polymorphism and the population structure of Atlantic cod tGadus morhua) in the northwest Atlantic. Can. J. Fish. Aquat. Sci. .53 ( 121:2706-2721. Bigelow H., and W. Schroeder. 1953. Fishes of the Gulf of Maine. U.S. Fish and Wildlife Service Fish. Bull. 53, 577 p. Brunner P. C, M. R. Douglas, and L. Bernatchez. 1998. Microsatellite and mitochondrial DNA assessment of population structure and stocking effects in Arctic charr Salvelmux alpinus (Teleostei: Salmonidael from central Alpine lakes. Mol. Ecol. 7 (2): 209-223. Buonaccorsi V. P., K. S. Reece, L. W. Morgan, and J. E. Graves. 1999. Geographic distribution of molecular variance within the blue marlin {Makaira nigricans): a hierarchical analysis of allozyme, single-copy nuclear DNA, and mitochondrial DNA markers. Evolution 53 ( 2 1:568-579. Chapman R. W., G. R. Sedberry, J. C. McGovern, B. A. Wiley, and J. A. Musick. 1999. The genetic consequences of reproductive variance: studies of species with different longevities. In Life in the slow lane: ecology and conservation of long-lived marine animals. Am. Fish. Soc. Symp. 23:169-181. Chow S., and K. Hazama. 1998. Universal PCR primers for 87 ribosomal protein gene introns in fish. Mol. Ecol. 7:1247-1263. Chow, S., and H. Takeyama. 2000. Intron length variation observed in the creatine kinase and ribosomal protein genes of the swordfishX;p/zJas gladius. Fish. Sci. 64(3):397-402. Cordes, J. F 2000. Application of genetic markers to provide species iden- tification and define stock structure: analyses of selected marine fishes of the mid-Atlantic bight. Ph.D. diss., 142 p. Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA. Cordes, J. F., S. L. Armknecht, E. A. Starkey, and J. E. Graves. 2001. Forensic identification of sixteen species of Chesa- peake Bay sportfishes using mitochondrial DNA restriction fragment-length polymorphism (RFLP) analysis. Estuar- ies 24(1 ):49-58. Cowan, J. H., Jn 1985. The distribution, transport and age structure of drums (family Sciaenidae) spawned in the winter and early spring in the continental shelf waters off west Louisiana. Ph.D. diss., 182 p. Louisiana State University, Baton Rouge, LA. Crawford, M. K., C. B. Grimes, and N. E. Buroker 1989. Stock identification of weakfish, Cynoscion regalis, in the middle Atlantic region. Fish. Bull. 87: 205-211. CSPI-Scanalytics. 1996. RFLPScan Plus, version 3.0. SCPI-Scanalytics, Bil- lerica, MA. DeVries, D. A., and M. E. Chittenden Jr 1982. Spawning, age determination, longevity, and mortal- ity of the silver seatrout. Cynoscion nothus. in the Gulf of Mexico. Fish. Bull. 80:487-500. Ditty J. G. 1989. Separating early larvae of sciaenids from the western North Atlantic: a review and comparison of larvae from the northern Gulf of Mexico off Louisiana and Atlantic coast of the U.S. Bull. Mar Sci. 44:1083-1105. Excoffier L., P. Smouse, and J. Quattro. 1992. Analysis of molecular variance inferred from metric dis- tances among DNA haplotypes: application to human mito- chondrial DNA restriction data. Genetics 131:479^91. Ginsburg, I. 1931. On the differences in the habitat and size of Cynoscion arenarius and Cynoscion nothus. Copeia 1931:144. Guo, S., and E. Thompson. 1992. Performing the exact test of Hardy-Weinberg propor- tion for multiple alleles. Biometrics 48:361-372. Graves, J. E., J. R. McDowell, and M. L. Jones. 1992. A genetic analysis of weakfish Cynoscion regalis stock structure along the mid-Atlantic coast. Fish. Bull. 90:469-475. Hildebrand, H. H. 1955. A study of the fauna of the pink shrimp (Penaeus duor- arum Burkenroad) grounds in the Gulf of Campeche. Publ. Inst. Mar Sci. 4:169-232. Leclerc, G. M., M. Diaz, and B. Ely 1996. Use of PCR-RFLP assays to detect genetic variation at 450 Fishery Bulletin 101(2) single-copy nuclear loci in striped bass (Morone saxatilis). Mol. Mar. Biol. Biotechnol. 5:138-144. Miller, L. M.. and A. R. Kapuscinski. 1996. Microsatellite DNA markers reveal new levels of genetic variation in northern pike. Trans. Am. Fish. Soc. 125(6):971-977. Moran, P., D. A. Dightman, R. S. Waples, and L. K. Park. 1997. PCR-RFLP analysis reveals substantial population- level variation in the introns of Pacific salmon (Oncorhyn- chus spp.). Mol. Mar Biol. Biotechnol. 6(4);315-327. Moshin, A. K. Mohammad. 1973. Comparative osteology of the weakfishes (Cynoscion ) of the Atlantic and Gulf coasts of the United States ( Pisces- Sciaenidae). Ph.D. diss., 148 p. Texas A&M Univ., College Station, TX. Nesbit, R. A. 1954. Weakfish migration in relation to its conservation. U.S. Fish Wildl. Serv., Spec. Sci. Rep. Fish. 115:81. Palumbi, S. R., and C. S. Baker 1994. Contrasting population structure from nuclear intron sequences and mtDNA of humpback whales. Mol. Biol. Evol. 11:426-435. Paschall, R. L. 1986. Biochemical systematics of the seatrouts of the western Atlantic genus Cynoscion. Master's thesis, 100 p. Univ. New Orleans, New Orleans, LA. Patton, J. C, B. J. Galloway, R. G. Fechhelm, and M. A. Cronin. 1997. Genetic variation of microsatellite and mitochondrial DNA markers in broad whitefish (Coregonus nasus) in the Colville and Sagavanirktok rivers in northern Alaska. Can. J. Fish. Aquat. Sci. 54: 1548-1556. Perlmutter, A., S. W. Miller, and J. C. Poole. 1956. The weakfish (Cynoscion regalis) in New York waters. N.Y. Fish Game 3:1^3. Pogson, G. H., K. A. Mesa, and R. G. Boutilier 1995. Genetic population structure and gene flow in the Atlantic cod ^adus morhua: a comparison of allozyme and nuclear RFLP loci. Genetics 139:375-385. Reece, K. S., M. E. Siddall, E. M. Burreson, and J. E. Graves. 1997. Phylogenetic analysis ofPerkinsus based on actin gene sequences. J. Parasitol. 83(3):417-423. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-225. Rowe, P. M., and C. E. Epifanio. 1994. Tidal stream transport of weakfish larvae in Delaware Bay, USA. Mar Ecol. Prog. Ser 110:105-114. Ruzzante, D. E., C. T. Taggart, and D. Cook. 1996. Spatial and temporal variation in the genetic compo- sition of a larval cod (Gadus morhua) aggregation: cohort contribution and genetic stability. Can. J. Fish. Aquat. Sci. 53:2695-2705. Sambrook, J., E. F Fritsch, and T Maniatis. 1989. Molecular cloning: a laboratory manual, 3 vols. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schlossman, P. A., and M. E. Chittenden Jr. 1981. Reproduction, movements, and population dynamics of the sand seatrout, Cynoscion arenarius. Fish. Bull. 79: 649-669. Schneider, S., J.-M. Kueffer, D. Roessli, and L. Excoffier 1997. Arlequin: a software for population genetics data analysis (vers. 2.0), 111 p. Univ. Geneva, Switzerland. Scoles, D. R. 1990. Stock identification of weakfish, Cynoscion regalis, by discriminant function analysis of morphometric characters. Masters thesis, 55 p. Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA. Seeb, J. E., C. Habicht, J. B. Olsen, R Bentzen, J. B. Shaklee, and L. W. Seeb. 1998. Allozyme, mtDNA, and microsatellite variants de- scribe structure of populations of pink and sockeye salmon in Alaska. Bull. NPAFC 1:300-319. Shepherd, G. R., and C. B. Grimes. 1983. Geographic and historic variations in growth of weak- fish, Cynoscion regalis, in the middle Atlantic Bight. Fish. Bull. 81:803-813. 1984. Reproduction of weakfish, Cynoscion regalis, in the New York Bight and evidence for geographically specific life history characteristics. Fish. Bull. 82:501-511. Thorrold, S. R., C. L. Latkoczy, P. K. Swart, and C. M. Jones. 2001. Natal homing in a marine fish metapopulation. Sci- ence 291:297-299. Turner, T. F, L. R. Richardson, and J. R. Gold. 1998. Polymorphic microsatellite DNA markers in red drum [Sciaenops ocellatus). Mol. Ecol. 7( 12):1771-1788. Weinstein, M. P., and R. W. Yerger 1976. Protein taxonomy of the Gulf of Mexico and Atlantic Ocean seatrouts, genus Cynoscion. Fish. Bull. 74:599- 607. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38:1358-1370. 451 Dynamic age-length keys Are Salthaug Institute of Marine Research Nordnesgaten 50 PO Box 1870 N-5817 Bergen, Norway E-mail address arese'imrno Information about age composition is important when analyzing fish popu- lation dynamics. Age determination of individual fish is more difficult and time consuming than the recording of length measurements but by using age-length keys, age distributions can be estimated without much difficulty from length distributions (Fridrikson, 1934). Knowledge of the age-length composition in the population or in a given subgroup of the population is required for constructing adequate age-length keys. Various methods for construction and evaluation of age-length keys are described in the literature (see e.g. Fridrikson, 1934; Macdonald and Pitcher, 1979; Schnute and Fournier, 1980; Kimura and Chi- kuni, 1987; Hayes, 1993; Terceiro and Ross, 1993; Goodyear, 1997). Because of individual variation in growth rates and the variation in mortality rates at different ages and sizes, the age and length composition of a fish stock are constantly changing. With sufficient information about a fish stock, the change in the age-length composition can be modeled and theoretical age- length keys can be constructed for specific time periods. Age distributions can then be estimated from length distributions taken at different times of the season. In this work, a simple but useful modeling approach for con- structing dynamic age-length keys is described and applied to data from the Atlantic cod [Gadus morhua) stock in the Barents Sea. probability of an individual being a certain length (/) within an age group (a ) at a given time is assumed to follow a normal probability density function (Fig. lA), Niji^, fj^), with expectation s^ and standard deviation o^. When lengths of individual fish are recorded, they are normally classified as discrete length groups (e.g. 1-cm or 5-cm length intervals). The probability (P) for an individual in age group a to belong in a discrete length group, s, at a given time is then given by P.S ] N{^i,,s ■ age=8 • \,f''^^ ^/ o • age=9 01 ^^ 1^ » age=10+ 0 0 or'-' n n 0.0 01 0.2 03 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1986 1990 ^ 08 r 0.5 ^^ ». /^ 0.4 /^ ° 06 • X^ o/d . \X 0.3 ^, ^y^ ay^ o, • 0.4 ^^ 0.2 A i^ 01 • ^^^..^ ^ 0.2 1/^' • m^ l^""^ -jTa! ■ - 0 0 ^ 0 0 t^*>^ ^ t o 0.0 0.1 0.2 0.3 0.4 0.5 0-0 01 0.2 0.3 0.4 0,5 0.6 0-7 0,8 Q. O Q. "O 0) 1987 1991 .S 0.5 r . 0,8 o y^ 0 ■ ^y^ 0.4 "^t^o ■ y^ A 5^ Op ■ ^^^ A o^ o 0.3 .,^^r 0^ ■ -^ 0.2 J*o ■ ■ ^y^ .^'^ • 02 ° /^ 0.1 -A m^^^ ■■ • 10^ .^^•' 00 J^ 0 0 LgjClr^ 0.0 0.1 0.2 0.3 04 05 00 0.1 0.2 0.3 0.4 0.5 0.6 07 08 1988 1992 0.6 ^^ 06 ♦ ^/ 0.5 ^< 05 • j^ 0.4 ^V^ * ° y^^ 0.3 A ^^ °^ ^y"^ 0.2 o^o uy^ ♦ 5^i" ■ 01 X m ■ 01 A \< "* • L^ k# ■ d^^ * 0 0 K0 0 n 00 0 1 0.2 0 3 0.4 0.5 0.6 0 0 0 1 0.2 0.3 0.4 0.5 0.6 Predicted proportion Figure 3 Observed anc predicted proportions of different age groups in monthly samples (f!>300) from commercial catche.s in the period 1985-2000. Each age group has its own symbol (see plot for 1985 and 1993). The diago- nal is shown, which is where the points should lie. Note that the range on the axes varies between years according to the maximum values NOTE Salthaug; Dynamic age-length keys 455 o ■c 1993 1994 00 01 02 03 04 1995 1996 1998 1999 2000 0 5 0 6 0 0 0 1 Predicted proportion Figure 3 (continued) 456 Fishery Bulletin 101(2) Literature cited Foumier, D. A., J. R. Sibert, J. Majkowski, and J. Hampton. 1990. MULTIFAN a likelihood-based method for estimat- ing growth-parameters and age composition from multiple length frequency data sets illustrated using data for south- em bluefin tuna (Thunnus maccoyii). Can. J. Fish. Aquat. Sci. 47:301-317. Fridrikson, A. 1934. On the calculation of age distribution within a stock of cod by means of relatively few age determinations as a key to measurements on a large scale. Rapp. P. V. Reun. Cons. Int. ExplorMer 86:1-14. Goodyear, C. P. 1997. Fish age determined from length: an evaluation of three methods using simulated red snapper data. Fish. Bull. 95:39-46. Hayes, B. 1 993 . A statistical method for evaluating differences between age-length keys with application to Georges Bank haddock, Melanogrammus aeglefinus. Fish. Bull. 91:550-557. Jorgensen, T. 1992. Long-term changes in growth of North-east Arctic cod (Gadus morhua) and some environmental influences. ICES J. Mar Sci. 49:263-277. Kimura, D. K., and S. Chikuni. 1987. Mixtures of empirical distributions: an iterative appli- cation of the age-length key. Biometrics 43:23-35. Macdonald, P D. M., and T. J. Pitcher 1979. Age-groups from length-frequency data: a versatile and efficient method of analyzing distribution mixtures. J. Fish. Res. Board Can. 36:987-1001. Pennington, M., L-M. Burmeister, and V. Hjellvik. 2002. Assessing trawl-survey estimates of frequency dis- tributions. Fish. Bull. 100:74-80. Schnute, J., and D. Foumier. 1980. A new approach to length-frequency analysis: growth structure. Can. J. Fish. Aquat. Sci. 37:1337-1351. Terceiro, M., and J. L. Ross. 1993. A comparison of alternative methods for the estima- tion of age from length data for Atlantic coast bluefish iPomatomus saltatrix). Fish. Bull. 91:534-549. Tuljapurkar, S., and H. Caswell. 1997. Structured-population models in marine, terres- trial, and freshwater systems, 570 p. Chapman & Hall, London. 457 Effects of blood extraction on horseshoe crabs iLimulus polyphemus) Elizabeth A. Walls Department of Fisheries and Wildlife Sciences Virginia Polytecfinic Institute and State University Blacksburg, Virginia 24061-0321 Present address: Center for Environmental Studies Virginia Commonwealth University 1000 West Cary Street, Box 843050 Richmond, Virginia, 23284 Jim Berkson Department of Fisheries and Wildlife Sciences Virginia Polytechnic Institute and State University Blacksburg, Virginia 24061-0321 E-mail address (for J Berkson. contact author); |berkson@vt.edu the United States, mandated that all biomedical companies actively bleed- ing horseshoe crabs estimate mortal- ity rates resulting from their bleeding process (Schrading et al.'^). Because of the unique methods of the different biomedical companies, each company was required to quantify its own rate of mortality. BioWhittaker, a CAMBREX com- pany, is the largest producer of LAL. In response to the ASMFC mandate, BioWhittaker requested that Virginia Tech conduct the mortality study for their company. Our objective was to determine horseshoe crab mortality for a two-week period following the bleeding process. Methods Horseshoe crabs ILimulus polyphemus) are caught by commercial fishermen for use as bait in eel and whelk fisheries ( Berkson and Shuster, 1999) — fisheries with an annual economic value of $13 to $17 million (Manion et al.M. Horse- shoe crabs are ecologically important, as well (Walls et al., 2002). Migratory shorebirds rely on horseshoe crab eggs for food as they journey from South American wintering grounds to Arctic breeding grounds (Clark, 1996). Horse- shoe crabs are also essential for public health (Berkson and Shuster, 1999). Biomedical companies bleed horse- shoe crabs to extract a chemical used to detect the presence of endotoxins pathogenic to humans in injectable and implantable medical devices (Novitsky, 1984; Mikkelsen, 1988). Bled horseshoe crabs are returned to the wild, subject to the possibility of postbleeding mortal- ity. Recent concerns of overharvesting have led to conflicts among commercial fishermen, environmentalists acting on behalf of the shorebirds, and biomedi- cal companies (Berkson and Shuster, 1999; Walls etal., 2002). In order to create an effective, sus- tainable management policy for the horseshoe crab resource, the comple- tion of a stock assessment that incor- porates human-induced mortalities is necessary. A stock assessment is not currently available because of a lack of critical information on the horseshoe crab population (Berkson and Shuster, 1999). One critical piece of information needed is an estimate of the mortali- ties involved in the biomedical bleed- ing process. With an estimated 260,000 horseshoe crabs bled in 1997 (HCTC2), the last year with data available, mor- talities may not be negligible. Five biomedical companies on the At- lantic coast of the United States bleed horseshoe crabs in the laboratory for the production of Limulus Ameobocyte Lysate (LAL). The horseshoe crabs are caught by fishermen under contract to biomedical companies, bled, then re- turned to their point of capture. The LAL test used to detect endo- toxins in humans is derived from the blue, copper-based blood of the horse- shoe crab. Although alternate tests exist for the detection of endotoxin, the LAL test is the most effective because it is capable of detecting as little as one millionth of a billionth of a gram of endotoxin (Mikkelsen, 1988). The LAL test is now a standard test used to pro- tect human health around the world, and horseshoe crabs are the sole source of LAL. Each biomedical company maintains its own procedures for harvesting horseshoe crabs, extracting the horse- shoe crabs' blood, releasing the bled horseshoe crabs, and developing the LAL substance. In 1998, the Atlantic States Marine Fisheries Commission (ASMFC), the Commission responsible for horseshoe crab management in We compared mortality rates between horseshoe crabs that underwent the bleeding process (bled) and horseshoe crabs that were suitable to undergo the bleeding process but were not bled (unbled). Throughout the 1999, 2000, and 2001 bleeding seasons (June through August), BioWhittaker obtained horseshoe crabs by trawling in the Atlantic Ocean off the coasts of Chincoteague, Virginia, or Ocean City, Maryland (or off both coasts). After capture, the horseshoe crabs were brought to BioWhittaker's bleeding ' Manion, M. M., R. A. West, and R. E, Uns- worth. 2000. Economic assessment of the Atlantic coast horseshoe crab fishery, 71 p. Division of Economics, U.S. Fish and Wildlife Service, Arlington, VA. 2 HCTC (Horseshoe Crab Technical Committee). 1998. Status of the horse- shoe crab {Limulus polyphemus) popula- tion of the Atlantic coast, 9 p -t- figures and tables. Horseshoe Crab Technical Com- mittee, Atlantic States Marine Fisheries Commission. Washington, D.C. 3 Schrading, E., T. O'Connell, S. Michels, and P. Perra. 1998. Interstate management plan for horseshoe crab, 59 p. Atlantic States Marine Fisheries Commission, Washington D.C. Manuscript accepted 6 November 2002. Manuscript received 9 January 2003 at NMFS Scientific Pubhcations Office. Fish. Bull. 101:457-459 (2003). 458 Fishery Bulletin 101(2) Table 1 Comparison of mortality rates between bled and unbled groups of horseshoe crabs captured near Chincoteague, Virginia, and Ocean City, Maryland, 1999-2001. Dates Unbled horseshoe crabs Bled horseshoe crabs No. of crabs No. of crabs % dead at No. of crabs No. of crabs % dead at monitored monitored that died study end monitored that died study end 8-22 Jul 99 10 0 0% 10 0 0% 22 Jul 99-5 Aug 99 10 0 0% 10 3 30% 19 Jun 00-3 Jul 00 30 0 0% 30 0 0% 7-21 Jul 00 30 0 0% 30 0 0% 1-15 Aug 00 30 1 3.3% 30 6 20% 6-20 Jun 01 30 0 0% 30 0 0% 20 Jun 01-04 Jul 01 30 0 0% 30 2 6.7% 15-29 Aug 01 30 0 0% 30 5 16.7% Total 200 1 0.5'J 200 16 SV, facility in Chincoteague, Virginia. At the bleeding facil- ity, we randomly selected a predetermined number ( 10 in 1999, 30 in 2000 and 2001) of newly matured male horse- shoe crabs (identified by pristine shell condition and the presence of boxing-glove lower claws [Shuster*] ) from all of the horseshoe crabs obtained in that day's trawls. We selected newly matured male horseshoe crabs to minimize covariance in our study. These horseshoe crabs were not bled and served as a control in the experiment. They were packed in coolers labeled "unbled," and set aside. The same number of newly matured male horseshoe crabs were then randomly selected from the remaining horseshoe crabs and underwent BioWhittaker's normal bleeding process. Upon completion of the bleeding process, the horseshoe crabs were packed in coolers labeled "bled." All coolers containing horseshoe crabs, both bled and unbled, were immediately packed in an air-conditioned ve- hicle and transported to the Virginia Seafood Agricultural Research and Extension Center in Hampton, Virginia. The horseshoe crabs were removed from the coolers and the unbled horseshoe crabs were marked with external tags to distinguish them from the bled horseshoe crabs. These markings were unobtrusive and did not cause any undue stress to the unbled horseshoe crabs. All of the horseshoe crabs were placed in four replicated, flow-through holding tanks, and equal numbers of bled and unbled horseshoe crabs were held in each tank. The horseshoe crabs remained in the tank system at Hampton for two weeks. Horseshoe crabs were maintained in appropriate conditions (Brown and Clapper, 1981), and monitored daily. Horseshoe crabs that died during the two-week period were removed and returned to the ocean at the time of their death. At the conclusion of each two-week period, the status of each horseshoe crab (dead or alive) was recorded. All •* Shuster, C. N., Jr 1999. Managing the horseshoe crab re- source: it's the adult age that counts, 32 p. Virginia Institute of Marine Science, College of William and Marv, (Gloucester Point, VA. surviving horseshoe crabs were removed from the tank, placed in coolers, packed in an air-conditioned vehicle, re- turned to BioWhittaker's bleeding facility in Chincoteague, Virginia, and returned to the Atlantic Ocean in accordance with BioWhittaker's standard operating procedures. This procedure was repeated eight times during summers 1999, 2000, and 2001. The results from each of the replicates were combined, and the overall percentage mortality was calculated for the bled and unbled groups. Using Fisher's exact test, we evaluated statistical sig- nificance of differences in mortality between the bled and unbled horseshoe crabs (Mehta and Patel, 1999). We then calculated a 957? confidence interval for average differen- tial mortality using the common odds ratio in the statistical program StatXact (Mehta and Patel, 1999). Results A Fisher's exact test for statistical significance showed differences between mortality rates in bled and unbled horseshoe crabs (P=2.085E-04). Bled horseshoe crabs (n=200) had an overall mortality rate of S'J't compared to the 0.5% mortality rate of the unbled horseshoe crabs («=200; P<0.001) (Table 1). Thus, this study estimates average differential mortality between bled and unbled horseshoe crabs to be 7.5'y. The 95^^ confidence interval for this average differential mortality ranges from 0.14% to 38.1% as calculated with the common odds ratio (Mehta and Patel, 1999). Discussion Our results indicate that horseshoe crab mortality due to bleeding is relatively low. Two small-scale studies had pre- viously estimated postbleedmg mortality. Rudloe (1983), observing bled and unbled horseshoe crabs in a penned cove in Florida, found that bleeding increased mortality by NOTE Walls and Berkson: Effects of blood extraction on Limulus polyphemus 459 10% during the first year after bleeding, and 11% during the second year. Thompson (1998) estimated that mortal- ity associated with LAL processing was 15% during the first week following blood extraction by observing bled and unbled horseshoe crabs in tanks in South Carolina. Each LAL producer has a unique bleeding method, meth- od of capture, distance and method of travel to the bleed- ing laboratory, a different holding time and conditions, and method of return of the bled crab that is most appropriate to that company's setting and situation. The results found in this study reflect those of BioWhittaker and may not be reflective of other companies' procedures. We examined the survival of the horseshoe crabs in a controlled environment (tank), as opposed to their natural environment. Our survival rate for horseshoe crabs may not reflect the survival rate of horseshoe crabs returned to the wild. Transfer and holding processes induce stress on the horseshoe crabs. Thus, the survival of the bled horseshoe crabs could be compromised by translocation and confinement in tanks. However, the tank environment may provide protection for horseshoe crabs when they are in a weakened state and are more susceptible to predation following blood-extraction. Further, this study looked only at newly matured male horseshoe crabs in an attempt to minimize variation of ex- ternal influences, so that the only difference between the two groups was whether or not they underwent the blood extraction process. Additional studies should examine dif- ferences in mortality in other age and sex classes. The Food and Drug Administration estimates that 260,000 horseshoe crabs were caught, bled, and returned by biomedical companies when last reported in 1997 (HCTC^). Assuming the 7.5% mortality rate found in our study is applicable to each biomedical company, and assuming that the number harvested for the biomedical companies has stayed relatively constant, we estimate that approximately 18,750 horseshoe crabs die yearly as a result of the biomedical procedure. In comparison, the commercial fishery reported landings of 5,543,000 pounds in 1999 and 3,756,000 pounds reported in 2000, all with a 100% mortality rate (NMFS, 2002). In the overall picture of the magnitude of horseshoe crabs caught and the associ- ated mortality rates, it is evident that the bleeding process has a substantially smaller impact than the commercial fishery on the horseshoe crab population. However, infor- mation on both biomedical and commercial fishery-induced mortality are necessary to determine the total harvest mor- tality of horseshoe crabs. The information presented in this study provides an estimate of the postbleeding mortality rate, an element ^ HCTC (Horseshoe Crab Technical Committee). 1998. Status of the horseshoe crab (Limulus polyphemus) population of the Atlantic coast, 9 p. + figures and tables. Horseshoe Crab Tech- nical Committee, Atlantic States Marine Fisheries Commission. Washington, D.C. of human-induced mortality on horseshoe crabs. This is one critical piece of information required to conduct a stock assessment and to develop an effective management strategy. Acknowledgments The authors wish to thank Carl N. Shuster Jr. and William McCormick for their helpful advice on the design of this study. Michael Schwarz, Ryan Cool, and Michael Jahnke of the Virginia Seafood Agricultural Research and Exten- sion Center, a unit of Virginia Tech, provided the facilities for holding the horseshoe crabs and maintained them. Funding for this study was provided by BioWhittaker, a CAMBREX company. We especially thank Tammy Newcomb and Michael Vaughan for their helpful advice throughout all stages of this study. Literature cited Berkson, J., and C. N. Shuster Jr 1999. The horseshoe crab: the battle over a true multiple use resource. Fisheries 24:6-10. Brown, G. G., and D. L. Clapper. 1981. Procedures for maintaining adults, collecting gametes, and culturing embryos and juveniles of the horseshoe crab, Limulus polyphemus. In Laboratory animal management: marine invertebrates, p. 268-290. National Academy Press, Washington, D.C. Clark, K. 1996. Horseshoecrabsandtheshorebirdconnection. //! Pro- ceedings of the horseshoe crab forum: status of the resource. (J. Farrell and C. Martin, eds.), p. 23-25. Univ. Delaware Sea Grant College Program, Lewes, DE. Mehta, C, and N. Patel. 1999. StatXact 4 Windows. CYTEL Software Corporation, Cambridge, MA. Mikkelsen, T. 1988. The secret in the blue blood, 125 p. Science Press, Beijing, China. NMFS (National Marine Fisheries Service). 2002. Annual commercial landing statistics. Available on- line at www.nmfs.gov. Accessed 03/01/02. Novitsky, T. J. 1984. Discovery to commercialization: the blood of the horse- shoe crab. Oceanus 27:13-18. Rudloe, A. 1983. The effect of heavy bleeding on mortality of the horse- shoe crab, Limulus polyphemus, in the natural environ- ment. J. Invert. Path. 42:167-176. Thompson. M. 1998. Assessments of the population biology and critical habitat for the horseshoe crab, Limulus polyphemus. in the South Atlantic Bight. M.Sc. thesis, 50 p. Univ. Charles- ton, Charleston, SC. Walls, E. A., J. Berkson, and S. A. Smith. 2002. The horseshoe crab, Limulus polyphemus: 200 million years of existence, 100 years of study. Rev. Fish. Sci. 10(1): 39-73. 460 Erratum Fishery Bulletin 98(1):127-138 (2000). 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 On page 128 (right column, second paragraph) the authors wrote ■". . . state agencies in Alabama. Florida, South Carolina, and Texas studied the feasibility of stock enhancement as a means of supplementing wild populations." Later in the same paragraph, they also stated that ". . . because broodstock for large-scale enhancement programs along the Atlantic seaboard have been obtained from Mosquito Lagoon and nearby esturaries. there is a potential for artificial genetic exchange between putatively separate gene pools (e.g. those of Mosquito Lagoon and the Carolinas)." The authors would like to clarify that the South Carolina Department of Natural Resources u.ses only locally obtained red drum for broodstock in its stocking programs. It was culturisls for private facilities in South Carolina who used broodstock from Florida Mosquito Lagoon to produce red drum for worldwide distribution. The risk posited in our Introduction, as it related to the Carolinas, pertained to escapement or mishandling of the imported broodstock and their progeny. Fishery Bulletin 101(1) 461 Superintendent of Documents Publications Order Form *5178 I I YES, please send me the following publications: Subscriptions to Fishery Bulletin for $55.00 per year ($68.75 foreign) The total cost of my order is $ . . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Please Choose Method of Payment: I I Check Payable to the Superintendent of Documents □ GPO Deposit Account | | | | | I I |— H I I VISA or MasterCard Account (Credit card expiration date) To fax your orders (202) 512-2250 (Authorizing Signature) Mail To: Superintendent of Documents PO. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! ^50 Fishery Bulletin Guidelines for contributors Content of papers Articles Articles are reports of 10 to 30 pages (double spaced) that describe original research in one or a combination of the following fields of marine science: taxonomy, biology, genetics, mathematics (including modeling), statistics, engineering, eco- nomics, and ecologj'. Notes Notes are reports of 5 to 10 pages without an abstract that describe methods and results not supported by a large body of data. Although all contributions are subject to peer review, responsi- bility for the contents of articles and notes rests upon the authors and not upon the editor or the publisher It is therefore important that authors consider the contents of their manuscripts care- fully. Submission of an article is un-derstood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts must be written in English. Authors whose native language is not EngHsh are strongly advised to have their manuscripts checked for fluency by English-speaking colleagues prior to submission. Preparation of papers Text Title page should include authors' full names and mailing addresses (street address required) and the senior author's telephone, fax number, e-mail address, as well as a list of key words to describe the contents of the manuscript. Abstract must be less than one typed page (double spaced) and must not contain any citations. It should state the main scope of the research but emphasize the author's con- clusions and relevant findings. Because abstracts are circulated by abstracting agencies, it is impor- tant that they represent the research clearly and concisely. General text must be typed in double- spaced format. A brief introduction should state the broad significance of the paper; the remainder of the paper should be divided into the following sec- tions: Materials and methods, Results, Discussion (or Conclusions), and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdi- vision (i.e. there can be no subdivision without at least two subheadings). The entire text should be intelligible to interdisciplinar\' readers; therefore, all acronyms and abbreviations should be written out and all lesser-known technical terms should be deflned 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 Sci- entific style and format: CBE manual for authors, editors, and publishers (6th ed.) for editorial style and the most current issue o( the American Fish- eries Society's common and scientific names of fishes from the United States and Canada for fish nomenclature. Dates should be written as fol- lows: 11 November 1991. Measurements should be expressed in metric units, e.g. metric tons (t). The numeral one (1) should be typed as a one, not as a lower-case el (1). Footnotes Use footnotes to add editorial comments regarding claims made in the text and to document unpub- lished works or works with local circulation. Foot- notes should be numbered with Arabic numerals and inserted in 10-point font at the bottom of the first page on which they are cited. Footnotes should be formatted in the same manner as citations. If a manuscript is unpublished, in the process of review, or if the information provided in the footnote has been conveyed verbally, please state this information as "unpubl. data," "manuscript in review," and "personal commun.," respectively. Authors are advised wherever possible to avoid ref- erences to nonstandard literature (unpublished lit- erature that is difficult to obtain, such as internal reports, processed reports, administrative reports, ICES council minutes, IWC minutes or working papers, any "research" or "working" documents, laboratory reports, contract reports, and manu- scripts in review). If these references are used, please indicate whether they are available from NTIS (National Technical Information Service) or from some other public depository. Footnote format: author (last name, followed by first-name initials); year; title of report or manuscript; type of report and its administrative or serial number; name and address of agency or institution where the report is filed. Literature cited The literature cited section comprises works that have been published and those accepted for pub- lication (works in press) in peer-reviewed jour- nals and books. Follow the name and year system for citation format. In the text, write "Smith and Jones (1977) reported" but if the citation takes the form of parenthetical matter, write "(Smith and Jones. 1977)." In the literature cited section, list citations alphabetically by last name of senior author: For example. Alston, 1952; Mannly. 1988; Smith, 1932; Smith, 1947; Stalinsky and Jones. 1985- Abbreviations of journals should conform to the abbreviations given in the Serial sources for the BIOSIS previews database. Authors are responsible for the accuracy and completeness of all citations. Literature citation format: author (last name, followed by first-name initials); year; title of report or article; abbreviated title of the journal in which the article was published, volume number, page numbers. For books, please provide publisher, city, and state. Tables Tables should not be excessive in size and must be cited in numerical order in the text. Headings in tables should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted (use italic arable numerals for footnote markers). Use asterisks only to indicate probability in statistical data. Place table legends on the same page as the table data. We accept tables saved in most spread- sheet software programs (e.g. Microsoft Excel). Please note the following: • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Use zeros before all decimal points for values less than one (e.g. 0.31). Figures Figures include line illustrations, computer-gener- ated line graphs, and photographs (or slides). They must be cited in numerical order in the text. Line illustrations are best submitted as original draw- ings. Computer-generated line graphs should be printed on laser-quality paper Photographs should be submitted on glossy paper with good contrast. All figures are to be labeled with senior author's name and the number of the figure (e.g. Smith, Fig. 4). Use Helvetica or Arial font to label ana- tomical parts (line drawings) or variables (graphs) within figures; use Times Roman bold font to label the different sections of a figure (e.g. A, B, C). Figure legends should explain all symbols and abbreviations seen within the figure and should be typed in double-spaced format on a separate page at the end of the manuscript. We advise authors to peruse a recent issue ot Fishery Bulletin for stan- dard formats. Please note the following: • Capitalize the first letter of the first word of axis labels. • Do not use overly large font sizes to label axes or parts within figures. • Do not use boldface fonts within figures. • Do not create outline rules around graphs. • Do not use horizontal lines through graphs. • Do not use large font sizes to label degrees of longitude and latitude on maps. • Indicate direction of degrees longitude and latitude on maps (e.g. 170°E). •Avoid placing labels on a vertical plane (except ony axis). •Avoid odd (nonstandard) patterns to mark sections of bar graphs and pie charts. Copyright law Fishery Bulletin, a U.S. government publication, is not subject to copyright law. If an author wishes to reproduce any part oi Fishery Bulletin in his or her work, he or she is obliged, however, to acknowledge the source of the extracted literature. Submission of papers Send four printed copies (one original plus three copies) — clipped, no/ stapled — to the Scientific Edi- tor, at the address shown below. Send photocopies of figures with initial submission of manuscript. Original figures will be requested later when the manuscript has been accepted for publication. Do not send your manuscript on diskette until requested to do so. Dr Norman Bartoo National Marine Fisheries Service, NOAA 8604 La Jolia Shores Drive LaJolla, CA 92037 Once the manuscript has been accepted for pub- lication, you will be asked to submit a software copy of your manuscript. The software copy should be submitted in WordPerfect or Word format (in Word, save as Rich Text Format). Please note that we do not accept ASCII text files. Reprints Copies of published articles and notes are avail- able 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. I mbl n •> U.S. Department of Commerce Volume 101 Number 3 July 2003 T •> Fishery Bulletin U.S. Department of Commerce Donald L Evans Seaetary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T Hogarth Assistant Administrator for Fisheries The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- cries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. 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: $45.00 domestic and $56.25 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Dr. Norman Bartoo Editorial Assistant Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Dr Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr Theodore W. Pietsch Dr Joseph E. Pow/ers Dr Harald Rosenthal Dr Fredric M. Serchuk Dr George Watters University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Mahne Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service 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 101 Number 3 July 2003 Fishery Bulletin Contents The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service (NOAA) or any other agency or institution. The National Marine Fisheries Service (NMFS) 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 pubUcation. Articles 463-475 Brule, Thierry, Ximena Renan, Teresa Colas-Marrufo, Yazmin Hauyon, Armin N. Tuz-Sulub, and Christian Deniel Reproduction in the protogynous black grouper (Mycteroperca bonaci (Poey)) from the southern Gulf of Mexico 476—483 Comeau, Michel, and Manon Mallet The effect of timing of tagging on streamer-tag recapture rates for American lobster (Homarus americanus) 484—500 Diamond, Sandra L. Estimation of bycatch in shrimp trawl fisheries: a comparison of estimation methods using field data and simulated data 501 -513 Hanseiman, Dana H., Terrance J. Quinn II, Chris Lunsford, Jonathan Heifetz, and David Clausen Applications in adaptive cluster sampling of Gulf of Alaska rockfish Companion articles 514-534 Itoh, Tomoyuki, Sachiko Tsuji, and Akira Nitta Migration pattems of young Pacific bluefin tuna (Thunnus orientalis) determined with archival tags 535-544 Itoh, Tomoyuki, Sachiko Tsuji, and Akira Nitta Swimming depth, ambient water temperature preference, and feeding frequency of young Pacific bluefin tuna {Thunnus orientalis) determined with archival tags 545-565 Jagielo, Thomas, Annette Hoffmann, Jack Tagart, and Mark Zimmermann Demersal groundfish densities in trawlable and untrawlable habitats off Washington: implications for the estimation of habitat bias in trawl surveys Wootfi Holt, CceaoograpriiC '^ Libra'» AUG 5 2003 Fishery Bulletin 101(2) 566-582 Loughlin, Thomas R., Jeremy T. Sterling, Richard L. Merrick, John L. Sease, and Anne E. York Diving behavior of immature Steller sea lions (Eumetopias jubatus) 583-589 McBride, Richard S., Justin R. Styer, and Rob Hudson Spawning cycles and habitats for ballyhoo (Hemiramphus brasiliensis) and balao (H. balao) in south Florida 590—602 Morato, Telmo, Encarnacion Sola, Maria P. Gros, and Gui Menezes Diets of thomback ray (Raia clavata) and tope shark (Galeorhinus galeus) In the bottom longline fishery of the Azores, northeastern Atlantic 603-613 Mullin, Keith, D., and Gregory L. Fulling Abundance of cetatceans in the southern U.S. North Atlantic Ocean during summer 1998 614—626 Rogers-Bennett, Laura, Donald W. Rogers, William A. Bennett, and Thomas A. Ebert Modeling red sea urchin (Strongylocentrotus franciscanus) growth using six growth functions 627—639 Skomal, Gregory B., and Lisa J. Natanson Age and growth of the blue shark (Prionace glauca) in the North Atlantic Ocean 640—652 Teel, David J., Donald M. Van Doornik, David R. Kuligowski, and W. Stewart Grant Genetic analysis of juvenile coho salmon (.Oncorhynchus kisutch) off Oregon and Washington reveals few Columbia River wild fish 653—672 Terceiro, Mark The statistical properties of recreational catch rate data for some fish stocks off the northeast U.S. coast 673—683 Williams, Ashley J., Campbell R. Davies, Bruce D. Mapstone, and Garry R. Russ Scales of spatial variation in demography of a large coral-reef fish— an exception to the typical model? Notes 684—692 Klimley, A. Peter, Salvador J. Jorgensen, Arturo Muhlia-Melo, and Sallie C. Beavers The occurrence of yellowfin tuna (Thunnus albacares) at Espiritu Santo Seamount in the Gulf of Mexico 693—697 Landaeta, Mauricio R, Francisco J. Neira, and Leonardo R. Castro Larvae of Dactylopsaron dimorphicum (Perciformes: Percophidae) from oceanic islands in the southeast Pacific 698-703 Pooler, Penelope S., David R. Smith, Robert E. Loveland, Mark L. Botton, and Stewart F. Michels Assessment of sampling methods to estimate horseshoe crab (Limulus polyphemus L.) egg density in Delaware Bay 704-711 Powell, Allyn B. Larval abundance, distribution, and spawning habits of spotted seatrout (Cynoscion nebulosus) in Florida Bay, Everglades National Park, Florida 712-718 Zimmerman, Christian E., and Roger L. Nielsen Effect of analytical conditions in wavelength dispersive electron microprobe analysis on the measurement of strontium-to-calcium (Sr/Ca) ratios in otoliths of anadromous salmonids 719 Subscription form 463 Abstract— An analysis was made of sexual pattern, spawning season, sizes at sexual maturation, and sex change in black grouper {Mycteroperca bonaci ) from the southern Gulf of Mexico. Samples were taken between 1996 and 2000, from industrial and small-craft commercial fisheries, in offshore and inshore waters of the continental shelf of the Yucatan Peninsula iCampeche Bank), including the shallow waters of National Marine Park Alacranes Reef For all collected specimens (/! = 1229), sex and maturation condition were determined by histological analysis of the gonads. The offshore sample con- sisted of 75.1% females, 24.3% males, and 0.6% transitional-stage fish. All individuals collected from inshore waters were females. Gonadal structure and population structure characteris- tics for Campeche Bank black grouper were consistent with the characteristics of monandric protogynous hermaphro- dism for a serranid fish. Sexually active males and females were observed year- round, although ripening females, with stage-Ill, -IV, and -V vitellogenic oocytes in the ovaries, dominated in samples taken between December and March. In addition, peak occurrence of ripe- running females with hyaline oocytes or postovulatory follicles (or both) in the ovaries was recorded in January and February. A few precocious females began spawning in October and Novem- ber, and others were still in spawning condition in May and June. Fifty per- cent maturity of females was attained at 72.1 cm fork length (FL). Median size at sexual inversion was 103.3 cm FL, and 50% of the females measuring 111.4 cm FL had transformed into males. The southern Gulf of Mexico grouper fishery was considered deteriorated and lacked a well-defined management strategy. Results of the present study provide helpful information on black grouper reproduction in this area and could help Mexican authorities choose appropriate management strategies for this fishery, such as minimum size limit, closed fish- ing season, and protection of spawning aggregations. Reproduction in the protogynous black grouper (Mycteroperca bonaci (Poey)) from the southern Gulf of Mexico Thierry Brule Ximena Renan Teresa Colas-Marrufo Yazmin Hauyon Armin N.Tuz-Sulub Centra de Investigacion y de Estudios Avanzados del IPN Unidad Merida Antigua Carretera a Progreso km. 6 Apartado postal 73 Cordemex Codigo Postal 97310 Merida Yucatan, Mexico E-mail address (lor T. Brule), ibrulecfimda.cinveslav.mx Christian Deniel Institut Universitaire Europeen de la Mer, Ressources Halieutiques-Poissons Manns Unlversite de Bretagne Occidentale Place Nicolas Copernic Technopole Brest Iroise 29820 Plouzane, France Manuscript approved for publication 11 February 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:463^75 (2003). The black grouper {Mycteroperca bon- aci) is one of the 20 most commonly sought serranid fishes in the tropical western Atlantic region (Sadovy, 1994). The species ranges from Massachusetts and Bermuda to southeastern Brazil (Bohlke and Chaplin, 1993; Fischer, 1978; Bullock and Smith, 1991; Begossi and Figueiredo, 1995). It is found on irregular bottoms such as coral reefs, drop-off walls, and rocky ledges, in depths from 10 to at least 100 m (Roe, 1977; Manooch and Mason, 1987; Bull- ock and Smith, 1991; Heemstra and Randall, 1993; Huntsman et al., 1994). According to Shapiro (1987), the sa- lient feature of grouper reproduction is protogynous hermaphroditism. The first reasonable evidence of protogyny in M. bonaci was published by Smith (1959), although there have been other occasional reports on black grouper re- production (Erdman, 1956; Smith, 1961, 1971, 1972; Naranjo in Garcia-Cagide et al., 1994). Systematic study of sexual pattern and sexual maturation in the species has only been carried out by Garcfa-Cagide and Garcia (1996) in Cuban waters and by Crabtree and Bullock (1998) in Florida waters. This grouper has been reported to form spawning aggregations in the Gulf of Mexico and Caribbean Sea (Fine, 1990; Carter and Perrine, 1994; Domeier and Colin, 1997; Eklund et al., 2000). Black grouper is an important com- mercial and recreational fin fish re- source in Bermuda, southern Florida, Cuba, the southern Gulf of Mexico, and Venezuela (Manooch and Mason, 1987; Cervigon, 1991; Heemstra and Randall, 1993; Claro et al., 1994). In the southern Gulf of Mexico between 1989 and 1999, groupers accounted for 18-307< of the total offshore commercial marine re- sources harvested from the Campeche Bank (the continental shelf surround- ing the northern coast of the Yucatan Peninsula) and resources landed in inshore waters off the state of Yucatan (SEMARNAP, 2000a). At least 18 grou- per species are commercially exploited in this region — the most important of these by catch number and weight are red grouper {Epinephelus morio), fol- lowed by black grouper and gag {Myc- teroperca microlepis) (Colas-Marrufo et al., 1998). Because grouper landings in 464 Fishery Bulletin 101(3) Figure 1 Map of the Campeche Bank, Mexico, showing the geographic distribution of sampling locations (•) for black grouper (Mycteroperca bonaci) observed during the period 1996-2000. Samping locations marked (O) are where ripe-running female black grouper were caught. Sample locations: 1 = April 1996; 2 = April and May 1996; 3,4 = May 1996; 5 = November 1996; 6, 7 = December 1996; 8 = Januai-y 1997; 9 = January and Febru- ary 1997; 11 = Febi-uary 1997; 12 = May 1997; 13 = June 1997; 14 = June and July 1997; 15 = July 1997; 16 = July and August 1997; 17, 18 = August 1997; 19 = September 1997; 20 = September and October 1997; 21 = October and November 1997; 22 = November 1997; 23 = December 1997; 24,25 = January 1998; 26 = January and February 1998; 27 = February and March 1998; 29 = March 1998; 30,31 = June 1998; 32 = July 1998; 33 = August 1998; 34 = August and September 1998; 35 = September 1998; 36 = April 1999; 37,38 = May 1999; Alacranes Reef was sampled November and December 1999, January, February, and August to November 2000. No black giouper were caught in sample locations 10 (22''33'-88°24'W; February 1997) and 28 (22°30'N-89°30W; March 1998). the Campeche Bank decreased between 1991 and 1997, the Mexican government proposed management measures to protect the grouper resource, but without considering the biological characteristics and fishery aspects of each exploited species (SEMARNAP, 2000a, 2000b). Given that sustainable resource management is founded on stock as- sessments and knowledge of the biology of exploited species (Sadovy, 1997 ), more information on the biology of the most abundant groupers from the southern Gulf of Mexico in general, including Campeche Bank, is necessary to imple- ment and refine management strategies. This lack of knowledge is especially acute for Campeche Bank black grouper. For example, although growth, feed- ing, and reproduction of the Yucatan red grouper are well documented, none of this information is available for the black grouper in this region (Brule and Deniel, 1994; Brule et al., 1994, 1999). This lack of information is alarming be- cause M. bonaci can account for 40% of the grouper catch by weight for some commercial vessels, and if this species is not included in stock monitoring and reproduction stud- ies, effective overall management of the southern Gulf of Mexico grouper fishery could be seriously undermined (Colas-Marrufo et al., 1998). With the final aim of defining more accurate and effi- cient management practices for the Campeche Bank gi-ou- per fishery, we present analyses of sexual status, sexual cycle, spawning season, size at sexual maturation, and sex change for black gi-ouper from the southern Gulf of Mexico. Materials and methods Black grouper were collected from commercial catches taken from rocky bottoms in both offshore and inshore waters of the Campeche Bank and in the shallow waters of the Alacranes Reef complex. Alacranes Reef is the most important complex of coral reefs located on the Yucatan continental shelf Because of its high scientific and eco- nomic potential, the Mexican government declared this reef a National Marine Park in June 1994 (Fig. 1). In offshore waters, black grouper (/i =880) were caught by the long-line industrial fleet from 38 locations mainly situated in the northeastern part of the Campeche Bank, at depths rang- ing from 40 to 210 m, between April 1996 and May 1999. In inshore waters, some specimens (n=39) were obtained Brule et a\ : Reproduction in Mycteroperca bonaa 465 from the small craft-fleet, whose crew captured them using spear guns at depths ranging from 4 to 20 m, in areas close to the port of Progreso, between November 1998 and May 1999. On Alacranes Reef, black groupers («=206) were cap- tured with spear guns by small-craft fishermen at depths of 10-12 m between November 1999 and February 2000. Fork and standard lengths (FL, SL), whole and gutted weights (WW, GW), and weight of gonads (gW), were re- corded for all collected fish. All lengths reported in the pres- ent study are fork length and all weights are gutted weight. In discussion, total length data for Cuba (Garcia-Cagide and Garcia, 1996) and Florida (Crabtree and Bullock, 1998) populations were converted to fork lengths by using the fork-length to total-length relationship calculated by Crabtree and Bullock ( 1998). Criteria presented by Sadovy and Shapiro (1987) were used to diagnose sequential hermaphroditism in black grou- per of Campeche Bank. Sex by size-frequency distributions for black groupers were compared by using the Kolmogorov- Smirnov nonparametric test, and differences between male and female mean fork lengths were analyzed by using a one-tailed 2-test (if n>30) or a one— tailed <-test (if n<30). The male-to-female ratio (M:F), excluding transitional-stage fish (referred to as "transitional fish" in this article), was calculated and Pearson chi-square or Yates's corrected chi- square goodness-of-fit tests were carried out to determine if sex ratio differed significantly from unity (Scherrer, 1984). Significance level, «. was 0.05 in all instances. Sex-dependent change in fin pigmentation, as described by Crabtree and Bullock (1998), was examined in a sub- set of fish (« = 104) caught from Alacranes Reef between August and November 2000. The colors of the pectoral, anal, dorsal, and caudal fins were recorded and histologi- cal sections of the gonads were prepared and assessed for sex identification. For all the fish sampled in all locations, sex and sexual development were determined by examination of the micro- scopic structure of the gonads. These were preserved in Bouin's fluid, embedded in Paraplast and sectioned to 6 ftm thickness. Ovaries and testes sections were stained in Gabe and Martoja's triple stain for light microscopy (Gabe, 1968). Fish were identified as female, male, or as transitional. Based on red grouper microscopic features for oogenesis (Brule et al., 1999) and for spermatogenesis (Moe, 1969), six descriptive stages were recognized in black grouper ovaries and five in the testes. Histological sections of the ovaries were also scanned for the presence of postovulatory follicles and atretic oocjrtes in alpha or beta stages (Lam- bert, 1970). Using the criteria defined by Smith ( 1959) and Sadovy and Shapiro ( 1987 1, we considered individuals with gonads containing primarily ovarian tissue, degenerating or not, with few clusters of spermatocytes, spermatids, or spermatozoa to be undergoing sexual inversion. According to the sexual classes defined by Brule et al. ( 1999) for red grouper, female and male black grouper were classified as resting, ripening, ripe-running, or spent, and fish in the process of sexual inversion were classified as transitional. Using the histological features considered by Shapiro et al. (1993) as sign of prior spawning activity for the red hind (Epinephelus guttatus) we were able to distinguish resting mature females from immature (virgin) females that had never spawned. Reproduction periodicity was evaluated for both sexes by examining seasonal variations in the gonadosomatic index (GSI=100 xgW/GW) and in the relative proportion of individuals in each sexual class. Specimens from offshore waters taken during different years were pooled by month, and mean GSI values and percent frequencies of sexual classes were generated monthly for a single year Immature individuals were discarded from this analysis. Size at which SO'X of females were sexually mature (L,;q) was determined by using a binary logistic regression (SYS- TAT statistical computer package for Windows, version 8.0, SPSS Inc., Chicago, ID. For our analysis, resting mature, ripening, ripe-running, and spent females were considered as sexually mature individuals. Moreover, the minimum size at which females become sexually mature (L^,;„) was recorded, and the percentage of females of maximum length at first maturity, ^^iJ^^nax with L^^^ ~ maximum length of females recorded in samples, was determined (Grimes, 1987). Sexual transition was analyzed by using a binary logistic regression to estimate the length at which 50% of the females transformed to males (P^q) according to Crabtree and Bullock (1998). The size range and median size at which sex inversion occurs were estimated by fol- lowing the procedures of Shapiro ( 1984). Furthermore, the variation in size at sex change was analyzed by using two ratios defined by Shapiro (1987): ratio 1, size range of tran- sitional fish divided by maximum size of fish in samples; and ratio 2, range of overlap in size of males and females divided by maximum size of fish in samples. Results Size-frequency distributions All individuals collected from inshore waters were females ranging from 25.6 to 58.0 cm in length (Fig. 2). The off- shore fish sample, which did not include the Alacranes Reef sample, was composed of 75.1% females, 24.3% males, and 0.6% transitional fish. Females ranged in size from 57.0 to 123.5 cm, males from 86.0 to 132.0 cm, and transitional fish from 99.0 to 121.5 cm. The Alacranes Reef sample was com- posed of 95% females ranging in size from 46.0 to 100.0 cm and 5% males from 97.0 to 135.0 cm. In the offshore sample, the male size range differed significantly from that of the females (Kolmogorov-Smirnov; 7!=875; P<0.05), and male mean fork length (114.6 ±7.1 cm; mean ±SD) was greater than female mean fork length (96.6 ±12.1 cm; one-tailed 2-test, n=875; P<0.05). Similar results for male and female size ranges (Kolmogorov-Smirnov; n=206;P<0.05) and male (115.7 ±12.9 cm) and female (67.6 ±11.2 cm) mean fork lengths (one-tailed t-test, «=206; P<0.05) were observed for black grouper from the Alacranes Reef sample. Sex ratio The male-to-female ratios were calculated for each 5-cm size class from 25.1 to 135.0 cm length (Table 1). The 466 Fishery Bulletin 101(3) 140 120 100 • 80 60 ■ 40 20 E D Female (inshore waters) D Female (offshore waters) ^tt^aie ■ Transitional -P-P- P r fn ,11 ,l,i ,1,1 ,H ,11 JL.ik.lb^ 20 30 40 50 60 70 80 90 100 110 120 130 B 40 1 n Female 30 • 0Male 20- 10- n n • n ,n,R], , ,n,n, , JL,^ 20 50 60 70 80 90 Fork length (cm) 100 110 120 130 Figure 2 Size-frequency distribution for females, transitional individuals, and males for black grouper {Mycteroperca bonaci) caught between April 1996 and February 2000 in offshore and inshore waters of the Campeche Bank (A), and in shallow waters of Alacranes Reef (B), Mexico. sex ratios were female-biased in size classes less than 110.1 cm, did not differ significantly from a 1;1 sex ratio in the 110.1-115.0 cm size class, and were male-biased in size classes larger than 115.0 cm. The overall black grouper sex ratio was 1:4, which differed significantly from unity (Xf-=400.8,P<0.05). Fin pigmentation Of the 104 black grouper analyzed to detect gender- associated color changes, 98 were females (size range 47.0-99.0 cm), five were males (99.0-115.0 cm), and one, which presented previtellogenic oocytes and nests of sper- matocytes and spermatozoa in its gonads, was classified as transitional (99 cm). All males, as well as the transitional specimen, displayed the male color phase with jet black pigmentation on pectoral, anal, and caudal fins. Only 59f of the females (size range 50.0-100.0 cm, «=5) had jet black pigments on their fins. Gonadal structure All ovaries presented a central cavity with a germinal epi- thelium forming the surface layer of a series of projecting ovigerous folds or lamellae of the tunica albuginea. Of the 225 males assessed histologically, 769f (;! = 170) presented a membrane-lined central cavity in the testes. This lumen remained unused in the transport of spermato- zoa, and sperm ducts or sinuses within the gonadal capsule were observed in ZT7< of the specimens (/;=84) (Fig. 3, A and B). Previtellogenic oocytes (stages I and II ) remained in the testes of 13''i of the males (?)=30), and only one of these (107.0 cm) presented previtellogenic oocytes in degenera- tion within lamellae in a fully developed testis dominated by crypts of spermatocytes, spermatids, and spermatozoa. Yellow bodies w(>re oi)sorved in the testes of 96'r of the males. Internal gonadal structure for the five black gi-oupers classified as transitional was very similar to that of im- Brule et al : Reproduction in Mycteroperca bonaci 467 Table 1 Number of sampled fish; proportion of females, males, and transitional-stage fish (transitional fish); and sex ratio by length class for black gi-ouper {Mycteroperca bonaci) collected in the inshore and offshore waters of the Campeche Bank and shallow waters of Alacranes Reef, Mexico, between April 1996 and February 2000. Alacranes collection = collection at the Alacranes Reef Fork length class (cm) Females Total collected Transitional fish Males Sex ratio (male: female) Offshore collection Inshore collection Offshore collection Alacranes collection Offshore collection Alacranes collection n (%) for total collected n (%) n (%) 25.1-30.0 2 0 0 2 100.0 0 0 0 0 30.1-35.0 4 0 0 4 100.0 0 0 0 0 35.1-40.0 7 0 0 7 100.0 0 0 0 0 40.1-45.0 19 0 0 19 100.0 0 0 0 0 45.1-50.0 4 0 5 9 100.0 0 0 0 0 50.1-55.0 1 0 22 23 100.0 0 0 0 0 55.1-60.0 2 2 34 38 100.0 0 0 0 0 60.1-65.0 0 9 34 43 100.0 0 0 0 0 65.1-70.0 0 10 27 37 100.0 0 0 0 0 70.1-75.0 0 22 28 50 100.0 0 0 0 0 75.1-80.0 0 32 21 53 100.0 0 0 0 0 80.1-85.0 0 33 8 41 100.0 0 0 0 0 85.1-90.0 0 62 9 71 98.6 0 1 0 1 1.4 1:71 90.1-95.0 0 101 3 104 99.0 0 1 0 1 1.0 1:104 95.1-100.0 0 108 4 112 93.3 2 1.7 3 3 6 5.0 1:18.67 100.1-105.0 0 130 0 130 91.5 1 0.7 11 0 11 7.7 1:11.82 105.1-110.0 0 81 0 81 66.9 0 40 0 40 33.1 1:2.03 110.1-115.0 0 49 0 49 40.8 1 0.8 69 1 70 58.3 1:0.70* 115.1-120.0 0 15 0 15 22.7 0 47 4 51 77.3 1:0.29 120.1-125.0 0 7 0 7 19.4 1 2.8 28 0 28 77.8 1:0.25 125.1-130.0 0 0 0 0 0 12 2 14 100.0 130.1-135.0 0 0 0 0 0 2 1 3 100.0 Total 39 661 195 895 79.6 5 0.4 214 11 225 20.0 1:3.98 * Value did not differs signifi cantly from 1:1 sex ratio (Xi^; P>0.05). mature or resting females. Stage! anii -II oocytes, yellow bodies, and sometimes bundles of muscle and connective tissue were present within the lamellae. Intermixed with the female tissue, these gonads contained a few nests of spermatogonia, spermatocytes, or spermatozoa, although degeneration of female germinal tissue was not observed (Fig. 3C). These transitional specimens were captured in September, November, and December 1997 and in January and March 1998. Sexual cycle Females captured from inshore waters during January, February, March, May, November, and December were immature and had low individual GSI values (GSI range 0.01-0.18%), and only oogonia and previtellogenic oocytes were observed in their ovaries. Mean GSI for mature females caught in offshore waters began to increase in December (0.5%), reached a maximum value in February (2.2%), and declined to a near minimum level in March (0.7%) and April (0.6%) (Fig. 4). Highest in- dividual GSI values for females were observed in October (4.9%), December (6.0%), January (6.7%), and February (9.6%). Mean GSI for males caught in offshore waters in- creased in December (0.13% ) and January (0.14% ) — reach- ing a maximum value in February (0.22%) and declining from March (0.12%) to August (0.11%) (Fig. 4). Highest in- dividual GSI values for males were observed in September (0.43%) and February (0.39%). Ripening females, with stage-Ill, -IV, and -V vitellogenic oocytes in their ovaries, were observed year-round, but dominated in collections made between December and March (42-56% of females) (Fig. 5). Advanced vitellogenic oocytes undergoing final oocyte maturation were noted on- ly for some females captured between January and March (Fig. 6A). Ripe-running females, with hyaline oocytes or postovulatory follicles (or with both) in their ovaries, were recorded between October and June, and peaked in occur- 468 Fishery Bulletin 101(3) rence in January (28%) and February (52%)(Fig. 6B). The gonads of 50 ripe-running females caught between Janu- ary and April, and during June and November, contained both postovulatory follicles and stages III-V vitellogenic oocytes without sign of degeneration (Fig. 6C). Spent fe- Figure 3 Photomicrographs of histological sections from male and transitional black grouper iMycteroperca honati) gonads collected from Campeche Bank, Mexico. (A) Section from a 1 14-cm-FL ripening male captured in July 1997, showing lamellae, lumina, spermatogonia, and spermatocyte cysts, and lamellae sinuses full of spermatozoa. (B) Section from a r2'2-cm-FL ripe-running male captured in September 1997, with sperm sinus full of spermatozoa in gonadal capsule. (C) Section from a 113-cm-FL transitional fish captured in January 1998, showing previtellogenic oocytes (stages I) and scattered spermatocyte cysts. CX' = gonadal capsule; L = lumen; LA = lamellae; PO = previtellogenic oocyte; SPC = spermatocyte; SPG = spermatogonia; SPZ = spermatozoa; SS = sperm sinus. Scale bars = 200 microns. males, with atretic and remaining vitellogenic oocytes in their gonads, were caught between January and August {3-297f ). Resting mature females, with stages -I and -II oo- cytes, bundles of muscle, and yellow bodies in their ovaries were abundant in samples taken from May to November (54-98%). Ripening or ripe-running males were recorded year-round and spent males were observed in November (4%), from January to March (10-40%), and from May to July (8-22%). Various females from the Alacranes Reef were ripening in November (GSI range: 0.03-4.44%, n=l2), December (GSI range: 0.22-7.18%, n=9) and February (GSI range: 0.10-6.61%, n=25). In February, some of them were ripe- running, with postovulatory follicles in ovaries (GSI=1. 77% and 1.91%, n=2) and others were spent (GSI=0.74% and 0.88%, n=2). Alacranes Reef males were ripening or ripe- running in November, December, and February (GSI range: 0.03-0.44%, «=11). Location and timing of spawning Between April 1996 and February 2000, 61 ripe-running females were caught at 11 offshore fishing locations situ- ated in the northeastern part of the Campeche Bank (depth range: 51-68 m), and from shallow waters of the Alacranes Reef (8-10 m) (Fig. 1). All had vitellogenic oocytes (stages- III-V) with hyaline oocytes or postovulatory follicles (or with both) in their ovaries. Most of these females were caught during, or close to, the new moon phase (Table 2). Sizes at maturity and at sexual transition The smallest mature female (L^j|jj=58.0 cm) was caught in shallow waters of the Alacranes Reef and had stage-Ill oocytes in its ovaries. Fifty-percent maturity of females was attained at 72.1 cm in size and all females larger than 95.1 cm were mature (Fig. 7). Because the largest female observed in samples was 123.5 cm (L^,^^), the percentage of females at maximum length at first maturity was l^„^,J Females changed sex between 85.5 and 125.0 cm in length (the overlap zone between male and female sizes) and the median size of sexual inversion was 103.3 cm. By the time they attained a length of 111.4 cm, 50'7f of the females in the sample had transformed into males (Fig. 8). Size range of transitional fish (99.0-121.5 cm) was 17% of maximum fish size ( 135.0 cm) (ratio 1, see "Materials and methods"'section), and sex change occurred over 29'7f of the maximum size observed for the species (ratio 2). Immature males were not observed during the study. Discussion Sexual pattern Previous research strongly suggests that sex reversal occurs in M. bonaci (Smith, 1959, 1961; Garcia-Cagide and Garcia, 1996; Crabtree and Bullock, 1998). Observations on gonadal and population structure characteristics for black Brule et al. : Reproduction in Mycteroperca bonaci 469 10- Female • 8- • c o o • a. 6- • • • • • « 4- t ■D ■/•.I > T3 • IT) u^m^ Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Female I I I I I I I I I \ — I — r 0.5-1 0.4- 0.3- 0.2- 0.1 - 0.0 0.3- 0.2- 0.1 0.0- Male I I \ I 1 i I 1 — I — ( — I — I Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Male I I I — I — I — i — I — I — I — I — I — I — I c,e100 mm) differently (with the tag inserted in only one abdominal muscle) than did Moriyasu et al. (1995) for smaller lobsters. The differ- ence could also be explained by the artificial conditions of our aquarium experiment. Under natural conditions tagged lobsters could shed their tags through interspecific interac- tions (Rowe and Haedrich, 2001), intraspecific interactions, and by being dislodged by obstacles in their habitat (Ennis, 1986; Krouse and Nutting, 1990). Streamer tag loss related to inter- and intraspecific interactions and the habitat has already been reported for the brown shrimp (Howe and Hoyt, 1982; P. aztecus) and the tiger prawn (Hill and Was- senberg, 1985;P. esculentus). However, more research would be needed to identify the cause of tag shedding in nature and assess its variability in relation to different lobster habitat before the recapture rate could be adjusted based on inter- and intraspecific interactions and the habitat. The overall level of streamer tag loss compared to sphy- rion tag loss seems to be lower, but also depends upon the molt stage of the lobster at tagging and molting. In their study, Moriyasu et al. (1995), suggested that sphyrion tag loss mainly occurs within days after tagging or during molting and is related to the lobster molt stage at tagging. We observed lower levels of tag loss compared to those of Moriyasu et al. ( 1995), except for the tag shedding during molting for early premolt lobsters and tag-induced mor- tality for lobsters tagged in late premolt. They observed 3% and 11% of tag shedding without molting for lobsters tagged in early and late premolt stages, respectively, com- pared to none in our study. Furthermore, the most striking difference is the level of tag loss that reached 10% and 30% for lobsters tagged in intermolt and postmolt stages compared to 0% and <1%, respectively, in our study. The difference in tag loss for lobsters tagged in the postmolt stage could be explained by the physical nature of the tags themselves and the tagging techniques. Compared to the streamer tag that is threaded through two abdominal muscles, the sphyrion tag is anchored to only one muscle by means of a hypodermic needle (Moriyasu et al., 1995). Be- cause the muscles of postmolt lobsters (in the early stages) are not well formed, it is difficult to firmly embed an object, such as a tag, and the probability of tag loss for a tag em- bedded into only one thin muscle is greater than that for a tag treaded through two muscles. Hence, it seems that streamer tags are more effective in terms of tag retention compared to sphyrion tags for lobsters tagged in intermolt and postmolt stages, but equally so for lobsters tagged in the premolt stage. In field tagging studies, streamer tags yielded a good re- capture rate within the first year following tagging for lob- sters tagged immediately before or after the molting sea- son. The efficiency of streamer tags compared to sphyrion tags had already been established (.Moriyasu et al., 1995; Comeau et al., 1999). Moriyasu et al. (1995) reported that there was a significantly greater recapture rate for lob- sters tagged with streamer tags (44%) compared to those tagged with sphyrion tags (19%). Based on the results of Comeau et al. (1999), the recapture rate of lobsters tagged with sphyrion tags is 22% and 16% for lobsters tagged in premolt and postmolt stages, respectively, compared to 33% and 45% for lobsters tagged with streamer tags. These recapture rates corroborate aquarium observations by Moriyasu et al. (1995) on the tag retention of sphyrion tags and ours on the tag retention of streamer tags for lobster tagged at various molt stages. Knowledge of the level of tag loss is paramount for ad- justing the recovery rate to estimate population character- istics and fishery parameters for the American lobster. We observed that the recapture rate dropped significantly in the second and third years at large; this finding suggests a high level of tag loss. A similar multiple-years recapture rate pattern was observed for six other sites within the southwestern Gulf of St. Lawrence (Comeau and Savoie, 2002). Rowe and Haedrich (2001) indicated that the streamer tag shedding level for lobsters that molted almost a year later reached 40%. This high level of tag shedding, probably related to the streamer tag remaining firmly attached to the old dorsal thoraco-abdominal membrane during molting, might explain the drastic decrease of tag recaptures observed between the first and the second tag- recapture periods in our field study. We believe that the adjustment of the recapture rate due to tag loss should be limited to lobsters recaptured within the first year at large and prior to the molting season for lobsters tagged in intermolt and postmolt stages. The multiple-years recapture pattern of a high recapture rate within the first year at large followed by low recapture rates in subsequent years could have a significant impact on multiyear tagging models. These models that were pro- posed by Sober ( 1970) could be used to estimate population characteristics and fishery parameters if underlying as- sumptions are followed. Based on these multiyear models originally developed for birds (Seber, 1970; Brownie et al., 1985), a suite of models adapted for fishery data and ad- justments, mainly by reparameterization, were proposed to address underlying assumption violations (Pollock et al., 1991, 2001; Heam et al., 1998, 1999; Hoenig et al., 1998a, 1998b; Frusher and Hoenig, 2001; Latour et al., 2001a, 2001b). Some of these models were developed to take into account fishing effort, incomplete mixing, and tag recovery rate. The latter is a composite parameter involving tag retention and tag-induced mortality (tag loss), exploitation rate, and tag reporting rate. There is no argument that the participation of fishermen in returning tags (tag reporting rate) is very important; however, for crustacean fisheries, underlying assumptions dealing with tag loss (see assumptions 2 and 3 in Pollock, 1991, 2001) are equally important and have to be addressed. In general. Pollock et al. (2001) indicated that the assumptions of no tag loss could be violated in two ways; by tag loss in the first 482 Fishery Bulletin 101(3) few days after tagging or by chronic tag loss spread over an extended period of time, the latter being more difficult to model. Furthermore, Seber (1970) mentioned that the usefulness of the multiyear models depends not only on the validity of the underlying assumptions but also on the number of recaptures (i.e. parameter estimates based on a small number of tag recaptures would be biased). From our aquarium observations and field studies that show a small number of lobsters caught during the second and third re- capture periods, we conclude that a significant chronic tag loss does occur for the American lobster due to molting (i.e. the molt stage of the lobster at tagging and molting itself). Chronic tag loss impedes the effectiveness of multiyear re- capture models currently used (Heam et al, 1998; Hoenig et al., 1998a; Frusher and Hoenig, 2001) because it is not taken into account. We believe that assuming only a con- stant short-term tag loss for lobsters tagged with streamer tags is inadequate and can only bias estimates of survival and exploitation rate. Correcting for chronic tag loss after the first year at large for the American lobster, however, requires further knowledge, and more studies would be required to fully understand long-term tag loss. In conclusion, a high level of streamer tag loss is a major obstacle for using tagging studies to estimate natural mor- tality or to apply multiyear models for the American lobster. Because streamer tag loss is related to molting, adjustment is difficult because the molting frequency is size, sex, and environment dependent (Comeau and Savoie, 2001). In our attempt to estimate mortality at molt, it was found that differences in recapture rates of lobsters tagged in premolt and postmolt stages for a given molting period were not sta- tistically significEmt, thus suggesting a low level of natural mortality during the molt. The recapture rate for the 1996 tagging, for instance, was even higher for lobsters tagged in the premolt stage. Hence, tagging with streamer tags to es- tablish the level of natural mortality during the molt, or any other mortality that could be low, for the American lobster is not recommended. The alternative would be to develop another insertion tag with better retention through the molting process. Nevertheless, the streamer tag remains an adequate choice for studying lobster ecology and population biology. Streamer tags could be used to tag intermolt and postmolt lobsters during single recapture tagging studies to estimate the exploitation rate (Xiao et al., 1999). Based on our observations, a minimum adjustment of 24.9% (SD 2.9%) and 4.4% (SD 1.6%) is suggested for lobsters tagged in premolt and inter- or postmolt stages, respectively, and recaptured during the first recovery period. Acknowledgments The authors wish to thank all fishermen from Caraquet and the adjacent wharves that returned lobster tags and T Brideau, B. Comeau, J. Roussel, and F. Savoie for their tech- nical assistance in the field and during the tag collection. We especially thank A. Godin and his staff at the Aquarium et centre marin in Shippagan, New Brunswick, for their professional help during the aquarium experiment. We also want to thank J. M. Hanson and M. Moriyasu for critically reviewing the manuscript, and three anonymous reviews for thoughtful suggestions that improved the quality of this manuscript. Literature cited Aiken, D. E. 1980. Molting and growth. In The biology and manage- ment of lobsters. Vol. I: Physiology and behavior (J. S. Cobb and B. F. Phillips, eds.), p. 91-163. Academic Press, New York, NY. Beverton, R. J. H., and S. H. Holt. 1957. On the dynamics of exploited fish populations. U.K. Minist. Agric. Fish., Fish. Invest. 19:1-533. Brownie, C, D. R. Anderson, K. P. Burhnam, and D. S. Robson. 1985. Statistical inference from band recovery data: a handbook, 2nd ed. U.S. Fish Wildl. Serv. Resour. Publ. 156, 305 p. Comeau, M., W. Landsburg, M. Lanteigne, M. Mallet, P. Mallet, G. Robichaud, and F. Savoie. 1998. Lobster (Homarus americanus) tagging project in Caraquet (1993)— tag return from 1994 to 1997. Can. Tech. Rep. Fish. Aquat. Sci. 2216, 35 p. Comeau, M., M. Lanteigne, G. Robichaud, and F. Savoie. 1999. Lobster (Homarus americanus) movement in the southern Gulf of St. Lawrence — summary sheets of tagging projects conducted between 1980 and 1997. Can. Ind. Rep. Fish. Aquat. Sci. 249, HI p. Comeau, M., and M. Mallet. 2001. Estimating mortality rates by capture-recapture, catch-effort and change-in-ratio models for a spring Ameri- can lobster (Homarus americanus) fishery (.LFA 23). Can. Tech. Rep. Fish. Aquat. Sci. 2373, 20 p. Comeau, M., and F. Savoie. 200 1 . Growth increment and molt frequency of the American lobster (Homarus aiiiericanus) in the southwestern Gulf of St. Lawrence. J. Crust. Biol. 21:923-936. 2002. Movement of American lobster (Homon/,s americanus) in the southwestern Gulf of St. Lawrence. Fish. Bull. 100: 181-192. Ennis, G. P 1986. Sphyrion tag loss from the American lobster Homarus americanus. Trans. Am. Fish. Soc. 115:914-917. Frusher, S. D., and J. M. Hoenig. 2001. Estimating natural and fishing mortality and tag reporting rate of southern rock lobster iJasus cdwardsii) from a multiyear tagging model. Can. J. Fish. Aquat. Sci. 58:2490-2.501. Heam, W. S., K. M. Pollock, and E. N. Brooks. 1998. Pre- and post-season tagging models: estimation of reporting rate and fishing and natural mortality rates. Can. J. Fish. Aquat. Sci. 55: 199-205. Heam, W. S., T Polacheck, K. H. Pollock, and W. Whitelaw. 1999. Estimation of tag reporting rates in age-structured multicomponent fisheries where one component has observers. Can. J. Fish. Aquat. Sci. 56:1255-1265. Hoenig, J. M., N. J. Barrowman, W. S. Hearn. and K. H. Pollock. 1998a. Multiyear tagging studies incorporating fishing effort data. Can. J. Fish. Aquat. Sci. 55:1466-1476. Hoenig, J. M., N. J. Barrowman, K. H. Pollock. E. N. Brooks, and W. S. Heam. 1998b. Models for tagging data that allow for incomplete mixing of newly tagged animals. Can. J. Fish. Aquat. Sci. 55:1477-1483. Comeau and Mallet: Effect of timing of tagging on tag recapture rates for Homarus amencanus 483 Hill, B. J., and T. J. Wassenberg. 1985. A laboratory study of the effect of streamer tags on mortality, growth, moulting and duration of noctur- nal emergence of the tiger prawn Penaeus esculentus (Haswell). Fish. Res. 3:223-235. Howe, N. R.. and R R. Hoyt. 1982. Mortality of juvenile brown shrimp Penaeus aztecus associated with streamer tags. Trans. Am. Fish. Soc. Ill: 317-325. Krouse, J. S., and G. E. Nutting. 1990. Effectiveness of the Australian western rock lobster tag for marking juvenile American lobsters along the Maine coast. Am. Fish. Soc. Symp. 7:94-100. Landsburg, A. W. 1991. A field comparison of recapture rates of polyethylene streamer and modified sphyrion tags through molting of lobster (Homarus americanus). J. Shellfish. Res. 10, 225 P- Latour, R. J., J. M. Hoenig, J. E. Olney, and K. H. Pollock. 2001a. Diagnostics for niultiyear tagging models with appli- cation to Atlantic striped bass (Morone saxatilis). Can. J. Fish. Aquat. Sci. 58:1716-1726. 2001b. A simple test for nonmixing in multi-year tagging studies: application to striped bass tagged in the Rappahan- nock River. Trans. Am. Fish. Soc. 130:848-856. MaruUo, R, D. A. Emiiani, C. W. Caillouet, and S. H. Clark. 1976. A vinyl tag for shrimp (Panaeus spp.). Trans. Am. Fish. Soc. 105:658-663. Moriyasu M.. W. Landsburg, and G.Y. Conan. 1995. Sphyrion tag shedding and tag induced mortality of the American lobster, Homarus americanus H. Milne Edwards, 1837 (Decapoda, Nephropidae). Crustaceana 68:184-192. Pollock, K. H., J. M. Hoenig, W. S. Heam, and B. Calingaert. 2001. Tag reporting estimation: an evaluation of the reward tagging method. N. Am. J. Fish. Manage. 21:521-532. Pollock. K.H., J. M. Hoenig, and C. M. Jones. 1991. Estimating of fishing and natural mortality when a tagging study is combined with a creel survey or port sampling. Am. Fish. Soc. Symp. 12:423-434. Ricker, W. E. 1975. Computation and interpretation of biological statistics offish populations. Bull. Res. Board Can. 191, 382 p. Rowe, S., and R. L. Haedrich. 2001. Streamer tag loss from American lobsters. Trans. Am. Fish. Soc. 130:516-518. Scarratt, D. J., and P. F. Elson. 1965. Preliminary trials of a tag for salmon and lobsters. J. Fish. Res. Board Can. 22:421-423. Seber, G. A. F. 1970. Estimating time-specific survival and reporting rates for adult birds from band returns. Biometrika 57: 313-318 1982. The estimation of animal abundance and related parameters, 2nd ed., 654 p. Griffin, London. Templeman, W. 1935. Lobster tagging in the Gulf of St. Lawrence. J. Biol. Board Can. 1:269-278. Wilder, D. G. 1953. The growth of the American lobster {Homarus americanus). J. Fish. Res. Board Can. 10:371-403. Xiao, Y., J. D. Stevens, and G. J. West. 1999. Estimation of fishing and natural mortalities from tag experiments with exact or grouped times at liberty. Can. J. Fish. Aquat. Sci. 56:868-874. 484 Abstract— Bycatch. or the incidental catch of nontarget organisms during fishing operations, is a major issue in U.S. shrimp trawl fisheries. Because bycatch is typically discarded at sea, total bycatch is usually estimated by extrapolating from an observed bycatch sample to the entire fleet with either mean-per-unit or ratio estimators. Using both field observations of com- mercial shrimp trawlers and computer simulations, I compared five methods for generating bycatch estimates that were used in past studies, a mean- per-unit estimator and four forms of the ratio estimator, respectively: 1) the mean fish catch per unit of effort, where unit effort was a proxy for sample size, 2) the mean of the individual fish to shrimp ratios, 3) the ratio of mean fish catch to mean shrimp catch, 4) the mean of the ratios offish catch per time fished (a variable measure of effort), and 5) the ratio of mean fish catch per mean time fished. For field data, differ- ent methods used to estimate bycatch of Atlantic croaker, spot, and weakfish jaelded extremely different results, with no discernible pattern in the estimates by method, geographic region, or spe- cies. Simulated fishing fleets were used to compare bycatch estimated by the five methods with "actual" (simulated) bycatch. Simulations were conducted by using both normal and delta lognormal distributions of fish and shrimp and employed a range of values for several parameters, including mean catches of fish and shrimp, variability in the catches of fish and shrimp, variability in fishing effort, number of observa- tions, and correlations between fish and shrimp catches. Results indicated that only the mean per unit estimators pro- vided statistically unbiased estimates, while all other methods overestimated bycatch. The mean of the individual fish to shrimp ratios, the method used m the South Atlantic Bight before the 1990s, gave the most biased estimates. Because of the statistically significant two- and 3-way interactions among parameters, it is unlikely that estimates generated by one method can be converted or cor- rected to estimates made by another method: therefore bycatch estimates obtained with different methods should not be compared directly. Manuscript approved for publication 17 December 2002 by Scientific Editor Manuscript received 3 April 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:484:-500(2003). Estimation of bycatch in shrimp trawl fisheries: a comparison of estimation methods using field data and simulated data Sandra L. Diamond Department of Biology Box 3131 Texas Tech University Lubbock, Texas 79409 E-mail address: Sandra Diamond'fflttuedu Bycatch, as used in the present study, is the incidental catch of nontarget organisms that occurs to some extent in almost all commercial fisheries (Alver- son, 1994). Some of these incidentally caught organisms may be protected spe- cies— such as marine mammals, marine turtles, and seabirds — or they may be fish or invertebrates that are either harvested as target species by other fisheries, or species that fishermen call "trash fish" because they have little or no economic value. Bycatch in most commercial fisheries has only been a major issue since the 1980s — primarily because individuals caught as bycatch have historically been discarded at sea, leaving fishery managers and the general public unaware of the extent of bycatch mortality. For many organisms, bycatch may be a significant source of mortality, and inclusion of bycatch mor- tality in stock assessments or manage- ment plans may be critical for effective management. Because bycatch species are not usu- ally landed, quantifying bycatch poses a very different problem from that of quantifying the catch of a target spe- cies. Several methods of quantifying bycatch have been tried, including the requirement that fishermen record catch and bycatch in logbooks (Walsh and Kleiber, 2001), use of research ves- sel surveys to model commercial fishing (Nichols et al.'), and the placement of observers aboard fishing vessels (Julian and Beeson, 1998). Although direct ob- servation is the most accurate method, unless observer coverage of the fleet is complete, estimation of bycatch from observation data requires sampling of the fleet and then extrapolating from the samples (the observations) to the entire fleet using statistical estimators. Two types of statistical estimators are used: mean-per-unit estimators and ra- tio estimators. In both types of estima- tors, the observed catch of the bycatch species iy) is linked to an auxiliary vari- able (x) for which the population total is known (Cochran, 1977). In mean-per- unit estimators, the auxiliary variable is a measure of fishing effort such as tow, day, trip, etc., where each unit of effort is the same as one observation. In ratio estimators, the auxiliary vari- able is a variable that is correlated with the catch of the bycatch species, such as the catch of the target species or the number of hours fished (Cochran. 1977). The major difference between these two types of estimators is that the auxiliary variable in the mean-per-unit estimator is a substitute for the number of obser- vations rather than a mean value with a variance, while the auxiliary variable in the ratio estimator is the mean value of a quantity that varies from sample to sample. Although the statistical proper- ties of these two types of estimators are well known, the choice of which estima- tor to use in bycatch research is often based on the ease of collecting fleet information on the auxiliary variable, and not on any inherent properties of the estimators themselves or on any specific information about the relation- ship between the catch of bycatch spe- cies and the auxiliary variable. ' Nichols, S., A. Shah, G. Pellegrin Jr, and K. Mullin. 1990. Updated estimates of shrimp fleet bycatch in the offshore waters of the US Gulf of Mexico, 22 p. Pascagoula Laboratory, Southeast Fisheries Science Center, NMFS, PO Drawer 1207, Pasca- goula, MS 39568-1207. Diamond: Estimation of shrimp trawl bycatch 485 Bycatch is a major issue in the shrimp trawl fisheries of the Gulf of Mexico and the South Atlantic Bight. These fisheries are the most valuable fisheries in the southeast- em United States; almost 136,000 metric tons of shrimp, worth over $700 million, were landed in 2000 (NMFS^). It is estimated that 60-80% of the catch by weight in these fisheries is bycatch. Over 150 species have been reported in shrimp trawl bycatch, including marine turtles (Grouse et al., 1987) and juveniles of species that are highly valued as adults in other fisheries, such as weakfish iCynoscion rega- lis rVaughan et al.-'] ) and red snapper (Lutjanus campecha- nus [Goodyear'*! ). Both types of statistical estimators have been used to estimate bycatch in shrimp trawl fisheries. In the South Atlantic, biologists have periodically partici- pated as observers aboard commercial shrimp trawlers since at least the 1950s to characterize bycatch and estimate its magnitude (Fahy, 1966; Latham, ^ Lunz et al.,6 Fahy,'' Fahy« Fahy,^ Wolff,"* Keiser.n Knowltoni^). For most of the studies conducted between the 1950s and 2 NMFS (National Marine Fisheries Service). 2002. Unpubl. data Website: http://www.st.nmfs.gov/stl/comniercial/index. html. ^ Vaughan, D. S., R. J. Seagraves, and K. West. 1991. As assess- ment of the status of the Atlantic weakfish stock, 1982-1988. Special Report 21, 29 p. Atlantic States Marine Fisheries Commission, 1444 Eye Street, N.W., Sixth Floor, Washington, DC 20005. ^ Goodyear, C. P. 1995. Redsnapper in US watersofthe Gulfof Mexico. Contribution MlA-95/96-05, 171 p. Miami Laboratory, Southeast Fisheries Science Center, NMFS,75 Virginia Beach Drive, Miami. Florida 33149-1099. ^ Latham, F F. 1951. Evidence of fish loss due to shrimping in Pamlico Sound. Appendix B in The destruction of small fish by the shrimp trawlers in Pamlico Sound, North Carolina (G. R. Lunz, J. L.,McHugh, E. W. Roelofs, R. E. Tiller, and C. E. Atkinson), p. 17-24. Committee Report to the Atlantic States Marine Fisheries Commission, 1 November 1951. Atlantic States Marine Fisheries Commission, 1444 Eye Street, N.W, Sixth Floor, Washington. DC 20005. « Lunz. G. R.. J. L. McHugh, E. W. Roelofs, R. E. Tiller, and C. E. Atkinson. 1951. The destruction of small fish by the shrimp trawlers in Pamlico Sound, North Carolina. Committee Report to Atlantic States Marine Fisheries Commission, 1 November 1951, 34 p. Atlantic States Marine Fisheries Commission, 1444 Eye Street, N.W., Sixth Floor, Washington, DC 20005. "Fahy, W. E. 1965a. Report of trash fish study in North Caro- lina in 1962. Division of Commercial and Sports Fisheries, NC Department of Conservation and Development, Special Scientific Report 5. 20 p. NC Division of Marine Fisheries, 3441 Arendell St., Morehead City, NC 28557. 8 Fahy, W. E. 1965b. Report of trash fish study in North Caro- lina in 1964. Division of Commercial and Sports Fisheries, NC Department of Conservation and Development, Special Scientific Report 7, 13 p. NC Division of Marine Fisheries, 3441 Arendell St., Morehead City. NC 28557. 3 FahyW.E. Unpubl.datacitedinBrown, J.,andE.McCoy 1969. A review of the North Carolina scrap fishery. Division of Com- mercial and Sports Fisheries. NC Department of Conservation and Development. Information Series 1, 12 p. NC Division of Marine Fisheries, 3441 Arendell St., Morehead City, NC 28557. 1" Wolff, M. 1972. A study ofNorth Carolina scrap fishery. NC Department of Natural and Economic Resources, Special Sci- entific Report 20, 29 p. NC Division of Marine Fisheries, 3441 Arendell St., Morehead City, NC 28557. the 1980s, fisheries bycatch was estimated by using a ratio estimator, that is to say by calculating the observed ratio of fish (F) bycatch to shrimp (S) by weight and then multiply- ing by the total pounds of shrimp landed by the fleet (the F:S ratio estimator). The catch of shrimp was used as the auxiliary variable primarily because better records were kept of shrimp landings than of any measure of fleet effort. By the late 1980s, the problem of shrimp trawl bycatch in the United States was considered to be of such magnitude that in 1990 the Magnuson Fishery Conservation and Management Act (Magnuson Act) was amended to include bycatch research. Beginning in 1992, observers trained by the National Marine Fisheries Service (NMFS) to use a standardized sampling protocol (NMFS'-*) rode aboard paid volunteer commercial vessels in the South Atlantic and Gulf of Mexico. The 1992-94 observation data collected in the South Atlantic were used to estimate bycatch by spe- cies with a mean-per-unit estimator, which was the weight or number of fish caught per observed trip multiplied by the total number of trips taken by the fleet (the CPUE- mean-per-unit estimator). Trips were used as the auxiliary variable because fleet effort data were available at the trip level and this method was thought to be less variable than the F:S ratio method (SEAMAP'"). To date, there have been no detailed studies on how these different techniques compare to each other, or how accurately they estimate bycatch. Vaughan and Nance''' in a draft paper compared the estimated bycatch of mackerels (Scomberomorus spp.) and cobia {Rachycentron canadum) using both methods and found much higher estimates with the F:S ratio estimator than with the CPUE-mean-per-unit estimator. Because of the wide range of estimation meth- ods used over the years, the discrepancy in the estimates generated by the different methods, and the increasing im- portance of bycatch estimation for shrimp trawl fisheries " Reiser, R. K. 1977. The incidental catch from commercial shrimp trawlers of the South Atlantic states. Technical Report 26, 38 p. South Carolina Wildlife and Marine Resources Depart- ment, South Carolina Department of Natural Resources, Rembert C. Dennis Building, 1000 Assembly Street, Columbia, SC 29201. 12 Knowlton, C. J. 1972. Fishes taken during commercial shrimping in Georgia's close inshore ocean waters. Con- tributed Series 21, 42 p. Georgia Department of Natural Resources, Coastal Resources Division, One Conservation Way, Suite 300, Brunswick, GA 31520. 1-^ NMFS (National Marine Fisheries Service). 1992. Shrimp trawl bycatch characterization. Sampling Protocol Manual for Data Collection, 62 p. Galveston Laboratory, Southeast Fisheries Science Center, NMFS, 4700 Avenue U, Galveston, TX 77551-5997. ''• SEAMAP (Southeast Area Monitoring and Assessment Pro- gram ). 1996. Estimates of finfish bycatch in the South Atlan- tic Shrimp Fishery, July 24, 1995 (R. Peuser, ed.), 64 p. Final report of the Southeast Area Monitoring and Assessment Pro- gram (SEAMAP), SEAMAP-South Atlantic Committee, Shrimp Bycatch Work Group. Atlantic States Marine Fisheries Com- mission, 1444 Eye Street, N.W., Sixth Floor, Washington, DC 20005. '^ Vaughan, D. and J. Nance. 1998. Estimates of bycatch of mackerel and cobia in US South Atlantic shrimp trawls. Re- port for Gulf of Mexico and South Atlantic Fishery Management Councils, February 16, 1998, 26 p. NMFS -SEFSC, Beaufort Laboratory, 101 Fivers Island Road, Beaufort NC 28516. 486 Fishery Bulletin 101(3) Sound 36°00' 35°00' Figure 1 Map of North Carolina waters. Shrimping operations were observed in northern Pamlico Sound (between the mouth of the Pamlico River and southern Roanoke Sound) and the lower third of the Cape Fear River. For total bycatch, fleet shrimp landings and fleet shrimp effort, the northern region includes Pamlico Sound and its tributaries, and the southern region includes from the Cape Fear River to the New River. and other fisheries, fishery biologists need clear guidance on which method to use to estimate bycatch and they need a definitive knowledge of which methods are best under the varying conditions that might be found in a field observer study. In this article, I use both field data and computer simula- tions to compare the methods of bycatch estimation used in past studies. First, using field observations of Atlantic croaker {Micropogonias undulatus), spot (Leiostomas xan- thurus), and weakfish bycatch from shrimp trawlers in North Carolina, I compare bycatch estimates generated by the CPUE-mean-per-unit estimator with two different forms of the F:S ratio estimator, the mean of the individual fish to shrimp ratios and the ratio of the mean catch offish to the mean catch of shrimp. I then simulate fishing fleets with different catches of fish and shrimp, and estimate bycatch using the following five different estimators, a mean-per-unit estimator and four forms of the ratio esti- mator, respectively: 1 ) the mean fish catch per unit of effort, where unit effort is a proxy for sample size, 2) the mean of the individual fish to shrimp ratios, 3) the ratio of mean fish catch to mean shrimp catch, 4) the mean of the ratios offish catch per time fished (a variable measure of effort), and 5) the ratio of mean fish catch per mean time fished. The simulations employ different mean catches offish and shrimp, different levels of variability around the catches offish and shrimp and around the variable measure of ef- fort in the ratio estimator, and different levels of observer coverage, or the number of observations. I also investigate the effects on the bycatch estimates of different underlying distributions offish and shrimp, including normal distribu- tions offish and shrimp with different levels of correlation between the catches of fish and shrimp, and delta lognor- mal distributions of both fish and shrimp, with differing probabilities of catching fish or shrimp. Materials and methods Field sampling To compare the methods described in the literature using field data, I observed shrimping operations aboard com- mercial shrimp boats from July through October 1995 in Pamlico Sound, North Carolina, and from August through October 1995, in the Cape Fear River, North Carolina (Fig. 1). These two areas have different levels of fishing effort, different fish-to-shrimp ratios, and different prob- abilities of catching fish and shrimp. All fishermen coopera- tors were unpaid volunteers, and I did not direct them in any way regarding where or how to fish. Although sampled boats were not randomly chosen, the fishermen appeared to use gear and fishing methods similar to those of other shrimpers, and other shrimpers were often seen fishing in the area near the sampled boats. Sampled shrimp boats towed one or two nets, and all nets contained some form of turtle excluder device (TED) and bycatch reduction device (BRD) required by regulation. Diamond: Estimation of shrimp trawl bycatch 487 To sample the catch, I used the NMFS bycatch samphng protocol as described below. If the boat carried two nets and no try net (the small net towed in front of the main nets which is used to survey the catch at short time intervals), I randomly picked one net (the "selected net") by flipping a coin. If the boat had a try net, I picked the opposite net. I weighed the total catch of the selected net on a flat agri- cultural scale by emptying the net into a plastic tub placed on a scale. After having been weighed, the catch of the se- lected net was dumped onto the deck or into a culling tray that was divided so that the catch of the selected net was separated from the catch of the unselected net. Following the NMFS protocol, I mixed the selected net contents thor- oughly with a shovel, then took a random sample and set it aside until after the rest of the net contents had been sort- ed. To sort the net contents, marketable shrimp, which are pink shrimp (Faifantepenaeus duorarum), brown shrimp {Farfantepenaeus aztecus), and white shrimp (Litopenaeus setiferus) larger than would comprise about a 70-80 count (i.e. 70-80 shrimp per pound) were separated from the rest of the contents of the selected net, weighed, and then re- turned to the fisherman. The unsampled bycatch from the selected net was discarded overboard. The random sample taken from the selected net was then weighed. Market shrimp in the sample were taken out, weighed and counted by species, and returned to the fisherman. The bycatch por- tion of the sample, including undersized market shrimp, mantis shrimp, and all other fish and invertebrates, was packaged in plastic bags and placed on ice for the remain- der of the trip. Bycatch samples were brought back to the laboratory and frozen. Samples, including market shrimp, averaged 12% by weight of the total catch of the selected net. and ranged from 5% to 37% by weight. Expansion of observed bycatch to the entire tow In the laboratory, I thawed and rehydrated the bycatch sample in water. I sorted each sample by species and weighed each species as a group. All individuals of each species were then weighed and measured separately. To account for differ- ences between the scales used on the boat and those used in the laboratory, and for weight loss due to freezing, I cor- rected the weight of the total catch of each net measured on the boat by the ratio of the sample weight from the labora- tory to the sample weight from the boat as follows: Corrected total weight ^ = lab sample weight boat total weighty x (1) boat sample weighty where corrected total weighty = the corrected weight of the f-^ selected net; boat total iveightj = the weight of the entire catch of the/^" selected net measured on the boat; lab sample iveightj = the weight of the bycatch sample of the/'' net mea- sured in the laboratory plus the shrimp sample weight from the boat; and boat sample iveightj = the weight of the entire sample (including shrimp) from the 7^'' net weighed on the boat. This correction averaged less than 5% across all selected nets. To expand the catch in weight of each bycatch spe- cies from the sample to the entire selected net (called the "species net weight"), the total corrected weight of each selected net was multiplied by the fraction of the sample from the selected net that consisted of the bycatch species, as follows: corrected total weight, x Species net weighty ^ = species sample weighty ^ total sample weight j (2) where species net weighty : = the estimated catch in weight of the i^^ species in the/'' net; corrected total weight = the corrected weight of the total catch of the/'' net from Equation 1; species sample iveightj , = the weight of the j"' species in the sample from the /'' net; and total sample iveightj = the weight of the bycatch sample from the/'' net mea- sured in the laboratory plus the weight of the market shrimp in that sample mea- sured on the boat. Because the net contents were thoroughly mixed before sampling, I assumed, following the NMFS protocol, that there would be minimal variance among samples if more than one were taken. Expanding the catch in numbers of each bycatch spe- cies from the sample to the entire selected net (called the "species net number") could not be done in the same way as the expansion for the species net weight because there were often organisms like sea lettuce or pieces of fish or crabs that were weighed but that could not be counted. The species net number was therefore calculated by dividing the species net weight by the average weight per whole individual: species net weight, ^ Species net number^ - = species sample weight, ^ species sample number,^ (3) where species net number species net weight, the estimated number of individuals of the i''' spe- cies in the/'' net; the estimated total weight of the i*^ species in the /•■ net from Equation 2; 488 Fishery Bulletin 101(3) species sample weight species sample number the weight of the t"^ species in the bycatch sample from the/*^ net; and the number of whole indi- viduals of the /''' species in the bycatch sample from the./"^ net. To expand the observed bycatch from selected net to the entire tow, I multiplied either the species net weight or the species net number from each net by the number of nets towed concurrently. Bycatch estimation To compare the methods of bycatch estimation used in past studies, I estimated the bycatch of Atlantic croaker, spot, and weakfish (three of the most commonly caught bycatch species) using two categories of statistical estimators: a mean-per-unit estimator using the mean observed bycatch per day expanded by the total number of days fished (the CPUE-mean-per-unit method) and a ratio estimator using the observed ratio of fish to shrimp expanded by the total shrimp landings (the F:S ratio method). Because my purpose was to compare bycatch esti- mation methods and not to generate bycatch estimates that could be used for management purposes, I estimated total bycatch of these three species only for certain months and geographic regions within North Carolina corresponding to the times and areas that I observed shrimp trawling. The term "shrimp fleet" in the following paragraphs therefore refers only to shrimpers operating in those times and areas. In the calculations, I used bycatch per day instead of the bycatch per tow or bycatch per trip. I could not use tow as the unit of effort because there was no information on the number of tows made by the fleet to use as an expansion factor Although information on the number of trips made by the fleet was available, I could not use trip as the unit of effort because, although trips can last several days, all of the trips that I sampled were one-day trips. If my observations had also included a random sample of multiday trips, the unit of effort would have been trips instead of days. The CPUE mean per unit method was based on the fol- lowing equations: Mean ohsened hyaiich^ day X"^- (4) where mean observed _ the observed average bycatch in bycatch per day weight or number of the i^^ spe- cies on the d^^ day; n = the number of days observed; and F^ J = the sum of the expanded weight or number of the (''' bycatch spe- cies observed in all tows made on the d"' day; and mean obser\'ed bycatch' Total bycalcli, .-/.yf = ■ x clay tiiiiil trips X mean days trip (5) where total bycatch ^(.p^,^. = the total fleet bycatch of the (''^ species esti- mated by the CPUE method; mean observed bycatch ^ per day = the observed average bycatch of the i^^ spe- cies per day from Equa- tion 4; total trips = the total number of trips taken by the shrimp fleet; and mean days per trip = the average number of days that each fishing trip lasted based on the fleet. The total trips and mean days per trip were calculated from the North Carolina Division of Marine Fisheries (NCDMF) trip ticket database, as follows. To obtain the total number of trips, I first collapsed the trip ticket database so that each fisherman could have only one ticket for shrimp on a single day. In the database, each trip ticket does not represent one trip, but the sale to one dealer Fishermen could obtain more than one trip ticket per day by selling different size catego- ries of shrimp (each size category commands a different price, and generates a separate trip ticket ), or by selling their catch to more than one dealer I then calculated the time (in days) between the first and last trips for each fisherman whose trips occun-ed between 1 July and 31 October in Pamlico Sound and its tributaries (called the northern region) and between 1 August and 31 October in the Cape Fear River and nearby waters (the southern region). Because inshore waters were closed to shrimping on weekends, I multiplied all time spans greater than 7 days by 5/7 (0.714) to obtain the number of days fished. The number of days fished was summed and then divided by the number of trips for each region to obtain the mean days per trip. The F:S ratio estimator method was initially undertaken in two ways: by using the mean of the fish to shrimp ratios, called the mean of the ratios or the "basic" F:S ratio esti- mator method (Equation 6), and by using the ratio of the average catch offish to the average catch of shrimp, called the ratio of the means or the "grand" F:S ratio estimator method (Equation 7). The two methods are shown math- ematically as follows: Total h\catch, , ^„ = — > -^ x total shrimp landed. (6) Total hvcatcli ,v,. =-^ x total slniinp landed, (7) rf=l where total bycatch, y,,i, = the total fleet bycatch of the ;"' species estimated by the basic F:S method; total bycatch, ^^g = the total fleet bycatch of the /"' species estimated by the grand F:S method; Diamond: Estimation of shnmp trawl bycatch 489 F, J = the sum of the expanded weight or number of the ;'"' species observed in all tows made on the d^^ day; Sj = the sum of the expanded weight of market shrimp observed in all tows made on the c/* day; and n = the number of days observed. Because of the small number of days observed in each area, I also used the basic F:S ratio estimator with the Hartley-Ross correction for biases caused by small sample size (Cochran, 1977): Toliil hvccilcli^ f^if = total hxcatch, ^ j„ H (\\-rx), (8) )i - 1 where total bycatch^ ^jf^ = the total fleet bycatch of the ;'"' species estimated by the bias- corrected F:S ratio estimator; total bycatch^ p,.g = the total fleet bycatch of the ("^ species estimated by the basic F:S ratio using Equation 6; n = the number of days observed; N = the total number of days fished from the trip ticket database; y, = the mean bycatch of of the ;"^ species observed per day in weight or numbers from Equa- tion 4; F = the mean of the F:S ratios from Equation 6; and .V = the mean catch of market shrimp observed per day in weight or numbers. Total shrimp landings used in Equations 6 and 7 were obtained from the NCDMF trip ticket database for the northern region from July to October and for the southern region from August to October. In the trip ticket database, some shrimp weights were reported as "heads-on" and others as "heads-ofT'; therefore I converted heads-off weight to heads-on weight with a conversion factor of 1.583, taken from the average of pink, brown, and white shrimp conver- sion information used by the National Marine Fisheries Service (Fisheries Statistics of the United States, 1977). Bycatch simulations For the bycatch simulations, I created different fishing fleets of 1000 "boats" in Matlab 5.0 (The Mathworks, Natick, MA). For the normally distributed catch data, the catch of fish, the catch of shrimp, and the hours fished for each boat in a fleet were generated by using multivariate random normal distributions with a mean and variance that was specific to that fleet. I simulated observer data for each fleet by taking a random sample of boats from the fleet, resampling the sample 1000 times, then using the mean of the bootstrapped observer data in the equa- tions described below to estimate fleet bycatch. In the dif- ferent fleets, the mean catches of fish and shrimp ranged from 0.01 to 1000, giving fish to shrimp ratios of 0.001 to 100,000. In some fleets, the catches offish and shrimp were correlated, with correlation coefficients ranging from 0.5 to -0.5 (Table 1). Coefficients of variation (CVs) for fish catch and hours fished ranged from 20% to 80%, CVs for shrimp catch ranged from 20% to 120%, and the number of obser- vations ranged from 20 to 500, giving observer coverages of 2% to 50%. of the fleet. Although the range of mean catches I used in the simulations may seem fairly broad, they are within the range of the field data, depending on whether these were the mean catches per tow, per day, or per trip. The ranges of CVs for fish and shrimp catches were fairly narrow compared to those from the field data because CVs vary up to several hundred percent, particularly for patchy species. Observer coverage in the field is usually much less than 50%>, but I picked 50% as the upper limit of the range to see if greater observer coverage (i.e., a greater sample size of observations per fleet) increased the accuracy of the bycatch estimates. Bycatch estimates were calculated by using a mean per unit estimator and four forms of the ratio estimator, as described below. The CPUE mean per unit estimator was calculated by using the following equations, which are more general versions of Equations 4 and 5: Total bycatch, , Mean obsen'ed bycatch, I UE Mean hvcatch (9) UE -xtntal fleet effort. (10) where mean observed the observed average bycatch of bycatch^ per UE = the i"" species per unit of effort (tow, day, or trip); n = the number of observed tows, days, or trips; F, ug = the expanded weight or number of the i'*' bycatch species observed on the [/£"' tow, day, or trip; total bycatch-,(,piji' = the total fleet bycatch estimated by the CPUE method; and total fleet effort = the total number of tows, days, or trips fished by the fleet. The four ratio estimators were as follows: 1) the mean of the individual F:S ratios, called the "basic F:S'' ratio estimator (Eq. 11), 2) the ratio of the F:S means, called the "grand F:S" ratio estimator (Eq. 12), 3) the mean of the individual catch per effort ratios using a variable measure of effort such as hours fished as the auxiliary variable, called the "basic CPE" ratio estimator (Eq. 13), and 4) the ratio of the mean catch per mean effort using a variable measure of effort such as hours fished as the auxiliary variable, called the "grand CPE" ratio estimator (Eq. 14). Both F:S ratio estimators (Eqs. 11 and 12) are similar to the ones used in the field study (Eqs. 6 and 7), except that the observations could be from a tow, day, trip, or other measure of unit effort, rather than one day, as used in the field study. 490 Fishery Bulletin 101 (3) 1 V- F. Total bycatch- ^^^ = — > — ^ x total shrimp landed (11) Total bycatch "-^. ■■=1 - X total shrimp landed ( 12) where total bycatch^ pgg = the total fleet bycatch of the i"^ species estimated by the basic F:S ratio estimator; total bycatch^ pgQ = the total fleet bycatch of the i'^ species estimated by the grand F:S ratio estimator; F, ^, = the expanded weight or number of the /''^ bycatch species ob- served in the e'*' tow, day, or trip; S^ = the expanded weight of market shrimp observed in the e'*^ tow, day, or trip; n = the number of tows, days, or trips observed; and total shrimp landed = the sum of the total weight of shrimp landed by the fleet. 1 -v-i F Tdtal b\calch^^p^g = —y —'^^x total hours fished (13) Total bycatch ^ p^^; = -^ x total hours fished (14 ) where total bycatch, cpi^g = the total fleet bycatch of the f* species estimated by the basic CPE ratio estimator; total bycatch, (-.p^-Q = the total fleet bycatch of the /"' species estimated by the grand CPE ratio estimator; F, ^ = the expanded weight or number of the i*-^ bycatch species observed in the e"" tow, day, or trip; //^, = the hours fished in the e''' tow, day, or trip; n = the number of observed tows, days, or trips; and total hours fished = the sum of all hours fished by the fleet. To avoid confusion, it is important to note how fishing effort is used in the five estimators. All five estimators use a unit measure of fishing effort, such as a tow, day, or trip, as one sample, and the sample size for a fleet is the number of tows, days, or trips observed. In the CPUE mean-per-unit estimator, the estimate of total bycatch is a simple expan- sion of the observed bycatch per sample to the whole fleet. In the F:S ratio estimators, the unit effort appears in the calculations because the ratios of fish to shrimp are the amounts caught per tow, day, or trip (i.e. per sample). In the CPE ratio estimators, two measures of effort are used. As before, one measure of effort is the unit effort, such as a tow, day, or trip, that is equivalent to a sample, and the second measure of effort is the variable measure of effort, such as the hours fished, the distance towed, or the area covered, that is used as the auxiliary variable. The CPE ratio estimator is thus based on the amount offish caught per hour fished (for example) in each tow, day, or trip. The delta lognormal simulations were very similar to the normal simulations, except that I simulated the catch of fish and shrimp by using probabilities of catching fish or shrimp ranging from 0.05 to 0.95, multiplied by average catches offish or shrimp generated from random lognormal distributions with means ranging from 0.01 to 1000. Log- normal functions have parameters of// and a'^. which are the mean and variance of the normally distributed variable before logarithmic transformation. To obtain values of;/ and a^ from a lognormal distribution with a given mean and variance, I used an iterative procedure (the Solver procedure in Microsoft Excel, vers. 2000, Microsoft Corpo- ration, Redmond, WA) to estimate fj and o^ based on the following equations: mean = e [-^) (15) (16) where mean = the mean of the lognormal distribution of the catch offish or shrimp; var - the variance of the lognormal distribution of the catch offish or shrimp; H = the mean of the normally distributed catch of fish or shrimp before logarithmic trans- formation; and o^ = the variance of the normally distributed catch of fish or shrimp before logarithmic transformation. Levels of observer coverage and CVs for fish catch, shrimp catch, and effort and were the same as in the normally distributed data. In these simulations, sampled shrimp catch could be zero if the probability of catching shrimp was low and the number of observations was small, leading to F:S ratios of infinity. In these cases, for the basic F:S ratio estimator, the fish-to-shrimp ratio was the catch offish divided by the expected catch of shrimp (probability of catching shrimp times the mean catch ). For the grand F: S ratio estimator, if the average bootstrapped sample catch of shrimp was zero, Matlab substituted a value of 65535 to avoid division by zero. To avoid biases, these grand F; S simulations were left out of the subsequent analyses. In field saini)liiig, tows that caught no shrimp at all were rare, but tows that caught only small unmarketable shrimp that were discarded as bycatch occurred occasionally early in the season and after big rainstorms. Diamond: Estimation of shrimp trawl bycatch 491 Table 1 Parameters and their values used in the bycatch simulations for normal and delta lognormal distributions of fish and shrimp. Abbreviations for the parameters are shown in parentheses Distribution Parameter Values Normal Mean fish catch (AvgF) 0.01, 0.1, 1, 10, 100, 1000, 10,000 Fish CV (FCV) 20%, 50%, 80%, 120% Mean shrimp catch (AvgS) 0.01, 0.1, 1, 10, 100, 1000, 10,000 Shrimp CV(SCV) 20%, 50%, 80% Mean hours fished 1.0 Hours fished CV (ECV) 20%, 50%, 80% F:S ratio 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10,000, 100,000 Correlation coefficient (r) -0.5,-0.25,0,0.25,0.5 Number of observations (n) 20,50,100,500 Observer coverage 2%, 5%, 10%, 50% Delta lognormal Probability of catching fish (P(F)) 0.05,0.2,0.5,0,8,0.95 Mean fish catch (AvgF) 0.01, 0.1, 1, 10, 100, 1000, 10,000 Fish CV (FCV) 20%, 50%, 80% Probabihty of catching shrimp (P(S)) 0.05,0.2,0.5,0.8,0.95 Mean shrimp catch (AvgS) 0.01, 0.1, 1, 10, 100, 1000, 10,000 Shrimp CV(SCV) 20%, 50%, 80%, 120% Mean hours fished 1.0 Hours fished CV (ECV) 20%, 50%:, 80% F:S ratio 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10,000, 100,000 Number of observations ( n ) 20,50,100,500 Observer coverage 2%, 5%, 10%, 50% To statistically analyze the overall biases shown by each estimator regardless of fishing conditions (i.e. mean catches of fish or shrimp, CV, etc.), I first used paired sample /-tests (SAS v. 8, The SAS Institute, Gary NC) to separately compare each of the five estimates of fleet by- catch with the "actual" bycatch for that fleet, based on the following equation: where % bias estimated bycatch^ ^^^ "actual" bycatch , estimated bycatch, ,„ ^ - "actual" bycatch, ,, x 1 00 ( j^ 7 ) "actual" bycatch,^ „ I, = the bias in the m"^ estimator for the 6'h fleet; ,„ f, = the bycatch of the i^^ species by the w'*' estimator for the b^^ fleet; and the simulated actual bycatch of the ;"^ species by the 6"^ fleet. For these statistical tests, all fleets with normal distribu- tions offish and shrimp were combined and analyzed sepa- rately from the fleets with delta lognormal distributions of fish and shrimp, giving sample sizes of 21,600 fleets for the normal distribution and 118,810 fleets for the delta lognor- mal distribution (Table 1). To look for significant factors influencing the bycatch estimates for each of the five esti- mation methods, I used ANOVAs on all main effects and all 2-way and 3-way interactions of the main effects for each estimator. Although 7-way interactions were possible in the normally distributed simulations and 8-way interactions were possible in the delta lognormal simulations (Table 1 ), I stopped the analysis at 3-way interactions because of the difficulty in interpreting higher level interactions. Main effects were the following: mean catches offish and shrimp, CVs of fish catches, CVs of shrimp catches, CVs of hours fished, number of observations or observer coverage, cor- relation coefficient in the normally distributed simula- tions, and the probabilities of catching fish and shrimp in the delta lognormal simulations. In these ANOVAs, the response variable was the percent bias for each method, as calculated above. Results Field sampling I observed 16 tows from five trips in Pamlico Sound between July and October 1995 and 24 tows from five trips in the Cape Fear River between August and October 1995 (Table 2). According to the 1995 trip tickets, these months comprised the peak of the summer brown shrimp and fall white-pink shrimp seasons; 77% of the total shrimp catch and 75% of the total trips in the northern region and 63% of the total shrimp catch and 54% of the total trips in the southern region occurred during those months. All observed tows were daytime tows, which is when fishing generally occurs in these areas. All nets sampled in July were 2-seam or 4-seam flat trawls, designed to catch brown shrimp, and 492 Fishery Bulletin 101(3) Table 2 Characteristics of boats and fishing operations observed in North Carolina waters in 1995 trawlers operating using their standard operating procedures. Each enti-y represents one c fishing trip lasted one day. Avg. h/tow = mean hours per tow. All boats were commercial shrimp bserved fishing trip. Each observed Area Month Boat name Boat length (m ) No. of tows No of nets Headrope length (m ) Avg. h/tow Pamlico Sound Jul Last Toy 8.3 5 2 9.8 1.2 Aug Islander 8.9 2 18.4 1.1 Aug Last Toy 8.3 2 13.1 1.3 Sep Islander 8.9 5 18.4 1.1 Oct Islander 8.9 2 18.4 1.2 Average 8.7 3.2 1.2 15.6 1.2 Cape Fear River Aug Cajiin Lady 16.0 9 2 18.0 0.9 Aug Cajiin Lady 16.0 2 2 18.0 0.7 Sep Sea Mullet 14.2 4 2 14.8 1.3 Sep Cajun Lady 16.0 5 2 18.0 1.0 Oct Dorothy Glen 11.4 4 1 19.7 1.9 Average 14.7 5 1.8 17.7 1.2 Table 3 Landed and observed shrimp catch (heads-on kg landed), effort (total number of trips and days/trip), and catch per unit of effort (CPUE, kg /trip and kg/day) for regions within North Carolina. Information on fleet totals was obtained from the North Carolina Department of Marine Fisheries trip ticket database for vessels fishing during July through October 1995 in the Northern region (Pamlico Sound and tributaries), and during August through October 1995 in the Southern region (the Cape Fear River and nearby waters). Observations were conducted during these same months in Pamlico Sound and the Cape Fear River. Shrimp catch No. of Mean CPUE Days CPUE Region or location (kg landed) trips days/trip (kg/trip) fished' (kg/day) Fleet totals Northern 2,018,612 3196 3.64 631.6 11,633 173.5 Southern 122,893 1716 3.48 71.5 5972 20.6 Observations Pamlico Sound 278 5 1 55.6 5 55.6 Cape Fear River 867 5 1 173.4 5 173.4 ^ This value represents the maximum days fished because the calculations are between landings. based on the assumption that fishing takes place every allowable day both tongue trawls and flat trawls were sampled in August through October. Tongue trawls are modified mongoose trawls that have a higher vertical profile for catching white shrimp. In addition, the tongue trawls had a greater headrope length than the flat trawls; therefore many of the fishermen switched from pulling two flat trawls to pulling one larger tongue trawl. Tows typically lasted around one hour The observed catch of shrimp per day in the Cape Fear River was almost three times higher than the observed catch of shrimp per day in Pamlico Sound (Table 3). Total shrimj) landings and total shrimp trips during the observed months from the 1995 trip ticket database were used as the expansion factors in the estimates. Over half of the total .shrimp landings, or 2,018,622 kg, were caught in the northern region between July and October and only 122,893 kg came from the southern region between August and October, the months that corresponded to the observa- tions. Although there were about twice the number of trips and days fished the northern region, the average catch per trip (kg/trip) from the northern region was almost nine times higher than the catch per trip from the southern region (Table 3). The different estimation methods made a tremendous difference in the estimates of bycatch, but the differences were exactly opposite in the two geographic regions and varied somewhat by species. Total bycatch estimates de- rived with the basic F:S ratio estimator (mean of the ratios) by both weight and number were two to seven times higher than those based on the CPUE-mean-per-unit method for all species in the northern region, and about two to five times lower by both weight and number for all species in the southern region (Table 4). For Atlantic croaker and Diamond: Estimation of shrimp trawl bycatch 493 Table 4 Total bycatch in weight and numbers estimated from observation data using different estimation methods. The CPUE-mean- per-unit estimator (CPUE= catch per unit of effort), w hich is based on the catch per day. uses day as a proxy for sample size. The basic F:S ratio estimator is the mean of indi vidual fish (F) to shrimp (S) ratios, and the grand F:S ratio estimator is the ratio of the mean catch of fish to the mean catch of s hrimp. The Hartley-Ross method is the basic F:S ratio estimator corrected for small | sample sizes. AC = Atlantic croaker, SP = spo t, and WF = weakfish. The northern region includes Pamlico Sound and ts tributaries. and the southern region includes from the C ape Fear River to the New River. Estimates are for July through October 1995 in the northern region and Angus t through Octobei 1995 in the southern region . See text for calculations. Equations for the 95% CL are | from Cochran (1977), Equations 2.24, 6.12, and 6.14. Bycatch estimate by weight (millions of kg) F:S ratio estimator CPUE-mean-per-unit estimator Region and Basic Grand Hartley-Ross species Total wt. ±95% CL Total wt ±95% CL Total wt. ±95% CL Total wt. ±95% CL Northern AC 0.6 0.5 2.1 4.1 1.7 0.1 2.6 0.1 SP 0.5 0.8 2.9 5.6 1.5 0.3 2.3 0.3 WF 0.6 1.2 1.5 1.5 1.9 0.4 1.7 0.4 Southern AC 0.1 0.2 0.03 0.05 0.02 0.09 N/Ai N/A SP 0.02 0.05 0.006 0.01 0.003 0.03 N/A N/A WF 0.2 0.2 0.03 0.04 0.2 0.1 N/A N/A Bycatch estimate by number (millions) F:S ratio estimator CPUE-mean-per-unit Region and estimator Basic Grand Hartley-Ross species Total no. ±95% CL Total no ±95% CL Total no. ±95% CL Total no. ±95% CL Northern AC 28.1 23.7 186.0 263.8 84.8 8.4 144.1 8.4 SP 18.7 27.2 146.5 305.5 56.5 11.7 109.2 11.7 WF 11.8 20.8 36.2 34.8 35.7 6.4 35.9 6.4 Southern AC 13.7 20.9 4.1 7.1 1.7 12.7 N/A N/A SP 1.5 3.4 0.4 0.6 0.2 1.8 N/A N/A WF 19.1 29.2 3.5 4.9 2.3 16.3 N/A N/A ' The estimator gave negative estimates for bycatch. spot, the grand F:S ratio estimate (ratio of the means) was intermediate between the basic F:S ratio estimate and the CPUE-mean-per-unit estimate in the northern region, but was lower than either of the other estimates in the south- em region. The grand F:S ratio estimate for weakfish was larger by both weight and number than either of the other two estimates in the northern region and was the smallest in the southern region. The Hartley-Ross bias-corrected F:S ratio estimator gave estimates for the northern region that fell between the basic and grand F:S methods but gave negative estimates for the southern region (Table 4). CVs of the catch rates were generally larger for the basic F:S ratio estimator method than for CPUE-mean-per-unit estima- tor for spot in the northern region and Atlantic croaker in both regions, and smaller for spot in the southern region and weakfish in both regions. CVs estimated by the grand F:S ratio method were much smaller than those for either of the other methods for the northern region and much larger than the others for the southern region (Table 5). The variance of the catch rates with both methods was usually much larger than the mean (sometimes by an or- der of magnitude), indicating that catches were aggregated (Table 5). The confidence intervals around the bycatch esti- mates were huge regardless of method because of the small number of observed fishing days and the large variability in catch rates (Table 4). 494 Fishery Bulletin 101(3) Table 5 Observed catch rates in weight and numhers for selected species obtained with different estimation methods from field data. Obser- vations in Pamlico Sound took place in July through October 1995 and observations in the Cape Fear River took place in August through October 1995. The CPUE-mean-per-unit estimator (CPUE=catch per unit of effort), which is based on the catch per day, uses day as a substitute for sample size. The basic F:S ratio estimator is the mean of the individual fish (F) to shrimp (S) ratios, and the grand F:S ratio estimator is the ratio of the mean catch offish to mean catch of shrimp. AC = Atlantic croaker, SP = spot, and WF = weakfish. See text for calculations. Equations for the standard deviations are taken from Cochran (1977), Equations 2.20 and 2.45. Observed catch rate by weight F:S ratio estimator Basic Grand CPUE-mean per-unit estimator kg fish/kg shrimp SD CV(%) kg fish/kg shrimp SD CV(%) species Avg. kg/day SD CV (%) Pamlico Sound AC 49.64 34.79 70 1.62 1.05 154 0.86 0.05 6 SP 41.57 52.40 126 1.46 2.21 152 0.72 0.17 16 WF 53.04 79.90 151 0.75 0.60 80 0.92 0.15 16 Cape Fear River AC 24.01 21.56 90 0.27 0.31 116 0.14 0.61 438 SP 4.14 6.71 162 0.05 0.06 129 0.02 0.18 757 WF 31.80 30.10 95 0.25 0.24 94 0.18 0.81 443 Observed catch rate by number CPUE-mean per-unit estimator F:S ratio estimator Basic Grand no. fish/kg shrimp SD CV(%) no. fish/kg shrimp SD CV(%) species Avg. no./day SD CV (%) Pamlico Sound AC 2418 1639 68 92.15 105.29 114 42.02 3.37 8 SP 1610 1885 117 72.56 121.92 168 27.97 4.69 17 WF 1014 1438 142 17.92 13.89 77 17.62 2.54 14 Cape Fear River AC 2287 2826 124 33.70 46.90 139 13.20 83.35 632 SP 257 464 181 3.05 4.16 136 1.48 12.10 816 WF 3198 3941 123 28.09 32.14 114 18.45 106.75 579 Bycatch simulations For the normally distributed data, the CPUE-mean-per- unit estimator was the only estimator whose estimated bycatch was not significantly different than the actual sim- ulated bycatch {% bias=0.006, P=0.94). All four of the ratio estimators significantly overestimated bycatch (Table 6), although the average bias was less than a 1% overestimate for the grand F:S and grand CPE ratio estimators. The basic F:S ratio estimator and the basic CPE ratio estimator both overestimated bycatch by 300-4007^ (Table 6). Using a model that included all main effects and all 2-way and 3- way interactions in the ANOVA, I found that the CV of the auxiliary variable (either shrimp catch or hours fished) was a significant main effect for all four of the ratio estimators, but there were no significant main effects for the CPUE mean-per-unit estimator (Table 7). Observer coverage was also a significant main effect for the F:S and CPE grand ratio estimators, but was not significant for the basic F: S or CPE ratio methods. The grand F:S ratio estimator, the grand CPE ratio estimator, and the basic F:S ratio estimator all showed several significant 2-way and 3-way interactions (Fig. 2), whereas the basic CPE ratio estimator had no significant 2-way or 3-way interactions. The CPUE- mean-per-unit estimator showed only two significant 3- way interactions among variables, and observer coverage occurred in both. The correlation between fish catches and shrimp catches was a significant main effect for the basic Diamond: Estimation of shrimp trawl bycatch 495 Table 6 Mean percent bias of each of the estimators with normal and delta lognormal distributions offish (F) and shrimp (S) from simulated data. Percent bias (Eq. 17) was calculated separately for each simulated fleet. N = the number of fleets in each analysis. The * indicates that the mean estimated bycatch is significantly different than the mean actual bycatch in a paired sample t-test (P<0.05). The CPUE-mean-per-unit estimator (CPUE=catch per unit of effort) uses unit effort as a proxy for sample size. The basic F;S ratio estimator is the mean of the individual fish to shrimp ratios, and the grand F:S ratio estimator is the ratio of the mean catch offish to the mean catch of shrimp. The basic CPE ratio estimator is the mean of the ratios of catch per effort, where effort is a variable measure such as hours fished, and the grand CPE estimator is the ratio of the mean catch offish to the mean estimate of effort. Estimator CPUE mean-per-unit Basic F;S ratio Grand F:S ratio Basic CPE ratio Grand CPE ratio Mean % bias Normal distribution TV =21,600 Delta lognormal distribution N= 118,810 0,006 427.80* 0.65* 336.13* 0.46* 0.09 9.98* 12.23* 30.75* 0.47* F:S estimator, and showed significant interactions with other parameters in both the grand F:S and grand CPE ratio estimators. For the delta lognormally distributed data, the CPUE- mean-per-unit estimator was the only estimator whose estimated bycatch was not significantly different than the actual simulated bycatch (% bias=0.087%, P=0,64), All four of the ratio estimators significantly overestimated bycatch (Table 6), with estimates ranging from a less than 1% overestimate using the grand CPE ratio estimator to a 30% overestimate with the basic CPE ratio estimator (Table 6). Using all 2-way and 3-way interactions in the ANOVA, I found that significant main effects for both the basic and grand F:S ratio estimators were the probability of catching shrimp and the CV of the shrimp catch. The CV of the fish catch and observer coverage were also main effects in the grand F:S ratio method. The probability of catching fish was an additional main effect in the basic F:S ratio method. The only significant main effect in both CPE ratio estimators was the CV of effort, and the only significant main effect in the CPUE-mean-per-unit method was the CV of the fish catch. All five methods exhibited several statistically significant 2-way and 3-way interactions (Table 7), Discussion The differences in bycatch estimates generated from the field data show how confusing bycatch estimation can be SV\n(^P \ 400 %' 300 ',, which is the catch of the bycatch species, and.*,, which is the auxiliary variable, is a straight line through the origin (indicating that the ratio of bycatch to shrimp or the catch of bycatch per hour fished is constant over all observations) and if the variance ofy, about this line is proportional to X,. In practice, these conditions rarely hold true. The ratio of fish to shrimp catches and the bycatch per hour fished from field data often vary considerably among observa- tions because of the patchy spatial distributions offish and shrimp, seasonal differences in the relative abundances of fish and shrimp, movements associated with development through different life stages, and environmental factors. In addition, the bias of a ratio estimator is on the order of 1/n, indicating that the bias will be small if « is large (Cochran, 1977). In practice, n, or the number of onboard bycatch observations, is often very small, particularly if the data are stratified by season or area, leading to large biases in ratio estimators. The Hartley-Ross ratio estimator, which is a form of the basic ratio estimator method, may in some cases be an unbiased or less biased ratio estimator for small samples (Cochran, 1977). However, the Hartley-Ross method was not effective for the field data in the present study, giving nonsensical negative estimates of bycatch for all species in the southern region, although the estimates in the northern region were generally (but not always) some- where between the basic and grand F;S ratio methods. The problems with the Hartley-Ross ratio estimator in the southern region may have been due to two factors: 1) the very low value for total shrimp landings from trip tickets in the southern region, and 2) discrepancies between the observed average catch of shrimp per day and the fleet shrimp catch per day from the trip ticket database. The Hartley-Ross equation starts with the mean of the indi- vidual fish to shrimp ratios expanded by the total shrimp landings (the basic F:S ratio estimator) and corrects the estimate based on the sampling fraction multiplied by a quantity that includes the average observed catch of shrimp per day (Eq. 8). The total shrimp landings recorded on trip tickets for the southern region were extremely low, about 16 times lower than the total shrimp landings in the northern region, although the number of days fished was about half as many in the southern region. In addition, the average shrimp catch per day on vessels that I observed in the southern region was much greater than the average reported on trip tickets (173.4 kg per day observed vs. 20.6 kg per day from trip tickets), whereas the average shrimp catch per day of shrimp that I observed in the northern region was much lower than the catch per day shown on trip tickets (55.6 kg per day observed vs. 173.5 kg per day from trip tickets). The result of this combination of factors was that the estimated total bycatch before correction in the southern region was very small due to the low amount of total shrimp landings, whereas the correction factor was very large because of the high observed average catch of shrimp, leading to negative estimates of total bycatch. These problems did not occur in the northern region. Low shrimp landings in the southern region compared to the northern region may have been due to an actual difference in the abundance of shrimp or differences in fishing habits such as a smaller number of nets per boat, tows per day, or tow times per tow in the southern region. However, it is also possible that more fishermen in the southern region than the northern region keep their catch or sell part of their catch independently without generating a trip ticket, which would reduce the total landings of shrimp in the trip ticket database. The differences in the average observed catch of shrimp per day were probably due to a combination of factors, most of them based on the prob- lem of nonrandom or nonrepresentative sampling of boats. Because I depended on volunteer fishermen, the observed shrimp boats and captains were not randomly selected. In addition, because no records are kept of the boat size, gear used, fishing habits, or effort history of fishermen in the fleet, sampled boats could not be compared to unsampled boats for these factors. However, most of the fishermen whose boats I observed in the Cape Fear River (the south- ern region) owned large boats and made an average of 5 tows per day, whereas the fishermen I observed in Pamlico Sound (the northern region) generally had smaller boats and made an average of 3.2 tows per day. If the fishermen whose boats I observed in the Cape Fear River fished more than the average number of tows per day and the observed fishermen in Pamlico Sound fished fewer than the aver- age number of tows per day, then the catch per day values would show these discrepancies. Other factors could have been differences between observed boats and average boats in the number of nets per boat, or tow times. All of the methods that I used for bycatch estimation for the field data were based on the summed catches over all tows on a single day, because in this study the variance of catches among tows within days was much less than the variance among days. The use of tows as the basic unit of effort would therefore have underestimated the total vari- ance. Sampling only day-trips probably contributed to the covariance among tows because tows spread over several days (and probably several locations) in a multiday trip would probably vary more among tows within a trip than tows in a single day. For randomly sampled multiday trips, estimation methods based on tows rather than days or trips may be preferred to those based on a trip as the unit of ef- fort because the sample size of tows increases faster than the sample size of days or trips, which would tighten the confidence intervals around the estimates. However, the use of tows as the unit of effort could be considered pseu- Diamond: Estimation of shrimp trawl bycatch 499 doreplication (Hurlbert, 1984) and could lead to erroneous variance estimates if the tows in a trip are not independent samples (Cochran, 1977). The choice of whether to use trips or tows as the unit of effort is dependent on two factors: 1) whether there is a high degree of covariance among tows in a trip, and 2) whether there is an independent estimate of the average number of tows per trip to use as an expan- sion factor. Confidence intervals around the bycatch estimates are not symmetrical, although they are shown in Table 4 as symmetrical to allow for easier comparisons between the methods in the field study. Because of the small numbers of observations, most of the confidence intervals in the field study were larger than the means, with the general exceptions of the grand ratio estimators for all species in the northern region, which were surprisingly small. Most grand ratio estimators underestimate the true catch rate and are positively skewed unless the sample size is greater than 30 and the CVs of both the observed fish catch and the auxiliary variable are less than 10% (Cochran, 1977). As seen in Table 5, CVs of the observed fish catch from field data are rarely as low as 10%, and many are over 100%. The very small confidence intervals for all species in the northern region, and the very large confidence intervals for all species in the southern region generated by the grand ratio estimators are due to the nonrandom sampling of boats for the average catch of shrimp in both areas. This nonrandom sampling affects the confidence intervals be- cause the average catch per day is a term in the denomi- nator of the equation used to estimate the variance of the grand ratio estimator (Eq. 2.45 in Cochran, 1977). A very large value for the average catch per day from trip tickets compared to the value from observations as in the north- ern region causes an underestimate in the variance and reduces the confidence intervals, whereas a small value for the average catch per day from trip tickets compared to the value from observations as in the southern region causes an overestimate in the variance and increases the confidence intervals. The field data shown here indicate some of the problems that are peculiar to observing and estimating bycatch in shrimp trawl fisheries in comparison to other fisheries. First, there are several hierarchical levels of variability that are ignored because of the logistical difficulties of sam- pling shrimp trawls. If the National Marine Fisheries Ser- vice (NMFS) protocol for shrimp trawl bycatch is followed, only one sample of the catch is taken from a net because of the large numbers of organisms caught in a typical tow. The NMFS protocol depends on the observer thoroughly mixing the catch so that a single sample characterizes the entire catch without variance, but mixing the catch to obtain a random sample is sometimes difficult because of the weight of the catch, the position of the culling tray, the size of the boat, or weather conditions. In addition, some species such as crabs may redistribute themselves after the catch is mixed by simply walking away. Stender and Barans (19941 found differences in fish-to-shrimp ratios when sampling the net compared with enumerating ev- erything in the net. However this source of variability is not measured when following the NMFS protocol and not included in the bycatch estimates. Second, only one net is generally sampled per tow, although the boat may tow two, four, or more nets. There is therefore an expansion from the sampled net to the number of nets per tow so that variance among nets is ignored, and this process also adds error. Third, the expansion term, regardless of whether the total shrimp landings or the total shrimp effort is used, is assumed known without error. To include the error in the expansion term further widens the confidence intervals around the final estimates (Diamond and Hanan'"). One of the most interesting findings from the simulations was that all the methods tended to overestimate bycatch. None of the overall bycatch estimates, and relatively few of the individual fleet simulations, generated underestimates of the actual values. Although the mean-per-unit and grand ratio estimates overestimated bycatch by less than 19f , if the bycatch is large enough, these estimators could errone- ously add hundreds of thousands offish to the catch-at-age matrices used in stock assessments. Inaccurate stock as- sessments could have consequences for the management of fisheries, particularly for species like red snapper that are managed by quotas on the directed fisheries that are based on the level of bycatch or that have target levels set for rebuilding fish stocks. One method that might be used to "correct" bycatch estimates for the mean-per-unit esti- mator would be to use the estimator to calculate the catch of the target species from the observations, and then to compare the estimated target species catch with the total landings. Although this correction method assumes that the total landings of the target species are accurate (which is rarely a valid assumption), comparison of the estimated total catch of the target species to the actual landed catch might help to pinpoint biases and to adjust the estimated bycatch. Because of the differences in estimates generated by the different methods of estimating bycatch, interpreta- tions of bycatch estimates and comparisons of bycatch studies should be made very cautiously. It is often difficult to tell in past studies whether estimates were generated by basic F:S or grand F:S methods, but basic F:S methods overestimate bycatch to a much greater degree. Because of the statistically significant 2-way and 3-way interactions among parameters, it is unlikely that estimates generated by one method can be converted or corrected to other meth- ods, so bycatch estimates made over time using different methods should not be directly compared. In addition, any bycatch estimate should include some indication of the variance around either the estimate or the catch rate, although variance estimates can be misleading if samples are not random. Finally, estimates of the weight or num- ber of species taken as bycatch, no matter how large or small, are meaningless without an estimate of population abundance. Small populations could be harmed by rela- tively small amounts of bycatch, whereas large populations '6 Diamond, S. L., and D. Hanan. 1986. An estimate of harbor porpoise mortality in California set net fisheries April 1, 1983 through March 31, 1984. National Marine Fisheries Service Admin. Report SWR-86-15, 40 p. 500 Fishery Bulletin 101(3) could be able to withstand even large amounts of bycatch. For this reason, the consequences of bycatch can only be evaluated if examined in conjunction with some estimate of stock size. Acknowledgments My sincerest gratitude goes to the fishermen who allowed me on their boats: Allan Hines, Bud George, Pete Dixon, Ben Ingraham, H.O. Golden, Tommy Peters, Al Gillikin, and Brad Styron. Bud George also provided many helpful suggestions on how to weigh the catch. I also appreciate the help given to me by Oliver and Tina Lewis, Bimbo Melton, Tony Cahoun, Gracie Golden, Jim Bahen, John Schoolfield, Beth Bums, Bob Hines, and Jim Murray. Trish Murphey, Mike Street, and Dee Lupton from NCDMF provided the shrimp trip ticket data. Peter Lamb, Tyler Stanton, Sue Zwicker, Pam Robinson, Walter Mayo, Amy Makepeace, Martin Gallagher, Dawn O'Harra, and Jim Armstrong helped to sort and identify the bycatch species. Jim Rice, Larry Crowder, Joe Hightower, Ken Pollock, and Doug Vaughan provided valuable input on earlier drafts of this manuscript. Thanks also to Rich Strauss and Richard Ste- vens for their help in using Matlab software. This manu- script was significantly improved by the comments of Scott Nichols and two anonymous reviewers. This research was supported by a National Science Foundation Pre-doctoral Fellowship, the J. Francis Allen Scholarship from the American Fisheries Society, the Joseph L. Fisher Disserta- tion Award from Resources for the Future, and MARFIN Grant no. NA57FF0299. Literature cited Alverson, D. L. 1994. 1994. A global assessment of fisheries bycatch and discards, 233 p. FAO (Food and Agriculture Organization) of the United Nations, Rome, Italy. [ISBN 92-5-103555-5.] Cochran, W. G. 1977. Sampling techniques, 428 p. John Wiley and Sons, New York, NY. Grouse, D. T., L. B. Crowder, and H. Caswell. 1987. A stage-based population model for loggerhead sea turtles and implications for conservation. Ecology 68(5): 1412-1423. Fahy, W. E. 1966. Species composition of the North Carolina industrial fish fishery. Comm. Fish. Rev. 28(7):l-8. Fisheries Statistics of the United States. 1977. U.S. Fish and Wildlife Service, Bureau of Commercial Fisheries, Statistical Digest 71, 407 p. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field studies. Ecol. Monogr. 54 (2):187-211. Julian, F, and M. Beeson. 1998. Estimates of marine mammal, turtle, and seabird mortality for two California gillnet fisheries: 1990-1995. Fish Bull. 96:271-284. Stender, B. W., and C. A. Barans. 1994. Comparison of the catch from tongue and two-seam shrimp nets off South Carolina. North Am. J. Fish. Manage. 14:178-195. Walsh, W.A., and P. Kleiber 2001. Generalized additive model and regression tree analysis of blue shark iPrionace glauca) catch rates by the Hawaii-based commercial longline fishery. Fish. Res. 52(2):115-131. 501 Abstract — Adaptive cluster sampling (ACS) has been the subject of many publications about sampling aggregated populations. Choosing the criterion value that invokes ACS remains prob- lematic. We address this problem using data from a June 1999 ACS survey for rockfish, specifically for Pacific ocean perch {Sebastes aliitus). and for shortraker (S. borealis) and rougheye (S. aleutianus) rockfish combined. Our hypotheses were that ACS would out- perform simple random sampling ( SRS ) for S. aliitiis and would be more appli- cable for S. alutiis than for S. borealis and S. aleutianus combined because populations of S. alutus are thought to be more aggregated. Three alterna- tives for choosing a criterion value were investigated. We chose the strategy that yielded the lowest criterion value and simulated the higher criterion values with the data after the survey. System- atic random sampling was conducted across the whole area to determine the lowest criterion value, and then a new systematic random sample was taken with adaptive sampling around each tow that exceeded the fixed criterion value. ACS jaelded gains in precision (SE) over SRS. Bootstrapping showed that the distribution of an ACS estima- tor is approximately normal, whereas the SRS sampling distribution is skewed and bimodal. Simulation showed that a higher criterion value results in substantially less adaptive sampling with little tradeoff in preci- sion. When time-efficiency was exam- ined, ACS quickly added more samples, but sampling edge units caused this efficiency to be lessened, and the gain in efficiency did not measurably affect our conclusions. ACS for S. alutus should be incorporated with a fixed criterion value equal to the top quartile of previ- ously collected survey data. The second hypothesis was confirmed because ACS did not prove to be more effective for S. borealis-S. aleutianus. Overall, our ACS results were not as optimistic as those previously published in the literature, and indicate the need for further study of this sampling method. Applications in adaptive cluster sampling of Gulf of Alaska rockfish Dana H. Hanselman Terrance J. Quinn II School of Fisheries and Ocean Sciences University of Alaska Fairbanks 11275 Glacier Hwy. Juneau, Alaska 99801 E-mail address (for D. H. Hanselman); ftdhh@uaf.edu Chris Lunsford Jonathan Helfetz David Clausen Auke Bay Laboratory Alaska Fisheries Science Center National Manne Fishenes Service 11305 Glacier Hwy. Juneau, Alaska 99801 Manuscript approved for publication •30 January 2003 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:501-.513 (20031. In nature, populations are sometimes distributed in a patchy, rare, or aggre- gated manner. Conventional sampling designs such as simple random sam- pling (SRS) do not take advantage of this spatial differentiation. Thompson (1990) introduced a sampling design called adaptive cluster sampling (ACS) to survey these types of distributions. Adaptive cluster sampling, in theory, can be much more precise for a given amount of effort than conventional sampling designs (Thompson, 1990). In practice, however, this is not always the case. In some cases, the variance is greatly reduced, but bias is induced from stopping rules and criterion values that are sometimes changed mid-survey (Lo et al., 1997). In 1998, we conducted a survey on Gulf of Alaska rockfish in which ACS was efficient and successful, but the gains in precision, if any, were small compared to those of a SRS of the same size (Quinn et al., 1999; Hansel- man et al., 2001). Recently papers about ACS have in- cluded efficiency comparisons (Christ- man, 1997), restricted ACSs (Lo et al., 1997; Brown and Manly, 1998), boot- strap confidence intervals (Christman and Pontius, 2000), and bias estimates (Su and Quinn, 2003). However, little work has been done on determining the criterion value that, when exceeded, invokes additional sampling. In the fol- lowing study, we examine the details for choosing this criterion value by using data from a 1999 field survey for Gulf of Alaska rockfish. We then simulate the outcome of the experiment with dif- ferent criterion values after the survey. We also compare the efficiency of ACS to SRS. In the basic adaptive cluster sam- pling (ACS) design, a simple random sample (SRS) of size n is taken; if _y (the variable of interest) exceeds c (a criterion value), then neighborhood units are added (e.g. units above, be- low, left, and right in a cross pattern. Fig. 1) to the sample. These are called network units. If any network unit has y>c, then its neighborhood is added. Units that do not exceed the criterion are called edge units, and sampling does not continue around them. This process continues until no units are added or until the boundary of the area is reached (Thompson and Seber, 1996). Neighborhoods can be defined in any general way. The only condition is that if unit i is in the neighborhood of^, then unit j is in the neighborhood of i. The "unbiasedness" of the estimators relies on all neighborhood units of >'>c being sampled. If logistics cause the sampling to be curtailed before the sampling is complete, then biased estimators can 502 Fishery Bulletin 101(3) Cross Pattern A3 A3 A2 A3 A3 A2 Al A 2 A3 A3 A 2 Al R Al A 2 A3 A3 A 2 Al A 2 A3 A3 A 2 A3 A3 Line ar Pattern A4 A3 A2 Al R Al A2 A3 A4 Figure 1 Diagram of a basic cluster sampling design, showing the maximum possible number of adaptive hauls for the cross (S=3) and the linear (S=4) patterns with the imposition of a stopping rule. The initial random tow is denoted as "R," anc the adaptive tows as "A" and their respective level number result. For our study, all samples were called "tows" because our study was a trawl survey. When little information is available to preset a fixed criterion value, order statistics are often used to choose a criterion value (Thompson and Seber, 1996). The basic idea is that an initial random sample is conducted. Next, the values of the random tows are ordered, and ACS is conducted around the top r stations. The variable r is de- cided by the experimenter and depends on the amount of resources available and the suspected aggregation of the population. The criterion value is then set at the value of the next highest tow ( ;+ 1 ). This was the design used in the 1998 adaptive cluster sampling survey for rockfish (Quinn et al., 1999, Hanselman et al., 2001). The use of order statistics has several limitations, however. First, initial random samples must be taken before the adaptive phase can begin. This procedure can be inefficient, because the experiment may have to move a large distance back to the previous tows that exceeded the criterion, by which time the aggregation may have moved or dispersed. In some cases, this procedure may result in a very small criterion value that leads to an overwhelming amount of adaptive sampling around some tows. Second, the process of achiev- ing simple unbiased estimates of abundance is more com- plicated with order statistics because the criterion value is dependent on the sampling. In our study, we address methods to avoid these limita- tions and illustrate these methods with a 1999 ACS survey for Gulf of Alaska rockfish. The primary target of the sur- vey was Pacific ocean perch iSebastes alutus [POP] ). These fish have extremely uncertain biomass estimates in the Gulf of Alaska (Heifetz et al.'). The estimates are based in part on a standardized stratified random survey conducted by the National Marine Fisheries Service every three years (every two since 2000). This uncertainty is likely due to their highly clustered distribution (Lunsford, 1999) and has led to two independent surveys ( 1998, 1999) to test the benefits of ACS in sampling POP. Shortraker (S. borealis) and rougheye (S. aleutianus) rockfish combined (SR-RE) are also tested to compare the results of a population that is considered highly clustered (POP) versus one that is considered more uniformly distributed (SR-RE). SR-RE are combined because they co-occur in identical habitat and are managed as a complex. Materials and methods In June 1999, ACS was carried out between 140° and 144° west longitude near Yakutat in the Gulf of Alaska (Fig. 2). Approximately 75% of sampling was directed toward the POP depth stratum ( 180-300 m) and 25% directed toward SR-RE depths (300^50 m). A 182-ft. factory trawler, the Unimak, was chartered to conduct trawl samples. Fish- ing and field operations are described in Clausen et al.^ Duration of all trawl hauls was 15 (POP) and 30 (SR-RE) minutes on the bottom. SR-RE tows were made parallel to the depth contours in a linear pattern (Fig. 1) because the slope that SR-RE inhabit is too steep for perpendicular tows. Travel time between all tows was recorded to exam- ine time efficiency. Initially, a set of systematic random tows was conducted from west to east across the entire study area to determine the criterion value. Samples were chosen systematically by longitude and distributed randomly by depth within each longitudinal strip. This procedure was a necessary proxy for simple random sampling because of poorly known bathym- etry in the area. The use of simple random latitudes and longitudes often results in the selection of sites that are well out of the sampling depth interval. After random sampling was completed, we compiled and examined the data to set the criterion value. Criterion values were chosen based on a hierarchy of three alternatives described below. Next, we conducted a new set of random tows from east to west across the area, in which any tows exceeding the criterion value were adaptively sampled. A distance of 0.19 km (0.1 nmi) was used between all adaptive tows and the initial random tow to avoid depletion effects on the catches. ' Heifetz, J., D. L. Courtney, D. M. Clausen, J. T. Fujioka, and J. N. lanclli. 2001. Slope rockfish. In Stock as.scssment and fish- ery evaluation for the groundfish resources of the Gulf of Alaska, 72 p. North Pacific Fishery Management Council. 605 W. 4"' Ave. Suite 306, Anchorage, AK 99501. 2 Clausen, D.M.,D.H. Hanselman, C. Lunsford, T. Quinn II, and J. Heifetz. 1999. fZ/i/niaA' enterprise cruise 98-01 rockfish adap- tive sampling experiment in the central Gulf of Alaska 1998, 49 p. Auke Bav Lab, NMFS, NOAA, 1 1305 Glacier Hwy, Auke Bay, Alaska, 99801. Hanselman et al : Applications in adaptive cluster sampling of Gulf of Alaska rockfisfi 503 145°W 144"W I43°W Ma-w 14 I '-'W 14()°W bU N 59°N Figure 2 Map of sampling area in the Gulf of Alaska on the Uniinak 99-01 adaptive sampling cruise. "R" symbols are the initial random tows for the criterion phase, "r" symbols are random stations in the survey phase, "A" symbols are adaptive cluster samples. Three methods were formulated for determining a fixed criterion value c of POP catch-per-unit-of-effort ( CPUE). ( 1 ) We combined and calibrated past survey and fishing data to provide the anticipated distribution of CPUE in the 1999 survey. Then we calculated the SO"^*^ percentile of that dis- tribution as the criterion value. Our rationale was that this value would correspond to that obtained from order statis- tics. (Three networks were sampled in 1998; therefore the criterion value was set to the 4'*^ highest of the ordered 15 initial tows, which corresponded approximately to the 80'^ percentile. ) ( 2 ) We used the mean CPUE of past survey and fishery data because when we compared the 80"^ percentile criterion against the 1998 ACS survey's data, the sampling would have resulted in primarily edge units. (3) After a representative random sample was taken across the entire area in 1999, we would use the initial mean CPUE for the criterion value for the return trip. The rationale for using mean CPUE above is that in an aggregated population, the majority of the tows would be less than the mean. The actual values of the criterion chosen under each alternative are described in the results. We chose the SR-RE criterion to be the mean CPUE of initial tows. We assumed this was a reasonable criterion value because if the population of SR-RE were somewhat uniform, a lower value would result in too much ACS, but mean CPUE would still be low enough to allow higher cri- terion values to be examined. Although we concentrated on evaluating criterion alternatives for POP, we present the SR-RE data to illustrate that different levels of aggregation could affect how much can be gained with ACS in terms of precision and efficiency. A major problem in applying adaptive sampling is that sampling may continue indefinitely because of a low crite- rion value. To limit the amount of adaptive sampling, an arbitrary stopping rule of S levels was imposed. For those strata where the cross pattern of adaptive sampUng was used (POP), the stopping rule was S = 3 levels, allowing for a maximum of 24 adaptive tows around each high-CPUE random tow (Fig. 1). For the strata with the hnear pattern of adaptive sampling (SR-RE), the stopping rule was S = 4 levels, for a maximum of eight adaptive tows around each high-CPUE random tow. This stopping rule differs from that of the previous year in which we used a stopping rule of six because we believed that the possible 30-km difference between the ends of the networks was too large for efficient sampling (Clausen^). In addition, no adaptive sampling ex- tended beyond a stratum boundary. The result of adaptive sampling around each high-CPUE tow was a network of tows that extended over and, in some cases, delineated the geographic boundaries of a rockfish aggregation. 504 Fishery Bulletin 101(3) Statistical analysis of the results was based on adap- tive cluster sampling (Thompson and Seber, 1996). First, we estimated the abundance (kg/km) for the targeted rockfish species from the n initial random tows using the standard simple random sampling (SRS) estimator Then, two adaptive estimators of abundance, a Hansen-Hurwitz estimator (HH) and a Horvitz-Thompson estimator (HT), were calculated. We computed standard error (SE) as a measure of precision. The unbiased HH estimator for the ACS mean is 1 " T "_ (1) where w^ and y* = the mean and total (respectively) of the X, observations in the network that intersects sample unit /. The HH estimator essentially replaces tows around which adaptive sampling occurred with the mean of the network of adaptive tows that exceeded the criterion CPUE. The unbiased HT estimator for the ACS mean is 1 ^ •' A'-f-i'or, (2) where y^ = the sum of the y-values for the kth network; K = the number of distinct networks in a sample; rtf. = the probability that network k is included in the sample; and N = the total number of sampling units. If there are .r^ units in the kth. network, then N-.v, (3) where N = the total number of sampling units; n = the initial random sample; and X)^ = the number of units in the network. The HT estimator is based on the probability of sampling a network given the initial tows sampled and involves the number of distinct networks sampled (in contrast to the HH estimator which is based only on the initial tows). The HT estimator often outperforms other estimators as seen in simulation studies (Su and Quinn, 2003). Both estima- tors use the network samples and initial random samples, but not the edge units. This sample size is referred to as v' (convention established by Thompson (1990) and used in Thompson and Seber (1996)). To include edge units into the estimates Thompson and Seber (1996) and Salehi (1999) used the Kao-Blackwcll theorem, which is a com- plex method that could theoretically result in more precise estimates. However, it had little effect for the 1998 survey data (<1% improvement, Hanselman, 2000); therefore these calculations were not used in our study. When a stopping rule is used, the theoretical basis for the adaptive sampling design changes. It may result in incomplete networks that overlap and are not fixed in rela- tion to a specified criterion — changing with the pattern of the population. In contrast, the nonstopping-rule scheme has disjoint networks that form a unique partition of the population for a specified criterion. This partitioning is the theoretical basis for the unbiasedness of pi^^ and fifjj. Thus with a stopping rule, some bias may be introduced. Recent simulation studies (Su and Quinn, 2003) have estimated the bias induced by using a stopping rule on each estimator with order statistics, but not with a fixed crite- rion. Because the use of a fixed criterion is design unbiased, its estimate should be less biased by the stopping rule than a sample with order statistics. Therefore, we can use the Su-Quinn simulation results to approximate the maximum bias induced by the stopping rule. With a stopping rule of three and the HH estimator, the maximum positive bias is 17% for a highly aggregated simulated population. With a stopping rule of three and the HT estimator, the maxi- mum bias is approximately 12%. Considering our design, we accepted the tradeoff of relatively small bias for gains in precision and logistical efficiency. Additionally, nonparametric bootstrap methods were adapted from Christman and Pontius (2000) and we used the HH version of the estimates to examine bias from our survey. Five thousand resamples were performed by using n for the SRS bootstrap, and the sample size from the origi- nal criterion value of 220 kg/km ( v) was used for the ACS bootstrap. Bootstrap distributions of the data were exam- ined for SRS and ACS designs to examine the capability of each design to clearly demonstrate a central tendency. We evaluated two hypotheses: 1) Adaptive sampling would be more effective in providing precise estimates of POP biomass than would a simple random survey design; and 2) Assessment of POP abundance would benefit more from an adaptive sampling design than would SR-RE be- cause POP are believed to be more clustered in their dis- tribution than SR-RE. SRS estimates were obtained from the initial random tows, and variance estimates were cal- culated for the initial sample size in ) and for the equivalent sample size that included the adaptive tows but not the edge units (\''). This procedure makes the theoretical com- parison fair because each estimate is based on the same number of samples. Total sample size including edge units (v) was not used in the theoretical precision comparison but was considered when efficiency issues were examined later These hypotheses were assessed by comparing the standard errors (SEs) of ACS to those of SRS. Substantial reductions in SE with ACS for POP would support the first hypothesis, whereas no reductions of SE using ACS for SR-RE would support the second hypothesis. This com- parison is qualitative because relevant significance tests are unavailable and the two methods are different in terms of efficiency. To evaluate different alternatives and criterion values, each network was reconstructed as if the higher criterion values had been used in the field. We also examined the tradeoff between amounts of additional sampling com- pared with the gains in precision. A comparison was made oftiu' SRS results by using sample sizes constructed with the number of possible samples with the time-per-sample Hanselman et a\: Applications in adaptive cluster sampling of Gulf of Alaska rockfish 505 Table 1 Data used to determine criterion values c for the 1999 adaptive cluster sampling (ACS) survey. Data from a 1998 ACS survey from a different area is divided by the National Marine Fisheries Service triennial survey data and fishery data from the same area to obtain gear efficiency values. The mean of these gear efficiencies are then multiplied against triennial and fishery data from the new area to yield gear-calibrated CPUEs for the new area. Only numbers in bold were used in calculations, n = the number of observations of that data set; 80% = the 80"^ percentile catch of that data set. Data source Year Mean CPUE (kg/km) 80% ACS results from different area and year (divided by) CPUEs of corresponding previous area from triennial and fishery data (equals) Gear efficiency of the Unimak (multiplied by) Prior CPUE data from area for 1999 ACS survey (equals) Calibrated CPUE data for 1999 ACS survey Criterion value c 1998 284.94 Triennial 1993 38.36 1996 46.64 1993-96 42.54 Fishery 1996-98 30.64 1993 7.44 1996 6.12 1993-96 6.71 1996-98 9.32 Mean 7.63 Triennial 1993 40.32 1996 26.50 1993-96 33.92 Fishery 1996-98 19.61 Triennial 1993 307.52 1996 202.06 1993-96 258.69 Fishery 1996-98 149.57 Mean 219.71 223.92 57 7.89 50 27.33 51 18.79 101 14.03 434 28.18 8.14 11.84 15.85 17.39 X 46.74 29 33.50 25 38.85 54 30.47 190 812.67 29 582.52 25 675.63 54 529.90 137 641.69 data we collected. In this comparison we used three new sample sizes: 1) v,, the number of samples that could have been taken in the same amount of time as that for a SRS if sampling time for edge units was neghgible; 2) v^, in which the edge units had taken the same amount of time as non-edge units; and 3) v^, in which the average distance between each tow type was used as effort instead of time (with edge units included). survey data (1993, 1996) and fishery data (1996-98) from the observer program for the same area. This gear coef- ficient was then multiplied by the same data for the new area to establish the expected catches. The data used and the calculations are shown in Table 1. To implement alter- native 3, we conducted 13 initial POP and 10 initial SR-RE random tows across the entire area. Catches from these initial tows gave us the following results for each criterion alternative: Results Alternative 1 Formulation of criterion alternatives A total of 164 tows were conducted for the ACS experiment. Nearly all tows were made successfully; only a few excep- tions were deemed untrawlable and moved to the nearest Alternative 2 trawlable bottom. We determined the POP criterion value for alternatives 1 and 2 (see below) before the survey by looking at the 1998 ACS results from a different geographic Alternative 3 area, as well as prior survey and fishery data in our study area. We obtained the criterion value by calculating a gear efficiency coefficient for the 1998 sui-vey by using NMFS For alternative 1, the mean of the 80"' per- centile of the data from Table 1 is 641.69 kg/km. We rounded this downward to c = 540 kg/km ( 1000 kg/nmi) for ease of opera- tion in the field (the design was originally in kg/nmi units). The mean calibrated CPUE for the area from Table 1 yielded a criterion value c of 220 kg/km (rounded). In this alternative, the mean CPUEs from the initial sample in 1999 yielded criterion values of c = 250 kg/km for POP and c = 418 kg/km for SR-RE. 506 Fishery Bulletin 101 (3) Table 2 Summary of density estimates (/J) and standard errors (SE) for the 1999 adaptive cluster sampling experiment for the Sebastes atutus and the S. borealis-S. aleutianus complex, c is the criterion value, r is the number of adaptive networks, n is the initial sample size, v' is the adaptive sampling size (excluding edge units). SRS = simple random sampling estimator, HH = Hansen-Hur- witz adaptive estimator, and HT = Horvitz-Thompson adaptive estimator. Alt. = criterion alternative. Sebastes alutus Sebastes borealis and S. aleutianus Alt. 2 Alt. 3 Alt. 1 — Alt. 3 — c (kg/km) >220 >250 >540 >1080 >418 >540 r 6 6 5 3 5 3 n 25 25 25 25 9 9 v' 74 73 55 48 30 14 ''sRS 904 904 904 904 447 447 SE„ 496 496 496 496 115 115 se; 288 290 334 358 63 92 ^'HH 498 501 566 526 511 486 SE 166 167 192 197 128 141 i'HT 471 472 567 527 511 486 SE 167 167 192 197 128 141 The second phase of the experiment began with random tows in an east to west direction. Complete location and CPUE data for both species are located in Appendix I. In order to analyze all alternatives, the lowest alternative was used in the field for adaptive sampling during the second phase, which resulted in the 220 kg/km criterion value for POP from alternative 2. For SR-RE, the criterion value was the mean CPUE of 418 kg/km from alternative 3. The remaining alternatives were simulated following the completion of the survey. POP results After the initial tows, 25 random tows were selected for the return trip across the area. All 25 were completed, of which six became networks of more than one unit. A total of 106 tows were completed in the POP stratum. At one of the tows that exceeded the criterion value, the captain deemed that further adaptive sampling was not feasible because of the presence of coral. Of the six networks, two overlapped, resulting in five distinct networks. In these networks, 81 adaptive samples were taken, of which 49 exceeded the criterion and 32 did not and were therefore edge units and not included in the sample estimates. We compared the results of the original adaptive sample ( alternative 2 ) with the simulated results of higher criterion values (Table 2). The precision of simple random sample estimates with both n (number of random samples) and v' (number of random samples plus the number of adaptive network samples, not edge units) was contra.sted with that of the adaptive estimators described above. As the criterion value increased, n remained the same, whereas v' and r (the number of networks) decreased. At the 220 kg/km criterion value (alt. 2), there were substantial reductions in SE over the SRS estimators by using ACS estimators for both the ;; and v' sample sizes. The 250 kg/km criterion value (alt. 3) resulted in a nearly identical sample to that of the 220 kg/ km (alt. 2) criterion value and the loss of only one network sample. Hence, the estimates were nearly identical. The HT mean estimates were slightly lower than the HH estimates for the two lowest criterion values (alts. 2 and 3) because two networks overlapped. These networks became separate at the next higher criterion value, which aligned the estima- tors. The next highest criterion value of 540 kg/km (alt. 1) showed that even though the sample size was reduced by 19 tows from the original criterion value, the ACS estimators performed nearly as well, yielding just slightly larger SEs. When the criterion was arbitrarily doubled to 1080 kg/km, the sample size was further reduced by seven, and had similar SEs to the 540 kg/km criterion value. The SRS and ACS bootstraps for POP resulted in very different distributions. Five thousand replications showed that the SRS distribution was bimodal and right skewed (Fig. 3). The SRS mean fell on the second mode, which is more than twice the ACS mean. This bimodal distribution is driven by the presence of the very large random catch (tow no. 60). If that haul is present in a bootstrap repli- cate, then the SRS estimate tends to be high, leading to the second mode in the bootstrap distribution. The ACS boot- strap distribution was symmetric and closely resembled a normal distribution (Fig. 3). The average estimates of bias showed that the bias of HH was +4^( and the bias of HT was -I7c. The standard error had an estimated bias of -t-S'/c for HH and HT The results from this POP study and the previous 1998 study were both greatly affected by one or two very large catches, as we expected for a highly clustered population. Of interest is what happened when the largest catch was changed to a nominal catch that still exceeded the criterion value. Appendix II shows the results of changing haul no. 60 from 12,000 kg/km to 540 kg/km. In the comparison at v', SRS outperforms ACS in terms of SE. However, it also Hanselman et al : Applications in adaptive cluster sampling of Gulf of Alaska rockfish 507 Table 3 Comparisons of time per travel (TPT) and time per sample (TPS) of adaptive sampling against simple random sampling for Pacific ocean perch (S. alutus) and for shortraker tSehastes borealis) and rougheye (S. aleutianus) rockfish combined, on a 1999 adaptive sampling cruise. TPT is the travel time between tows in hours; TPS is the travel time plus haul time in hours. "Distance between" is the average travel distance (km) between two adaptive stations and between two random stations. "Adjusted distance" is the distance if the random sample size was increased to 106. S. alutus S. borealis and S. aleutianus Random Adaptive Random Adaptive Time(h) 10.4 11.4 4.4 12.0 No. of hauls 23 72 9 24 TPT 0.45 0.16 0.49 0.50 TPS 0.95 0.66 1.49 1.50 Distance between 20.2 3.22 Adjusted distance 4.73 3.22 shows that the mean of ACS is stable because it changes little by removing a high catch, whereas the SRS mean is reduced by half. SR-RE results At every third POP random tow, a tow was made in the SR-RE depth stratum. A total of 35 tows were made in the SR-RE stratum. Nine random tows jrielded five distinct networks with 21 network tows and five edge units. The stopping rule was invoked for three of the five networks. At the mean CPUE criterion (418 kg/km, alt. 3), the adaptive estimators performed approximately the same in terms of SE compared to the SRS esti- mator using 71 (Table 2). With v', the SRS estimator yielded a lower SE than both adaptive estimators. When the criterion value increased to an arbitrarily higher value (540 kg/km), the adaptive estimators performed worse than SRS estimates for both /; and v'. Time efficiency We recorded and compared travel time between adaptive tows and simple random tows for 149 of the tows (Table 31. Not all the tows were used because of mechanical failure or because the factory capacity was reached. In the survey, 38 hours out of 10 days were spent in transit between sampling tows, which for a short survey was a substantial amount of the available time. For POP, substantial gains in travel-time efficiency were achieved with ACS. Average travel time for simple random tows (0.45 h) was nearly triple that of adaptive tows (0.16 h) for POP, which indicated that ACS can maximize sampling tows for POP when time is limited. In the SR-RE sampling, travel time for adaptive sampling (0.5 h) was about the same as simple random sampling (0.49 h), which was due to long linear samples that are not as close together as POP tows (Fig. 1). Also, determina- 500 1000 1500 2000 Mean abundance 2500 3000 Figure 3 Bootstrap distributions for the 1999 adaptive sampling survey (25,000 replicates). Dotted line is the sampling estimate of mean abundance (kg/km) from the survey. Top graph is the distribution of mean abundance estimates for simple random sampling. Bottom graph is the distribution of mean abundance estimates for adaptive cluster sampling (obtained with the Hansen-Hurwitz estimator). tion of CPUE required processing of the catch, which took various amounts of time after the completion of the tow. Because of this delay, we went to the opposite tow on the other side of the random tow when sampling SR-RE with the linear pattern, whereas there were many nearby tows when sampling POP with the cross pattern. The travel time was added to the average tow time irom gear deployment to full retrieval of 0.5 h for POP and 1.0 h for SR-RE to obtain total sampling time (per sample). Travel time was reduced by 31% with adaptive sampling (0.66 h/sample) in relation to simple random sampling 508 Fishery Bulletin 101(3) (0.95 h/sample) for POP. Sampling time efficiency for SR- RE was approximately the same for adaptive sampling (1.5 h/sample) and simple random sampling (1.49 h/sample) for SR-RE. These results are confounded by the fact that the random tows are spread apart because of the lesser effort applied to them. The average distance between random tows (20.2 km) was adjusted to a distance of 4.73 km as if there were 106 random tows distributed throughout the area. This distance is still larger than the average distance between tows in adaptive sampling (3.22 km). From these time and distance data, we re-estimated the precision of SRS under three new sample sizes in order to further compare the relative efficiency of ACS. We denoted the sample size that could have been taken under SRS, using the same amount of time as was used during the adaptive sampling including edge units, as v.. An alternative sample size V, was the equivalent SRS sample size if the amount of time to sample edge units in ACS was negligible. This sta- tistic would be useful if edge units could be determined (i.e. hydroacoustically or visually [presence or absence] ) without actually trawling them. A third alternative was to find the equivalent SRS sample size v^^ that would result from apply- ing the total distance traveled in the ACS design on random stations instead. For \'^., more random POP samples would have been taken than were included in the adaptive estima- tors (Table 4). The SEs of ACS were still much lower across all criterion values (Table 2). When we used v, (Table 4), SRS was much less precise than ACS (Table 2). Finally, when we used distance instead of time (\'^), the results were almost exactly the same as those for \\, (Table 4). Discussion Our two hypotheses were that ACS would be more precise than SRS for POP and no more precise for SR-RE com- bined. The results from the 1999 field study showed that the SEs for the adaptive POP estimates were smaller than both SRS estimates, with n and \'', and thus support the first hjrpothesis. One curious result is that in both 1998 and 1999, the SRS estimate of density was substantially larger than the ACS estimate, even though, on average, they were both essentially unbiased. We attributed this curiosity to the more variable and skewed SRS distribution in which large sampling error on the high side is possible more often than in the ACS estimation. Of course we fully expected that both estimates would average to be the same value if the experiment could be repeated many times. ACS reduced the influence of one large CPUE in the relatively small initial sample, as illustrated by the symmetric and near-normal shape of the ACS bootstrap distribution. Con- sequently, we concluded that ACS is a more robust estima- tor of density than SRS for aggregated populations. One caveat is that the precision of the estimates, if measured in terms of coefficient of variation, is similar between the two methods because of the nuich larger moan estimate for the SRS estimate. Monte Carlo simulations would be useful to examine the properties of the estimators under different criterion values and population densities along the lines of Su and Quinn (2003). Table 4 Comparison of simple random sampling (SRS) precision estimates with the inclusion of time and distance informa- tion, c is the criterion value, v' is the original adaptive clus- ter sampling adjusted sample size, v,. is the time adjusted sample size, including edge units, v, is the time adjusted sample size with edge unit cost set to zero, v^^ is the dis- | tance-adjusted sample size including edge units ii is the mean SRS density estimate, SE is the standard error for that sample size. c (kg/km) >220 >250 >540 >1080 A' 904 904 904 904 v' 74 73 55 48 SE 294 296 341 365 y, 81 80 67 55 SE 281 283 309 341 ^', 59 58 46 41 SE 329 332 373 395 \;, 80 79 67 54 SE 283 285 309 344 The SR-RE adaptive estimates all have higher SEs than the SRS estimates, and this finding supports the second hy- pothesis. More than twice as many samples were directed toward POP than SR-RE, yet the POP density estimates are much more variable than those for SR-RE. This much larger variability for POP was indicative of the clustering that we expected. This experiment showed that for POP, ACS with a fixed criterion has some distinct advantages over simple random sampling and over adaptive cluster sampling with order statistics, which was used in the previous 1998 survey. Lower SEs were obtained, at one third less effort than if we just added an equivalent number of random samples. Sampling over a broader area yielded better results than the tightly stratified 1998 design. Our study also assumed stationary aggregations of fish. This assumption may have been better satisfied with a fixed criterion because the adaptive sampling was conducted immediately after a sample exceeded the criterion value. Although the fixed criterion eliminates bias induced by a variable criterion value, we still used stopping rules. If bootstrapping is a good indicator of bias, then the bias in- duced by stopping rules is negligible. Additionally, we have shown that a relatively high criterion value could be used to help minimize the use of these stopping rules. Our study showed that ACS is a fast and efficient way to gain a large number of samples. However, if edge units do not contribute to a better estimate and they have a sim- ilar cost or time expense as included samples, then little is gained. This deficiency shows the need for some method of determining edge units without actually sampling them. In fisheries surveys, this use might be a double sampling design with hydroacoustics as an auxiliary variable Hanselman et al : Applications in adaptive cluster sampling of Gulf of Alaska rockfisfi 509 (Fujioka^) or a design called TAPAS that hydroacoustically delineates clusters ( Everson et al., 1996). In other surveys, it might be possible to detect the presence of the item of interest without actually surveying the unit (as in aerial surveys. ) An ACS design should not be attempted without some prior knowledge of the population distribution. Populations for which the design would be useful should have an aggre- gated distribution that can be described by correlated varia- tion with distance, not just a large variance in relation to the mean. One way to examine the data is to fit variograms to examine spatial autocorrelation (Hanselman et al., 2001). If no prior data exist, it would not make sense to attempt ACS as an initial sampling design. We have shown that a wide range of criterion values can be used without considerable differences in the results. Therefore, only enough prior data are needed so that an adequate range of population density can be estimated. If the criterion value chosen resulted in too many or too few samples, the criterion could be adjusted, and then the design stratified into two different areas. Most commercial fish species have survey data that can be used to determine a fixed criterion. If possible, criterion val- ues should be determined prior to the survey, so that maxi- mum efficiency can be attained. We have shown that it may be appropriate to choose a relatively high sampling criterion such as the 80'^ percentile of past CPUE without sacrificing estimation capabilities. This high sampling criterion has sev- eral practical advantages. First, the design is attractive for commercial boats to perform the adaptive phase at no-cost because only large catches are sampled. The current design does not use the fish sampled during the survey, which, in the case of deepwater rockfish, would cause certain mortality. Under an adaptive design, a commercial boat would take the larger catches and could put them to use. Second, fewer over- all networks would be sampled because the higher criterion would evoke less adaptive sampling, which may mean less overall sajnpling in the survey. Finally, precision would be gained at a minimal cost and effort. Stopping rules would be unnecessary, ensuring an unbiased estimate. However, clus- ter sampling is most effective when the cluster samples are as heterogeneous as possible. Therefore, caution is required not to set the criterion too high, or the resulting clusters will be either too homogeneous or contain only edge units, leading to no improvement in the estimators. Similarly, if there are large changes in density from year to year, a fixed criterion may not be appropriate. In conclusion, adaptive cluster sampling is appropriate for surveys of highly clus- tered species with low temporal fluctuations, for which a fixed criterion can be determined beforehand. Acknowledgments We thank the crew of the FV Unimak, in particular Cap- tain Paul Ison and Production Manager Rob Elzig, for their 3 Fujioka, J. 2001. Unpubl. manuscr. Using hydroacoustics and double sampling to improve rockfish abundance estimation, 8 p. Auke Bay Laboratory, National Marine Fisheries Service, NOAA, 11305 Glacier Hwy, Auke Bay, AK 99801. excellent cooperation in this study. We also acknowledge the hard work of the scientists that participated in the cruise and the NMFS personnel who prepared for the charter. We greatly appreciate the helpful comments from three anonymous reviewers that helped us refine the paper. This publication is the result of research sponsored by Alaska Sea Grant with funds from the National Oceanic and Atmospheric Administration, Office of Sea Grant, De- partment of Commerce, under grant no. NA90AA-D- SG066, project number R/31-04N, from the University of Alaska with funds appropriated by the state. Further support was provided by the Auke Bay Laboratory, Alaska Fisheries Science Center, National Marine Fisheries Ser- vice and by a Population Dynamics Fellowship to Hansel- man through a cooperative program funded by Sea Grant and NMFS. Literature cited Brown, J. A., and B. J. F. Manly. 1998. Restricted adaptive cluster sampling. Environ. Ecol. Stat. 5:49-63. Christman, M. C. 1997. Efficiency of some sampling designs for spatially clus- tered populations. Environmetrics 8:145-166. Christman, M. C, and J. S. Pontius. 2000. Bootstrap confidence intervals for adaptive cluster sampling. Biometrics 56:503-510. Everson, I., M. Bravington, and C. Goss. 1996. A combined acoustic and trawl survey for efficiently estimating fish abundance. Fish. Res. 26:75-91. Hanselman, D. H. 2000. Adaptive sampling of Gulf of Alaska rockfish. M.S. thesis, 72 p. Univ. Alaska, Fairbanks, AK. Hanselman, D. H., T. J. Quinn, C. Lunsford, J. Heifetz, and D. M. Clausen. 2001. Spatial inferences of adaptive cluster sampling on Gulf of Alaska rockfish. In Proceedings of the 17th Lowell- Wakefield symposium: spatial processes and management of marine populations, p. 303-325. Univ. Alaska Sea Grant Program, Fairbanks, AK. Lo, N., D. Griffith, and J. R. Hunter 1997. Using a restricted adaptive cluster sampling to esti- mate Pacific hake larval abundance. Calif Coop. Oceanic Fish. Invest. Rep. 38:103-113. Lunsford, C. 1999. Distribution patterns and reproductive aspects of Pa- cific ocean perch {Sebastes alutus) in the Gulf of Alaska. M.S. thesis, 154 p. Univ. of Alaska Fairbanks, Fairbanks, AK Quinn II, T. J., D. H. Hanselman, D. M. Clausen, J. Heifetz, and C. Lunsford. 1999. Adaptive cluster sampling of rockfish populations. Proceedings of the American Statistical Association 1999 Joint Statistical Meetings, Biometrics Section, 11-20. Am. Statist. Assoc, Baltimore, MD. Salehi, M. M. 1999. Rao-Blackwell versions of the Horvitz-Thompson and Hansen-Hurwitz in adaptive cluster sampling. Environ. Ecol. Stat. 6:83-195. Su, Z., and Quinn, T. J., II. 2003. Estimator bias and efficiency for adaptive cluster sam- pling with order statistics and a stopping rule. Environ. Ecol. Stat. 10, pp. 17-41. 510 Fishery Bulletin 101 (3) Thompson, S. K. 1990. Adaptive cluster sampling. 1050-1059. J. Am. Stat. Assoc. 412; Thompson, S. K., and G. A. F. Seber. 1996. Adaptive sampling, 265 p. Wiley, New York, NY. Appendix 1 CPUE (kg/km) data from the 1999 adaptive cluster sampling survey. CPUE is giver in kg/km. The format of "Adaptive 26-1" corresponds to the first adaptive tow around haul no. 26. POP = Pacific ocean perch; SR-RE = shortraker and rougheye rockfish | combined. Summary table Tow type Initial random 2""' phase random Adaptive network Adaptive edge unit Total 1 POP 13 25 49 32 106(119) SR-RE 10 9 21 5 35(45) Total 23 34 70 37 141 (164) ' Values in parenthesis include initial random tows that are not included in estimation results. Criterion determining random tows Tow Latitude Longitude Tow type POP CPUE SR-RE CPUE 3 59.59 -143.81 POP random 39.3 43.7 4 59.54 -143.55 POP random 49.2 13.7 5 59.51 -143.55 SR-RE random 3.4 870.9 6 59.58 -143.28 POP random 174.8 112.0 7 59.56 -143.28 SR-RE random 17.7 582.3 8 59.67 -143.01 POP random 72.7 21.0 9 59.69 -142.75 POP random 21.3 6.1 10 59.64 -142.75 SR-RE random 6.3 6.3 11 59.60 -142.49 POP random 9.6 36.2 12 59.59 -142.48 SR-RE random 3.8 608.0 13 59.40 -142.22 POP random 20.7 113.0 14 59.28 -141.96 POP random 25.3 394.4 15 59.27 -141.96 SR-RE random 19.1 713.1 16 59.17 -141.68 POP random 185.4 68.5 17 59.16 -141.68 SR-RE random 24.9 48.5 18 59.04 -141.41 SR-RE random 1.7 450.4 19 59.03 -141.41 POP random 196.5 21.9 20 59.01 -141.14 SR-RE random 30.0 676.9 21 58.78 -140.88 POP random 2271.6 0.0 22 58.75 -140.88 SR-RE random 65.9 80.6 23 58.67 -140.61 POP random 80.6 101.1 24 58.66 -140.35 POP random 98.2 55.0 25 58.66 -140.35 SR-RE random 21.2 140.5 Beginning of adaptive random tows 26 58.70 -140.64 POP random 576.7 0.0 27 58.68 -140.65 SR-RE random 16.3 115.8 28 58.73 -140.71 POP adaptive 26-1 138.1 12.0 29 58.72 -140.65 POP adaptive 26-2 138.4 9.7 30 58.69 -140.62 POP adaptive 26-3 2294.2 0.0 31 58.70 -140.64 POP adaptive 26-4 290.1 0.4 32 58.70 -140.63 POP adaptive 26-8 334.8 0.0 33 58.69 -140.62 POP adaptive 26-9 56.5 21.2 34 58.69 -140.63 POP adaptive 26-10 16.4 1.9 35 58.71 -140.67 POP adaptive 26-11 20.7 3.7 36 58.72 -140.67 POP adaptive 26-12 30.2 1.0 continued Hanselman et al.: Applications in adaptive cluster sampling of Gulf of Alaska rockfish 511 Appendix 1 (continued) Criterion determining random tows Tow Latitude Longitude Tow type POP CPUE SR-RE CPUE 37 58.69 -140.61 POP adaptive 26-18 1299.4 1.2 38 58.69 -140.61 POP adaptive 26-17 965.0 55.9 39 58.70 -140.75 POP random 62.0 148.0 40 58.76 -140.85 POP Random 3591.0 58.4 41 58.79 -140.89 POP adaptive 40-1 5934.1 0.0 42 58.77 -140.86 POP adaptive 40-2 4521.0 0.0 43 58.74 -140.83 POP adaptive 40-3 515.7 9.1 44 58.76 -140.86 POP adaptive 40-4 4453.7 37.3 45 58.79 -140.90 POP adaptive 40-5 1338.8 0.0 46 58.79 -140.88 POP adaptive 40-6 393.9 0.0 47 58.77 -140.86 POP adaptive 40-7 109.4 0.0 48 58.75 -140.82 POP adaptive 40-8 85.0 0.0 49 58.73 -140.80 POP adaptive 40-9 67.9 0.1 50 58.74 -140.83 POP adaptive 40-10 128.0 17.6 51 58.76 -140.86 POP adaptive 40-11 1597.3 0.0 52 58.78 -140.89 POP adaptive 40-12 268.5 3.8 53 58.80 -140.90 POP adaptive 40-24 1282.9 0.0 54 58.81 -140.92 POP adaptive 40-13 2304.4 0.0 55 58.80 -140.90 POP adaptive 40-14 776.2 0.0 56 58.79 -140.88 POP adaptive 40-15 882.6 0.0 57 58.75 -140.86 POP adaptive 40-22 168.1 2.7 58 58.78 -140.89 POP Adaptive 40-23 253.9 0.2 59 58.83 -140.95 SR-RE random 24.1 290.2 60 58.88 -140.95 POP random 12001.5 0.0 61 58.87 -140.96 POP adaptive 60-4 10659.3 0.0 62 58.91 -140.97 POP adaptive 60-1 1179.0 0.0 63 58.89 -140.95 POP adaptive 60-2 3050.4 0.0 64 58.86 -140.95 POP adaptive 60-3 2984.7 0.0 65 58.86 -140.95 POP adaptive 60-10 3590.4 0.0 66 58.88 -140.96 POP adaptive 60-11 1086.9 0.0 67 58.91 -140.98 POP adaptive 60-12 1311.7 8.7 68 58.92 -140.98 POP adaptive 60-5 1581.0 0.0 69 58.91 -140.96 POP adaptive 60-6 4148.4 0.0 70 58.89 -140.95 POP adaptive 60-7 1297.4 0.0 71 58.86 -140.94 POP adaptive 60-8 214.1 0.0 72 58.84 -140.94 POP adaptive 60-9 2190.3 0.0 73 58.84 -140.94 POP adaptive 60-20 1502.2 0.0 74 58.83 -140.93 POP adaptive 60-19 2828.9 0.0 75 58.84 -140.93 POP adaptive 60-18 102.9 0.0 76 58.86 -140.94 POP adaptive 60-17 46.6 0.0 77 58.89 -140.95 POP adaptive 60-16 27.8 0.0 78 58.89 -140.95 POP adaptive 60-15 53.4 0.0 79 58.92 -140.97 POP adaptive 60-14 495.7 0.0 80 58.93 -140.98 POP adaptive 60-13 1323.4 0.0 81 59.05 -141.05 POP random 1448.8 0.4 82 Coral encountered N/A N/A 83 59.03 -141.08 POP random 560.6 102.8 84 59.03 -141.19 POP random 283.6 298.5 85 59.04 -141.19 POP adaptive 83-1 1119.7 101.3 86 59.04 -141.26 POP adaptive 83-2 1407.0 21.7 87 59.02 -141.22 POP adaptive 83-3 398.1 29.2 continued 512 Fishery Bulletin 101 (3) Appendix 1 (continued) Criterion determining random tows Tow Latitude Longitude Tow type POP CPUE SR-RE CPUE 88 59.03 -141.16 POP adaptive 83-4 264.6 87.0 89 59.05 -141.20 POP adaptive 83-5 416.6 47.3 90 59.04 -141.29 POP adaptive 83-6 2186.1 7.0 91 59.04 -141.25 POP adaptive 83-7 482.0 8.7 92 59.03 -141.22 POP adaptive 83-8 115.2 36.6 93 59.02 -141.19 POP adaptive 83-9 182.5 36.4 94 59.02 -141.13 POP adaptive 83-10 41.4 45.5 95 59.02 -141.16 POP adaptive 83-11 29.2 41.1 96 59.04 -141.20 POP adaptive 83-12 261.4 80.6 97 59.04 -141.25 POP adaptive 83-24 109.3 32.0 98 59.04 -141.29 POP adaptive 83-23 62.0 69.4 99 59.05 -141.26 POP adaptive 83-13 186.4 56.2 100 59.05 -141.32 POP adaptive 83-14 443.8 4.5 101 59.04 -141.29 POP adaptive 83-15 1497.1 5.4 102 59.04 -141.25 POP adaptive 83-16 892.0 21.4 103 59.03 -141.22 POP adaptive 83-17 604.8 26.1 104 59.03 -141.16 POP adaptive 84-3 123.5 91.4 105 59.03 -141.22 POP adaptive 84-4 129.3 285.3 106 59.04 -141.26 POP adaptive 84-1 231.2 602.5 107 59.02 -141.32 SR-RE random 49.3 721.9 108 59.05 -141.26 POP adaptive 84-5 214.6 1408.9 109 59.04 -141.35 POP adaptive 84-6 215.0 123.6 110 59.04 -141.31 POP adaptive 84-12 61.5 664.5 111 59.04 -141.32 SR-RE adaptive 107-1 57.5 758.1 112 59.02 -141.37 SR-RE adaptive 107-2 0.0 490.7 113 59.05 -141.20 SR-RE adaptive 107-3 0.0 408.6 114 59.01 -141.42 SR-RE adaptive 107-4 0.0 669.1 115 59.00 -141.14 SR-RE adaptive 107-6 0.0 760.8 116 58.97 -141.09 SR-RE adaptive 107-8 0.0 1540.6 117 58.11 -141.06 SR-RE random 0.0 443.2 118 59.14 -141.60 SR-RE adaptive 117-1 0.0 1052.8 119 59.09 -141.64 SR-RE adaptive 117-2 0.0 1042.0 120 59.16 -141.50 SR-RE adaptive 117-3 51.3 621.6 121 59.07 -141.69 SR-RE adaptive 117-4 25.7 2096.7 122 59.05 -141.46 SR-RE adaptive 117-6 68.4 480.5 123 59.19 -141.40 SR-RE adaptive 117-5 41.2 924.3 124 59.21 -141.73 SR-RE adaptive 117-7 189.0 731.9 125 59.04 -141.78 SR-RE adaptive 117-8 82.3 772.2 126 59.14 -141,34 POP random 61.9 4.8 127 59.15 -141.60 POP random 82.6 55.8 128 59.21 -141.65 POP random 68.5 8.1 129 59.29 -141.75 POP random 84.6 0.0 130 59.23 -141.85 SR-RE random 6.1 1024.1 131 59.27 -141.85 SR-RE adaptive 130-1 2.6 626.9 132 59.21 -141.94 SR-RE adaptive 130-2 1.5 451.9 133 59.27 -141.81 SR-RE adaptive 130-3 4.2 2208.3 134 59.28 -142.00 SR-RE adaptive 130-5 7.4 1605.6 135 59.31 -142.06 SR-RE adaptive 130-7 5.0 1305.2 136 59.19 -142.11 SR-RE adaptive 130-4 0.0 432.4 137 59.17 -141.75 SR-RE adaptive 130-6 1.6 457.4 138 59.39 -141.70 POP random 181.8 25.9 139 59.36 -142.05 POP random 62.9 12.2 continued Hanselman et al.: Applications in adaptive cluster sampling of Gulf of Alaska rockfish 513 Appendix 1 (continued) Criterion determining random tows Tow Latitude Longitude Tow type POP CPUE SR-RE CPUE 140 59.40 -142.15 SR-RE random 3.7 772.3 141 59.45 -142.25 SRRE adaptive 140-1 1.1 222.7 142 59.38 -142.31 SRRE adaptive 140-2 0.0 209.0 143 59.42 -142.22 POP random 177.2 36.0 144 59.67 -142.25 POP random 45.4 33.5 145 59.60 -142.35 POP random 8.3 117.8 146 59.71 -142.45 POP random 4.3 32.0 147 59.67 -142.65 SR-RE random 2.0 47.0 148 59.64 -142.65 POP random 18.0 50.8 149 59.67 -142.95 POP random 34.2 3.4 150 59.61 -142.85 POP random 125.0 18.8 151 59.57 -143.05 SR-RE random 3.6 530.5 152 59.59 -143.05 POP random 139.0 39.7 153 59.56 -143.15 SR-RE adaptive 151-1 5.1 555.2 154 59.59 -143.16 SR-RE adaptive 151-2 2.6 255.5 155 59.55 -143.00 SR-RE adaptive 151-3 0.0 314.5 156 59.56 -143.22 POP random 23.5 567.4 157 59.57 -143.25 POP random 43.3 399.3 158 59.54 -143.35 SR-RE random 9.3 82.2 159 59.58 -143.36 POP random 74.9 493.0 160 59.55 -143.45 POP random 2838.5 1.8 161 59.57 -143.65 POP adaptive 160-1 1674.5 54.5 162 59.53 -143.69 POP adaptive 160-2 2912.8 1.8 163 59.55 -143.63 POP adaptive 160-3 196.5 0.0 164 59.52 -143.65 POP adaptive 160-4 148.2 0.5 165 59.52 -143.60 POP adaptive 160-5 75.6 21.0 166 59.58 -143.63 POP adaptive 160-6 863.1 9.4 167 59.56 -143.69 POP adaptive 160-7 41.3 0.0 Appendix il Results of estimation with haul no. 60 changed from 12000 kg/km to 540 kg/km. c is the criterion value (kg/km), /i is the mesm Pacific ocean perch density (kg/km) for each estimator, n is the random sample size, v' is the adaptive sample size without edge units. SE is the standard error of the mean. c (kg/km ) c (kg/km) >220 >250 >540 >1080 >220 >250 >540 >1080 ^'srs^"> 445 445 445 445 SE 148 149 175 158 SE 179 179 179 179 ^'HT 442 443 536 413 SE(V) 104 104 104 104 SE 149 149 175 158 i'HH 470 473 535 412 514 Abstract — We investigated the migra- tion and behavior of young Pacific bluefin tuna (Thunnus orientalis) us- ing archival tags that measure envi- ronmental variables, record them in memor>-, and estimate daily geographi- cal locations using measured light levels. Swimming depth, ambient water temperature, and feeding are described in a companion paper. Errors of the tag location estimates that could be checked were -0.54° ±0.75° (mean ±SD) in lon- gitude and -0.12° ±3.06° in latitude. Latitude, estimated automatically by the tag, was problematic, but latitude, estimated by comparing recorded sea- surface temperatures with a map of sea-surface temperature, was satisfac- tory. We concluded that the archival tag is a reliable tool for estimating location on a scale of about one degree, which is sufficient for a bluefin tuna migra- tion study. After release, tagged fish showed a normal swimming behavioral pattern within one day and normal feeding frequency within one month. In addition, fish with an archival tag maintained weight-at-length similar to that of wild fish; however, their growth rate was less than that of wild fish. Of 166 fish released in the East China Sea with implanted archival tags, 30 were recovered, including one that migrated across the Pacific Ocean. Migration of young Pacific bluefin tuna appears to consist of two phases: a residency phase comprising more than 80% of all days, and a traveling phase. An indi- vidual young Pacific bluefin tuna was observed to cover 7600 km in one trav- eling phase that lasted more than two months (part of this phase was a trans- Pacific migration completed within two months). Many features of behavior in the traveling phase were similar to those in the residency phase; however the temperature difference between viscera and ambient temperature was larger, feeding was slightly more fre- quent, and dives to deeper water were more frequent. Migration patterns of young Pacific bluefin tuna (Thunnus orientalis) determined with archival tags Tomoyuki Itoh Sachiko Tsuji National Research Institute of Far Seas Fisheries 5-7-1 Shimizu-Orido, Shizuoka Shizuoka, 424-8633, Japan E-mail address (for T Itoh): itou@fra.affrcgo.|p Akira Nitta Japan NUS Co., Ltd. Loop-X BIdg., 3-9-15 Kaigan, Minato Tokyo, 108-0022, Japan Manuscript approved for publication 22 October 2002 by Scientific Editor. Manuscript received 3 January 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:514-534 (2003). Pacific bluefin tuna (Thunnus orien- talis), a highly migratory species, is mainly distributed in the temperate zone of the northern Pacific Ocean (Yamanaka, 1982; Bayliff, 1994) in con- trast to T. thynnus, which inhabits the Atlantic Ocean (Collette, 1999). Current knowledge on the migration of Pacific bluefin tuna is summarized in the following studies: Aikawa ( 1949); Bell ( 1963a ); Okachi ( 1963 ); Orange and Fink ( 1963 ); Nakamura ( 1965 ); Clemens and Flittner ( 1969); Shingu et al. ( 1974); Yorita( 1976); Bayliff( 1980); Yamanaka (1982); Yonemori (1989); and Bayliff et al. (1991). The majority of bluefin tuna spawn in the northwest Pacific Ocean in an area from the Philippines past Taiwan to Okinawa from April to June, and small numbers spawn off southern Honshu in the Pacific Ocean in July and in the Sea of Japan in August (Yabe et al., 1966; Ueyanagi, 1969; Okiyama, 1974; Yonemori, 1989; Kitagawa et al., 1995). Carried by the Kuroshio Cur- rent, juveniles arrive near the coast of Japan, move northward during sum- mer and early autumn, and then most turn around and move back southward during late autumn and winter along the Japanese coast. During the first few years of their lives, the majority of young fish repeat a similar north-.south seasonal migration. However a small fraction, increasing each year, moves away from the Japanese coast and often reaches the eastern side of the Pacific Ocean, off the United States and Mexico. These fish stay in the eastern Pacific Ocean for 1-3 years. Some time later, as mature fish, they gather in the north- west Pacific Ocean to spawn and then disperse after the spawning season. This information on Pacific bluefin tuna movements has been accumu- lated through analyses of fishery catch data and tag-recapture data. Fishery catch data are based on different indi- viduals from limited areas where, and particular seasons when, fishing took place. Conventional tagging data pro- vide migration information regarding only two points: release and recapture. Acoustic tracking, another method used for investigating behavior and migra- tion of individuals, can collect detailed information on fish movements and behavior on a time scale of seconds, but the duration of the tracking period of each individual has usually been less than several days in studies of Pacific bluefin tuna (Marcinek et al., 2001; Hisada et al.') as well as in other stud- ies of Thunnus species (e.g. Carey and Olson, 1982; Holland et al., 1990; Cayre, 1991; Cayre and Marsac, 1993; Block 1 Hisada, K., H. Kono, and T Nagai. 1984. Behavior of young bluefin tuna during migration. In Progress report of the marine ranching project 4, p. 1-7. Nat. Res. Inst. Far Seas Fish. Pelagic Fish Resource Division, 5-7-1 Shimizu-Orido, Shizuoka, Shizuoka, 424-8633, Japan. [In Japanese, the title was translated by authors.! Itoh et al,: Migration patterns of Thunnus orienta/is determined with archival tags 515 et al., 1997). These methods have not yielded detailed information regarding migration, behavior, and their rela- tion to environmental factors for Pacific bluefin tuna over a long period. An archival tag is an electronic device that measures environmental variables and records data in its memory. When attached to an animal, it allows direct examination of the relationship between an animal's behavior and physiological condition, or the ambient environment. One type of "archival" tag merely stores data; however, another type not only stores data but also provides daily geographi- cal locations of the fish by processing the measured envi- ronmental data. This type of archival tag was anticipated since the 1980s as a tool that could collect detailed infor- mation on individual fish behavior (Hunter et al., 1986; Anonymous, 1994). Metcalfe and Arnold (1997) estimated the geographical locations and tracks of plaice, a demersal species, by comparing tidal depth variations with the time series depth data recorded by archival tags attached to the fish. However, this method is not suitable for pelagic fish, which change swimming depth freely. A type of archival tag that can estimate geographical locations based on change of light levels during a day — a method more suitable for pelagic species — has been commercially available since the early 1990s. So far, archival tags of this type have been used in several tagging projects (Arnold and Dewar, 2001). The results published in a few reports on southern bluefin tuna (T. maccoyii) (Gunn and Block, 2001), and Atlantic bluefin tuna (T! thynnus) (Block et al., 2001), show the remarkable value of archival tag data. As archival tags have come into wide use, results of sev- eral experiments conducted to evaluate the reliability of its geolocation estimates have been published (Welch and Eveson, 1999; Musyl et al., 2001; Gunn et al.^). However, several points remain to be tested: tag reliability when a number of tags are deployed for long duration, reliability of sensors for variables other than light, and the effects of attaching the tag to fish. After two preliminary experiments with tags in 1994, the first with tags placed at a known outdoor location on land and in air, and the second with tags attached to young Pacific bluefin tuna held in pens, we applied archival tags to wild young Pacific bluefin tuna to investigate their be- havior and migration. In the present study we report the characteristics of migration for this species based on data on daily geographical location, as well as the reliability of archival tag data and the effect of attachment of the tag to fish. Analyses for swimming depth, ambient water temperature, and feeding frequency of the species are undertaken in other papers (Kitagawa et al., 2000; Itoh et al.,2003). 2 Gunn, J., T. Polacheck, T. Davis, M. Sherlock and A. Betlehem. 1994. The development and use of archival tags for studying the migration, behavior and physiology of southern bluefin tuna with an assessment of the potential for transfer of the technol- ogy to groundfish research. Proc. ICES mini-symposium on migration, St. Johns, Newfoundland. ICES CM. Mini:2.1, 23 p. International Council for the Exploration of the Sea, Palaegade 2-4, DK-1261 Copenhagen K, Denmark. Materials and methods Outline of the archival tag used in this study The archival tag used in this study (Northwest Marine Technology, Inc. Shaw Island, WA ) had a cylindrical stain- less-steel body ( 16 mm in diameter and 100 mm long, and weighing 52 g) that was implanted in the animal. A flexible sensor stalk 2.2 mm in diameter and 150 mm long extended from the tag through the skin of the animal into the water. The end of the stalk housed an external temperature sensor and a light capture region. Light was led from the capture region by optical fiber to a photodiode sensor in the body of the tag, which also housed sensors for pressure, internal temperature, and light. Response times for the tempera- ture sensor were three seconds for the external sensor and 20 seconds for the internal sensor, and temperature resolu- tion was 0.2°C for both sensors. Resolution of the pressure sensor record was 1 m at shallow depths up to 126 m, then changed to 3 m from that depth to the scale limit of 510 m. Clock drift was less than 30 seconds per year. The tag had a data measurement interval of 128 seconds, a 256-kByte data memory, and an operating life exceeding seven years. Data were downloaded from recovered tags by using a per- sonal computer and a fiber-optic connector. Two types of data files were created within the tag memory. One data file stored daily records containing date, estimated times of sunrise and sunset, water temperatures at 0 m plus two other selectable depths (we selected 60 m and 120 m), and other information required for, or produced in, the course of location estimates for each day. This file is referred as the "summary file" in the "Results" section, and it stored data for all days after the memory was last cleared. The times of sunrise and sunset were estimated within the tag from sea-surface light intensities, which were inferred from measured depth and measured light intensity at depth and a water opacity factor determined from the measured data each Universal Time (UT) day. The time of midday was determined as the midpoint between sunrise and sunset times, and longitude was calculated from the difference between the midday time and 1200h UT, at a rate of 15 degrees longitude per hour, corrected for astronomical effects. Latitude was estimated from the duration of daylight (Hill, 1994). The second data file contained unprocessed time series data records taken at 128-second intervals. The tag could record at any integer multiple of its 128-second measure- ment interval and a multiple of one was chosen. Each record consisted of external temperature, internal tem- perature, pressure, and light intensity, and corresponded to a known time. This is referred to as the "detail file" in the "Results" section. It could hold about 54,000 records, or about 80 days of steady recording at the high data rate chosen — a small fraction of the tag's overall lifetime. The time-series memory was divided into two sections, and the size allocations for the two sections were determined by the user The first section filled first and did not change there- after. The second section filled next, but once full, it was continually overwritten by new data. Thus the first section always contained the earliest data retrieved from a tag; the 516 Fishery Bulletin 101(3) second always contained the latest data. We divided the file into two 40-day sections for releases in 1995 and 1996, and into 20- and 60-day sections for releases in 1997. Reliability and calibration of archival tags in air To examine the reliability of location estimates made by archival tags, 117 archival tags were left outdoors (34°59'N; 138°59'E) where they were not affected by artifi- cial light during July-September 1996 (55 days, five tags), May-August 1997 (86 days, 14 tags), and October 1997 (five days, 100 tags). Two of the tags were used in two of the experiments. Calibration tests of internal and external temperature sensors were conducted for all tags before being implanted in fish that were released, and the sensors were recalibrat- ed for nine tags after they were recovered. Temperature calibration was done by immersing tags into a series of water tanks that were set to temperatures ranging from 5.0° to 30.0°C by 5°C intervals. Calibration tests of pres- sure sensors were also conducted for all tags before release and on 27 tags after being recovered. Tags were placed in a pressure chamber with a resolution of 0.1 bar and exam- ined up to 20 bar. The tags were left at least five minutes at each temperature or pressure to obtain at least two measurements at the 128-second recording interval. Experiment with pen-held fish Archival tags were attached to three pen-held young Pacific bluefin tuna of 93-97 cm fork length (FL) at Kasasa in Kagoshima Prefecture (31°25'N; 130°11'E) in November 1994. The fish had been reared in a net pen (40 m x 25 m with 12 m depth) for more than two years and were accli- mated to the environment at the time of the experiment. Archival tags were inserted into the abdominal cavities of two of the three fish by the following method. A fish caught by hook and line was put into a styrofoam box, and its eyes were covered with a black polyethylene bag. The belly of the fish was cut with a scalpel about 4 cm anterior to the anus, 3—4 mL of antibiotic (artificial penicillin, Doil, Tanabe Seiyaku Co., Ltd., Osaka, Japan) was injected into abdominal cavity of the fish, and an archival tag was inserted there with the stalk extending through the inci- sion. A stitch was made in the middle of the incision with an absorbable suture (Coated Vicryl, type J583G, Ethicon Inc., Cornelia, GA ), and the fish was released back into the pen. All tools and tags were disinfected with 100% ethanol. No anesthetic was used because with their eyes covered, the fish remained quiet during the surgery. This simplified procedure (from making the incision to releasing the fish) could be completed in less than 90 seconds, thus minimiz- ing total stress on the animal and, in later experiments on wild fish, providing the best chance for the animal to rejoin its original school. In this pen study, the third fish was tagged externally instead of internally, the tag being connected by a thin wire rope to a small metal arrowhead inserted in a muscle near the second dorsal fin base. During the pen-held fish experiment, none of the fish were observed to die as a result of tagging. The tag that had been attached externally came loose from the fish and was retrieved from the bottom of the pen four days after tagging. One tagged fish escaped when the pen was broken. The remaining tag was recovered 453 days later when the fish was caught from the pen as part of a com- mercial catch. Experiments with wild fish Tag and release experiments on wild young Pacific bluefin tuna were conducted near Tsushima, at the northeastern end of the East China Sea, by using chartered commercial trolling vessels, every November and December from 1995 to 1997. A total of 166 fish, ranging from 43 to 78 cm FL (age 0 or 1), were internally tagged as described above and released immediately. Two dart-type conventional tags were also attached to the second dorsal-fin base of each fish in the 1997 experiment as visual markers in an attempt to improve the recovery rate. Thirty archival tags (18.1%) were recovered. The dura- tions at sea were 50 days or less for 13 fish, 96-211 days for 13 fish, and 359-375 days for three fish, all recaptured around Japan. One additional fish was recaptured off the west coast of Mexico, on the east side of the Pacific Ocean, at 610 days after release. Data could not be downloaded from one archival tag released in 1995 and recovered 30 days after release; all other tags returned data. Results Reliability of location estimates The tag recovered from a fish penned in a known location for 453 days yielded a record of positions automatically estimated during that time. Figure 1 plots the errors in those estimates and the date when each was made. This tag provided the only position sequence of long duration obtained from a captive fish. Unfortunately it was discov- ered later, after the experiment was completed and after this particular tag was no longer available for further test- ing, that the light sensitivity of this tag, as well as that of the tag that yielded data for four days in the captive fish experiment, was at least a factor of ten lower than that of other tags. This discrepancy in light sensitivity could be seen in the daily noon-light intensity data in the summary file, both during the in-water experiment (when compared with typical values for tags in wild fish) and when tested in air (compared with other tags of the group tested in air). On dark days there was an unusual pattern of early sunset times and late sunrise times that the tag manufacturer interpreted as being associated with the low light sensi- tivity. Thus, although the general trends of error size with season can be expected to be representative, the absolute size of the errors was likely inflated in this, the only long- term record obtained from a captive fish. Longitude error showed no change with season, but latitude error increased dramatically near the equinoxes as expected because day length does not vary significantly with latitude at that time, and therefore carries little in- Itoh et al.: Migration patterns of Thunnus orientalis determined with archival tags 517 -1 1 n 10Nov94 18Feb95 29 May 95 6 Sep 95 15Dec95 _ 40 -, 10 Nov 94 18 Feb 95 29 May 95 6 Sep 95 15 Dec 95 Date Figure 1 Seasonal change of errors in location estimate. Upper and lower panels show longitudinal and latitudinal errors, respectively. Triangles show equinoxes. Data are from an archival tag implanted in a fish held in a pen for 453 days. formation about latitude. The tag did not provide a latitude estimate for 18 days around and at the vernal (autumn) equinox and had large errors for one month before (or after) as well as 10 days after (or before) that period, respectively. The same pattern was observed in the test with archival tags that were left in air. In addition, the latitude estimates were biased toward south in summer and toward north in winter, that is, toward erroneously short day lengths. Occasional large deviations were observed in both lati- tude and longitude estimates. These were easily identified as outliers in our analyses of data obtained from wild fish by comparing them with estimated locations for adjacent days. When evaluating the accuracy of location estimates for practical use in analyses of wild fish movements, we excluded longitude or latitude estimates that differed more than 10° from the real location and the latitude estimates not provided by the tag near the equinoxes. These account- ed for 2.8% of longitude estimates and 8.9% of latitude estimates obtained in the tests in air, as well as 4.8% of longitude data and 47.5% of latitude data obtained in the tests of pen-held fish. Figure 2 shows the position estimates and error bars corresponding to one standard deviation for 1 17 tags tested in air — most of them for a .5-day period, five for 82 days, and twelve others for various intermediate durations. The aggregate of all observations in air yields an error estimate (mean ±standard deviation) of -0.54° ±0.75° for longitude, and -0.12° ±3.06° for latitude. When individual tags tested in air were examined separately, 96%- of tags (112/117) showed average posi- tion errors within a range of ±1.5° in longitude. Among these 112 tags with small longitude errors, 95 had been manufactured within the last half year and had an av- erage and standard deviation of position error equal to -0.50° ±0.19°, and the other 17 tags were more than one year old and had an average position error of -0.51 ±0.75°. The average is not significantly different (ANOVA F=0.01,P>0.05) and the younger tags had a smaller stan- dard deviation (F=412, P<0.01). No significant difference of accuracy was observed among the 17 older tags that could be related to their history, i.e. among four tags kept in air without release and 13 tags released with fish and recovered (F=1.01 for average and F=2.81 for standard deviation, both P>0.05). For the two tags attached to fish in pens, one tag mea- sured only five positions with a resulting error estimate of -2.38 ±0.39 for longitude, and -1.82° ±1.58° for latitude. The other measured 432 positions, with a resulting error estimate of -0.53° ±2.46° for longitude and 1.26° ±5.33° for latitude. This is the data series presented earlier in Figure 1. The large standard deviation in longitude error — much larger than that obtained in other tests — initially raised questions regarding the effect of water on the posi- tioning techniques. However as mentioned earlier, the low light sensitivity of both tags used in captive fish was identi- fied as the likely cause of these large errors. 518 Fishery Bulletin 101 (3) 10.0 1 -10.0 -10.0 0.0 Longitudinal error (degree) 10.0 Figure 2 Average errors of location estimates from each archival tag in the benchmark tests. Smaller circles are the average errors of the tags tested in air and the larger circles are the average errors for archival tags implanted in pen-held fish. Bar shows a standard deviation. A more useful measure of in-water accuracy was pro- vided by comparison between actual recapture locations of 18 tags and the locations that those tags estimated one or two days prior to capture (thus avoiding the disturbed light data on the final day). Average differences of the 18 tags were -0.1 ±0.8° (range: -2.0 ±1.7°) in longitude and -1.6 ±1.8° (range: -5.7 ±0,6°) in latitude. Because the tag's latitude estimate based on day length was found to have limited reliability, we estimated latitude using sea-surface temperature (SST) as recorded in the summary file for each day. The temperature reference field used was the SST map published by Japan Fisheries In- formation Service Center, which gave average SST weekly for the western Pacific Ocean (west of 160°E), and every 10 days for the eastern Pacific Ocean (east of 160°E). The longitude value determined automatically by the tag was used to choose a longitude on the SST map. Along that lon- gitude line a point was sought where the map SST matched the SST value recorded by the tag. If multiple points were found to satisfy this criterion, the point that gave the most plausible movement was selected, based on fish locations on several adjacent days. If a location still could not be determined, it was interpolated as a midpoint between the adjacent two days' locations. One example of location re-estimation is shown in Figure 3. After consulting with the SST maps, we used 1.4% of the locations estimated automatically from 29 recaptured tags. and 79.7% of latitudes were changed by -(-0.3 (±2.8°) on av- erage wdth the SST method. The remaining 18.9% of days did not provide any reasonable location estimates for various rea- sons, including anomalous longitudinal estimates, no match points of SST along the estimated longitudinal line, or the existence of a wide latitudinal area showing the same SST. Reliability of temperature and pressure sensors One hundred tags calibrated within half a year of manufac- ture showed average errors of 0. 1 ±0. 1°C for both internal and external temperature sensors. Nine tags recovered from fish and tested more than one year after manufacture showed average errors of 0.0 ±0.1°C for both sensors. It thus appears that no deterioration of the temperature sensors occurred because of release-recapture or the passage of time. No large error in pressure sensors was observed dur- ing calibration of tags before release. However, 20 of 27 tags recovered from tagged and released fish were found on recalibration to record substantially lower than actual pressure. One example is shown in Figure 4. No further deterioration of pressure sensors was observed when these tags were kept in air for an additional half year There was no way to know exactly when the sensor deteriora- tion had occurred during the time the fish were in water However, the frequency of records showing swimming at 0 m depth was remarkably higher in the second part of Itoh et al ; Migration patterns of Thunnus orientalis determined with archival tags 519 130°E MO-E f 1 40°N "-l\ll ^/ ^^ ^-M^y^ ^S^/ ^> 30°N ^1 w '■ • V \ A -■■J f ' J^ ^^A . \ \ I y^ 4\ 20°N A 130E 140E 40°N 30°N 20°N Figure 3 Locations estimated by an archival tag with a young Pacific bluefin tuna before (upper panel) and after (lower panel) replacement of the original latitudinal estimate based on day length by one using sea- surface temperature. Locations out of the range of the figure and those for which latitude was not estimated were not drawn in the upper panel. Estimated locations for all days are shown in the lower panel. Open circles in the lower panel are interpolated locations. the detailed file (i.e. just before recapture) when compared to the first part of the detailed file (i.e. just after release). The tag manufacturer analyzed this deterioration in the pressure sensors, and expected the sensor characteristics to remain constant after an initial change (if one occurred), and agreed that the early and late pressure data should be treated separately. We assumed that the deterioration oc- curred sometime during the middle period of the time the fish was free, when no record was being kept in the detail file. Recorded depths in the second part of the detail file for eight tags with relatively large deterioration detected were corrected by using two regression lines joined at around 30 m in real depth for each tag (Fig. 4). Effect of the archival tag on fish The effect of both the implantation process and the pres- ence of the implanted tag in the fish was investigated by macroscopic observation of recovered fish. Further infor- mation was obtained by comparing the weight at length 520 Fishery Bulletin 101(3) Real depth (m) Figure 4 An example of observed deterioration in a pressure sensor of an archival tag in a postdeployment recalibration. The horizontal axis shows the test pressure, vertical axis is pres- sure recorded by the tag. Dots are observed data. The solid line bent at 25 m of real depth is formed from two regres- sion lines, one fitted to data below and one to data above 25 m depth. This approximation to the deteriorated sensor characteristic was used to correct pressure data for this tag. Pressure values are converted to depth in meters. A broken line is that of observed depth equal to real depth. o m and the monthly average growth rates of tagged fish with wild fish, and also by evidence of feeding to be found in the records returned in the tags. The bodies of two fish among 30 recoveries were avail- able for observation. One fish recaptured 27 days after release still had a scar on its skin but no trace of the tag insertion surgery was found in its belly muscle. Another fish recaptured 200 days after release had no trace of sur- gery either on its skin or in its belly muscle. Surface skin around tag stalks was ulcerated in both fish. The stalks were immobilized in the belly muscle. The cylindrical bod- ies of both tags were covered with membrane and located between the stomach and the pyloric caeca. No infection or necrosis was observed in the visceral organs around tag bodies or in the muscle around tag stalks. Body weights of all recaptured fish that were measured (^=8, 17-203 days after release) were within the range of those of wild fish of the same fork length (Fig. 5). An average growth rate of recaptured fish was 1.4 ±0.5 cm per month (n=6, three fish recaptured at short durations of liberty that showed no or negative growth were excluded). A subgroup of four fish recaptured after more than 5 months from release, i.e. fish at liberty during the summer when growth might be expected to be faster, had an average growth rate of 1.3 ±0.6 cm — similar to that from all durations. The average number of daily feeding events, which were found by specific changes of visceral temperature ( Itoh et al., 2003), increased linearly from no feeding on the day of release up to a steady rate beginning about 30 days after release. Thus, it appeared that fish did not feed normally during this initial period (Fig. 6). 9 -1 /O203 8 - / 7 - 6 - y027 5 - / 4 - 3 - ? - ^/^"l7 -£1^ , , 1 1 40 50 60 Fork length (cm) 70 80 Figure 5 Comparison of weight at length of young Pacific bluefin tuna between recaptured fish tagged with archival tags and wild (untagged) fish. Numerals show days at liberty. An average (thick solid line) and upper and lower 95'7f confidence limits (thin solid lines) are derived from 11,777 wild fish from 40 to 80 cm in fork length caught in 1995 and 1996 around Japan. Equations for average is H''=2.844x 10''xL-'"", upper 95% limit is W=.3.028x lO^'x/-^^™, and lower 95% limit is W=2.745xl0'^xL2 906 where L = fork length in cm and VV = body weight in kg. Horizontal movement Estimated tracks of all fish that traveled out of the East China Sea along with one fish that remained in the East China Sea are shown in Figure 7 and Figure 8. All of these fish were released off Tsushima in November or Decem- ber and remained in the East China Sea at least 90 days. After that, four fish entered the Sea of Japan and moved northward from April to July (Fig. 7, A-D). Two of them moved southward in November one year after release (Fig. 7, C and D), and one of the two fish returned to the region off Tsushima where the fish were released (Fig. 7C). One fish remained and was recaptured within the East China Sea in November, one year after release, although it had moved to the east coast of the Korea Peninsula for a period in August and September (Fig. 7E). Ten fish remained within the East China Sea for more than five months and were recaptured from May to June, five to seven months after release (Fig. 7F). Two fish moved to the Pacific Ocean (Fig. 7G and Fig. 8). One of these fish entered the Pacific Ocean on 7 March 1996, and then traveled east- ward straight from a position off the south coast of Kyushu (31°N, 131°E) to one off the east coast of Choshi (36°N, 142°E), then stayed for a while Itoh et al.: Migration patterns of Thunnus orientalis determined with archival tags 521 H 4-1 2 - 1 - cP O qO o (HDtDD cm «„„ "^X) On «DO o % O coxm canxoDODD cramaDari C3D0 ojjoo 5^ o o ■ ■ aD c^^o 0(5b° ° oOo(5c£ o ID 00 — I—l 50 100 Days after release 150 200 Figure 6 The average number of daily feeding events plotted against days after release. in an area of 32-37°N, 143-147°E (Fig. 7G). This fish was recaptured by purse seine on 7 June 1996. The other fish traveled from the western Pacific Ocean to the eastern Pacific Ocean as follows (Fig. 8). It was re- leased off Tsushima on 29 November 1996 at 55 cm FL and remained for a period within the East China Sea. It moved to the Pacific Ocean on 1 May 1997 and then traveled east- ward straight from a position off the south coast of Kyushu to one off the east coast of Choshi then stayed for a while in an area of 34-39°N, 143-150°E. It moved northeastward from 30 July to 18 August 1997, then stayed in the area 40-44°N, 152-163°E. It began the trans-Pacific migration on 11 November 1997 at 41°N, 163°E, and traveled straight to northern California, U.S.A. (36°N, 127°W) arriving on 15 January 1998. After arriving in the eastern Pacific Ocean, this fish initially stayed in an area of 33-40°N, 122-128°W, then moved southward from 25 February to 3 March, then again stayed in an area of 25-29°N, 116-119°W. It started mov- ing northward on 9 May and reached 40°N, 127°W on 25 May, but without staying there moved again southward and reached an area of 25-29°N, 116-120''W on 12 June, close to the place from which it had departed. The fish was recaptured by a recreational fishing vessel on 1 August 1998, 610 days after release, off Baja California, Mexico (31°48'N, 117n8'W), at 87.6 cm FL The track of this fish consisted of apparently separable segments, five traveling periods and six resident ones. All of the fish that moved out of the East China Sea showed the same type of pattern, staying resident in an area for a relatively long period and then traveling continuously for at least several days in a stable direction. The terms "traveling phase" and "residency phase" are used in the following description. If a fish moved continu- ously for more than three days in a stable direction covering more than 700 km in total distance, it was considered to be in a traveling phase — at all other times in a residency phase. A few movements for short periods or short distances (or both) were also observed during periods of a residency phase: a fish resident off the east coast of Hokkaido (40-44°N, 152-163°E) shifted eastward gradually within the area during a period of two months (Fig. 8). Another fish resident in the northern area of the East China Sea moved rapidly to the southern area of the East China Sea at the end of December and came back rapidly to the northern area in early May (Fig. 9). In- dividual movements were completed within a few days and the total distances moved were far shorter (380 and 310 km, respectively) than those seen in traveling phases. A total of 12 traveling phases were identified in records of six fish (Table 1). The direction of travel stayed constant within each phase, except in one case where a fish com- pletely turned around in the middle of traveling (in the east- em Pacific Ocean in May and June 1998). Daily distances moved during those traveling phases were calculated. To reduce the infiuence of scatter in the estimated locations, three-day running averages of latitude and longitude were used for calculation. Excluding the one trans-Pacific migra- tion phase of exceptional length, 7636 km (66 days), the total distance per traveling phase ranged from about 730 to 3406 km (average: 1430 km). The duration of a traveling phase was four to 35 days (average: 17 days) and the dis- tance traveled ranged from 59 to 182 km (average: 104 km). Six residency phases which occurred between two clearly identified traveling phases lasted from 40 to 125 days (aver- age: 81 days). In total, 83% of days were in a residency phase and 17% of days were in a traveling phase. If residency phases for which the beginning or end could not be defined because offish release or recapture were also included, the average duration of residency phases increased to 110 days, and the proportion days belonging to each phase became 87% in residency and 13% in traveling phases. Comparison of fish behavior and ambient water temperature between traveling and residency phases Several points regarding fish behavior, described in detail in Itoh et al. (2003), were compared between all days in a traveling phase and ten days in the residency phase that for four fish immediately preceded the traveling phase. In the case of one fish (no. 241) where data for preceding 522 Fishery Bulletin 101 (3) 130°E 140°E 40°N 30°N \ .'. Red 20596 130°E 140°E 40°N 30°N Rec300497 Figure 7 Tracks of young Pacific bluefin tuna estimated by archival tags. "Rel" and "Rec" mean release and recapture, respectively. Numerals are dates in ddmmyy. Each panel shows the track of one fish with an archival tag. days were not available, ten days from the residency phase immediately following were used instead. Because errors in determining geographical positions introduced scatter in the sequence of estimated positions, the onset (or end) of a traveling phase was not always easy to define. In response to this situation, three days of the residency phase nearest Itoh et al; Migration patterns of Thunnus orientolis determined with archival tags 523 130°E 140°E 40°N \ '■ 270898 30°N 130°E 140°E D / a1\ 1 * fj^\ y '^S °N x2 ■ 080898 \ \ AmJ^ 111198 A_^. 'V, \ 140798 f ^t^ V/ / ?^Rel231197 "' fl^^ ^ ■'.. Rec291198 °N fj Br ' '-'o' ' Figure 7 (continued) to the traveling phase were not included, that is to say, ten days between the fourth and thirteenth days preceding (or following) the onset (or end) of a traveling phase were used. Among 12 features investigated, three differed between the two phases (Table 2). The temperature difference be- tween fish viscera and ambient water (thermal excess) was more than 1.0°C higher during the traveling phase for four out of five fish. In the one remaining fish (no. 241), data used for analysis were those from days at the end of the traveling phase. The larger thermal excess during a trav- eling phase was observed at both daytime and nighttime (Fig. 10). The second significant feature was that all fish dived to water deeper than 150 m more frequently during the traveling phase. Except for one fish (no. 241), which spent a long time in water deeper than 150 m, almost all 524 Fishery Bulletin 101(3) 130°E 140°E E f ^'.^ N A / '*TV J ' ' r^ >'V~^^ ;^ /^ "jl 170898 ■^c 40°N Rel211197__ ; m ■"150998 J" ■J 30°N ^f.4 Rec301198 , ar • 130°E 140°E 40°N 30°N Figure 7 (continued) records of excursions below 150 m were due to spikes of deep diving of short duration, less than 10 minutes. Fi- nally the frequency of daily feeding events, detected by a change of visceral temperature, was slightly higher dur- ing a traveling phase (1.6 ±0.6) than a residency phase (1.1 ±0.4). Records of surface temperature from the summary file were examined to answer three questions regarding the re- lation of water temperature to traveling. The first question was whether any water temperature change, an increase in spring and summer or a decrease in winter, was observed several days prior to the onset of the traveling phase. Such a temperature change was observed in 10 out of 12 cases ( Table 3, Fig. 11). In those 10 cases, the water temperature increased in spring and summer to 19-26''C (average of 22°C), and decreased in winter to 15-17°C (16°C). Itoh et al.: Migration patterns of Thunnus orientalis determined with archival tags 525 130°E 140° E 40°N 30°N Figure 7 (continued) 10898 Figure 8 Track of a young Pacific bluefin tuna that traversed the Pacific Ocean, estimated with an archival tag. "Rel" and "Rec" mean release and recapture, respectively. Numerals are dates in ddnimyy. The second question is whether the water temperature at the onset or end of traveling was within the temperature range of 14— 20°C, considered to be in the temperature range preferred by young Pacific bluefin tuna (Itoh et al., 2003). If the act of traveling was simply a reaction to the water tem- perature, fish would be expected to travel from water with a temperature out of that range to one within that range. How- ever that was observed in only two of 12 cases (Table 3). The third question was whether the temperature at the end of traveling was a temperature that the fish encoun- tered for the first time since the onset of traveling. We found, however, that there was no specific trend in water temperatures during traveling phases (Fig. 11). Six of 12 temperatures at the ends of traveling phases were not the first one that the fish had experienced during the phase (Table 3). 526 Fishery Bulletin 101 (3) 125°E 130°E 35° N 30° N • rr«.».-^ 35° N 30° N ■' Figure 9 Rapid movements of a fish with an archival tag in the residency phase in the East China Sea. The fish moved rapidly south (upper panel; open circles are in December 1995 and solid circles are in January 1996) and north (lower panel; open circles are in April 1996 and solid circles are in May 1996). Discussion Reliability of archival tag data The reliability of archival tag data for geolocation estimates based on measured light inten.sity ha.s been examined by implanting the tags in pen-held fish or attaching the tags to a stationary subsurface mooring (Welch and Eveson, 1999; Musyl et al., 2001; Gunn et al.-». About one degree of reli- ability for both longitude and latitude were the results. Our study included further tests: for a large number of tags; for sensors other than light sensors; for reliability of tags over time; for tags manufactured by Northwest Marine Technol- ogy that were applied to wild young blucfin tuna and not fully examined in previous studies; and finally for the effect of tag attachment on Pacific bluefin tuna. The benchmark test in this study showed that longitude estimated by archival tags had an error (mean ±standard deviation) of -0.54° ±0.75°, which differed by only -0.1° ±0.8° from a comparison of in-water tag position results with actual recapture locations. In the on-land benchmark test, the mean error did not change with tag age, although the standard deviation slightly increased. Ninety-six percent of all tags tested were considered to have suffi- cient reliability in longitudinal estimation. We concluded, therefore, that the archival tag is a reliable tool to estimate longitude on a scale of about one degree. Latitudes estimated automatically from day length car- ried larger errors than estimations of longitude, and the accuracy of estimation changed with season as well as with the latitude itself (Hill, 1994; Hill and Braun, 2001). Smith and Goodman (1986) recommended estimating latitude by Itoh et al.: Migration patterns of Thunnus orientalts determined with archival tags 527 Table 1 Information on traveling phases of young bluefin tuna as recorded by archival tags. Daily travel distances were estimated from a | three-day ru nning average of latitude and longitude. ID offish and number Onset of End of Duration of Total Daily (e.g. Ml) traveling phase traveling phase traveling distance distance of traveling phase traveled traveled phase Area Date Location Date Location (days) (km) (km/day) 241 Ml western Pacific 1 May 1997 31N 132E 15 May 1997 36N 143E 15 1,341 89.4 241 M2 western Pacific 30 Jul 1997 40N 147E 18 Aug 1997 42N 152E 20 1,161 58.1 241 M3' central Pacific 11 Nov 1997 41N 163E 15 Jan 1998 36N 127W 66 7,636 115.7 241 M4 eastern Pacific 25 Feb 1998 32N 126W 3 Mar 1998 29N 121W 7 1,043 148.9 241 M5- eastern Pacific 9 May 1998 29N 119W 12 Jun 1998 27N 120W 35 3,406 97.3 209 Ml western Pacific 7 Mar 1996 30N 131E 6 Apr 1996 35N 142E 31 1,860 60.0 164 Ml- Sea of Japan 4 May 1996 36N 134E 12 May 1996 38N 138E 9 770 85.6 319 Ml- Sea of Japan 18 Apr 1997 35N 132E 30 Apr 1997 38N 138E 13 1,346 103.5 688 Ml Sea of Japan 23 Jun 1998 37N 132E 30 Jun 1998 39N 138E 8 877 109.7 688 M2-' Sea of Japan 3 Nov 1998 44N 140E 14 Nov 1998 34N 128E 12 1,611 134.2 760 Ml Sea of Japan 13 Jul 1998 35N 129E 8 Aug 1998 40N 137E 27 1,590 58.9 760 M2-' Sea of Japan 11 Nov 1998 44N 14 IE 14 Nov 1998 38N 136E 4 727 181.7 Average 16.5-^ 1430^ 103.6 ' Trans-Pacific migration. - Detail file during the traveling phase exists and was used for analyses in Table 2. * The trans-Pacific migration, which was too long in time and distance, was not included in the calculation. comparing measured water temperatures at three depths to the water temperature maps at each depth. However, it is quite difficult to obtain water temperature maps for the whole range of times and areas where tuna migrate other than those for SST. Therefore we decided for the latitudinal estimation to rely on the SST maps and on longitude esti- mated by the tags. One difficulty with the SST method for latitude is that water of the observed surface temperature might occur at two different latitudes, thus not implying a single unique position. However because the latitude of about 80% of all days could be uniquely determined from SST, we considered the adjustment method taken here to be acceptable for the purpose of the present study. Although it was not possible to check the accuracy of the latitudinal estimation independently, judging from the accuracy of the longitude values we used to locate the appropriate North- South stripe on the SST maps and from the rapidity of temperature variations found along those stripes, which in most cases tightly constrained our estimates, we expected the accuracy of latitudinal estimation to be around one de- gree, which would be sufficient for a study of Pacific bluefin tuna migration. Some deterioration was observed in pressure sensors. The need for recalibration of sensors after recovery should be emphasized. Effect of the tag on fish Fish in this study were much smaller than those in other archival tag studies of southern bluefin tuna and Atlantic bluefin tuna (Block et al., 2001; Gunn and Block, 2001). The tagging success achieved confirms that the type of archival tag we used can be applied to fish at least down to 43 cm FL. No fish died because of the attachment of the tag during the experiment on pen-held fish. The recovery rate ( 18.1% ) for fish tagged with archival tags was similar to the rate (19.1%) for those in the conventional tagging experiment conducted in the 1980s off Nagasaki Prefecture, including Tsushima, for the same size fish of the species (Bayliff et al., 1991). This comparison should be made cautiously for the following reasons. The unusual appearance of an archi- val tag body would attract the attention of the finder who gutted the fish and might lead to a higher reporting rate. Increased fishing effort for young Pacific bluefin tuna in the northern part of the East China Sea in the 1990s compared to that in the 1980s might lead to a higher recapture rate. The inconspicuous stalk of an archival tag which was the only externally detectable sign of its existence might lead to a low discovery rate. Indeed, because many recoveries of archival tags were made by consumers while gutting the fish, archival tags implanted in the body of the fish must have been overlooked by fishermen and by sellers at fish markets. However, judging not only by the similar but also high recovery rates, it seems that damage and stress of handling at implantation and that due to the archival tags being carried by the fish did not have much more effect on fish survival than did the conventional tags. Macroscopic observations of two wild fish recovered with archival tags showed that the surgical injuries that occurred during archival tag implantations healed after one month 528 Fishery Bulletin 101(3) Table 2 Comparison of various averaged environmental, physiological, and behavioral values between a phase for young Pacific bluefin tuna. raveling phase and a residency Subject Phase ID offish Average of difference 164 241 319 688 760 Number of days Traveling Residency 9 10 2 10 13 10 12 10 4 10 Swimming depth (m) Traveling Residency Difference 8.7 14.0 -5.3 22.7 15.0 7.7 12.0 45.3 -33.3 22.8 10.2 12.5 7.5 17.4 -9.9 -5.6 Ambient water temperature (°C) Traveling Residency Difference 12.9 17.2 -4.3 17.4 17.9 0.5 12.8 15.6 2.8 19.8 18.7 1.1 16.3 15.3 0.9 -1.1 Temperature of viscera (°C) Traveling Residency Difference 17.8 20.7 -2.9 21.1 21.8 -0.7 17.5 19.1 -1.6 25.9 23.2 2.7 22.7 20.8 1.9 -0.1 Temperature difference between ambient water and fish viscera (°C) Traveling Residency Difference 4.9 3.6 1.4 3.7 3.9 -0.2 4.7 3.6 1.2 6.1 4.5 1.6 6.4 5.5 1.0 1.0 The number of depth records deeper than 150 m per day Traveling Residency Difference 5.4 1.3 4.1 36.0 13.7 22.3 1.6 0.0 1.6 3.7 0.3 3.4 4.3 0.2 4.1 7.1 The number of feeding events per day Traveling Residency Difference 2.6 1.7 0.9 1.0 0.6 0.4 1.6 1.2 0.4 1.7 0.9 0.8 1.3 1.3 0.1 0.5 Percentage of days when a rapid ascent at dawn was observed Traveling Residency Difference 11% 70% -59% 100% 70% 30% 8% 50% -A2% 67% 40% 27% 0% 0% 0% -9% Percentage of days when a rapid descent at dusk was observed Traveling Residency Difference 11% 80% -69% 50% 100% 50% 8% 50% -42% 27% 10% 17% 50% 40% 10% -27% Percentage of days when swimming depth was significantly deeper during daytime than at nighttime Traveling Residency Difference 89% 80% 9% 100% 70% 30% 50%. 90%. -40% 75% 80% -5% 25% 100% -75% -16% Percentage of days when ambient water temperature was significantly lower during daytime than at nighttime Traveling Residency Difference 56% 100% 70% 100% -14% 0% 58% 30% 28% 67% 60% 7% 50% 70% -20% 0% Percentage of days when temperature offish viscera was significantly higher during daytime than at nighttime Traveling Residency Difference 89%. 80% 9% 0%. 20% 20% 92% 100% -8% 75%. 90% -15% 100% 100% 0% -7% Accumulated swimming depth change per day (m) Traveling Residency Difference 5097 5639 -542 6031 4533 1498 6224 6123 101 7884 5266 2618 4576 6386 -1810 373 and there was no scar after a half year. No damage to vis- ceral organs was observed for the two fish. The finding is con- sistent with that for southern bluefin tuna (Gunn et al.'^). Fish tagged with archival tags usually showed similar behavioral patterns, such as diurnal change of swimming depth and vertical excursions at dawn and dusk (Itoh et al., Itoh et al.: Migration patterns of Thunnus onentalis determined with archival tags 529 Table 3 Water temperature changes associated with the travehng phase of Pacific bluefin tuna. Test 1: Circles mark the occurrence of a sea-surface temperature change before the onset of a traveling phase (increasing in spring-summer, decreasing in autumn-winter). T-change (temperature change) shows the maximum (in spring-summer) and minimum (in autumn-winter) water temperature (in the temperature change before traveling. Test 2: Comparison of temperatures at the onset and end of traveling to a temperature range of 14-20°C, which is thought to be a preferred temperature range for young Pacific bluefin tuna. Temperature values within the range of 14-20°C are under- lined. Test 3: Circles mark cases where the temperature at the end of the traveling was a temperature that the fish encountered for the first time since the onset of the traveling Pacific. Traveling phase number Test 1 Test 2 Season ID offish T-change T-onset T-end Tests Spring 209 164 Ml Ml O O 20 19 20.7 14,7 17.0 12.4 O 319 Ml — 16 15.6 13.0 — 241 Ml o 23 21.7 19.1 O Summer 241 M2 o 23 21.9 17.4 o 241 M5 o 20 18.2 19.3 — 688 Ml o 21 19.5 20.7 — 760 Ml o 26 24.0 23.6 — Autumn-winter 241 M3 o 15 14.7 14.3 — 241 M4 — 13 16.0 17.4 o 688 M2 o 17 19.9 22.3 o 760 M2 o 15 14.5 20.1 o 2003 ) from the second day after release, and their feeding frequency reached a constant level one month after release. Also, the fish maintained weight-at-length similar to that of wild fish. The average growth rate of fish tagged with archival tags observed in our study ( 1.3 cm/month) was less than the growth rate of wild fish observed between ages one and twoin previous studies (1.7-3.3 cm/month, Aikawa and Kato, 1938; Yokota et al., 1961; Yukinawa and Yabuta, 1967; Bayliffet al., 1991; Bayliff, 1993; Foreman, 1996), except for the result of Bell (1963b) (1.3 cm/month). Because similar growth rates were observed for fish recaptured more than a half year after release that spent the summer at large, it appears that the lesser growth rate in the present study is not due to the fact that some fish spent only winter at liberty, when the growth rate is less than that in summer (Yukinawa and Yabuta, 1967; Bayliff, 1993). Judging from these facts, we suggest that the effect of archival tags on fish behavior and physiology seems to be minor, although there is a possibility that carrying an archival tag caused a reduction in growth rate of the fish. Horizontal movement Archival tags revealed the movement pattern of young Pacific bluefin tuna individuals, which could be divided into the two clearly-separable phases of traveling and residency. These two phases were observed for all individuals that moved out from the East China Sea. T3 0) 8 12 16 20 24 Time of day (fi) Figure 10 Hourly averages of temperature difference between ambient water and fish viscera in both traveling (circle) and residency phase (triangle) of young Pacific bluefin tuna with archival tags. Data from five individuals are all combined. Bar indicates the standard deviation. The residency phase is considered a normal condition for young bluefin tuna, comprising 83-87% of their time. Fish with archival tags tended to stay in the areas of the East 530 Fishery Bulletin 101 (3) 1 Oct 1997 20 Nov 1997 9 Jan 1998 28 Feb 1998 19 Apr 1998 8 Jun 1998 28 Jul 1998 Figure 11 Changes in ambient water temperature recorded by an archival tag in a young Pacific bluefin tuna. Data are for the fish that traversed the Pacific Ocean. The symbols from Ml to M5 designate from the first to fifth traveling phase of the fish. Each dot is an ambient water temperature at 0 m depth recorded in the summary file of the archival tag each day. Large black and brighter circles are temperatures at the onset and end of traveling, respectively. Arrows indicate the observed ambient water temperature changes several days before the onset of the travel phase. China Sea, off the east coast of Choshi, off the east coast of Hokkaido in the western Pacific Ocean, off Southern California and Baja CaHfornia, off northern California in the eastern Pacific Ocean, and off the west coast between Akita and Hokkaido in the northern Sea of Japan. The first four areas correspond to the major known fishing grounds of young Pacific bluefin tuna. The last two areas do not correspond to previously known fishing areas for young Pacific bluefin tuna. In the eastern Pacific, young bluefin tuna are usually caught in an area from 23 to 34°N, off California to Baja California, from May to October by purse seine (Calkins, 1982). Catch records in the northern area around 40°N were scarce, and all of them were for catches from summer to autumn (Radovich, 1961; BaylifT, 1994). It was not expected that young Pacific bluefin tuna were to be found around 40°N in winter, but the archival tag records showed fish staying in an area of 33-40°N, off northern California, from winter to spring. In the northern Sea of Japan, young Pacific bluefin tuna are usually caught by set nets in coastal areas, and are not caught in the offshore area. The archival tag records again showed fish staying in this area in summer and moving southward without being captured. These cases clearly in- dicate the ability and advantage of archival tags to provide information on fish distribution and migration when and where fishing has not been conducted. An archival tag demonstrated that an individual young Pacific bluefin tuna was able to travel more than 7000 km without pause and to travel for more than two months. The daily moving distance during the traveling phase ranged from 59 to 182 km, and averaged 104 km. Assuming a con- stant swimming speed, the daily average swimming speed was estimated as a range from 1.3 to 4.1 knots (average of 2.3 knots). Itoh et a\ Migration patterns of Thunnus onentalis determined with archival tags 531 These horizontal swimming speeds (1-4 knots) are com- parable to those of larger young Pacific bluefin tuna and same-size fish of other Thunnus species, namely yellowfin tuna ( T. albacares ), bigeye tuna ( T. obesus ), and albacore (T! alalunga), determined from acoustic tracking experiments (Laurs et al., 1977; Carey and Olson, 1982; Holland et al., 1990; Block et al., 1997; Marcinek et al., 2001 ). Sustainable swimming speeds based on oxygen demand and supply for yellowfin tuna and skipjack tuna (Katsuwonus pelamis) of 1.5-2 kg in body weight were estimated to be 2-4 times FL per second (Brill, 1996;Korsmeyeret al., 1996), correspond- ing to 2.3-4.7 knots for fish of 60 cm FL. Applying to young bluefin tuna the same rule (2 -4 times FL) used by those workers as a summary of their data, the expected range of sustainable swimming speeds that do not accumulate an oxygen debt would cover the range of estimated aver- age swimming speeds during traveling phases. Of course, the swimming speed of young Pacific bluefin tuna based on a constant moving speed between two successive daily locations obviously carries some errors. First, a fish might not maintain a constant swimming speed all day long. For example, the daytime swimming speed of albacore observed in an acoustic tracking experiment was reported to be 1.3-2.1 times as great as that at night (Laurs et al., 1977). Second, the influence of water current which should be taken into consideration (Brill, 1996) was completely ignored. Third, assuming straight-line travel between two daily positions would lead to underestimation of actual daily distances traveled, even though the direction during traveling phases could be approximated as a straight line. Even if these errors had been large and the true swimming speed had been twice as large as that which was estimated, these estimated average swimming speeds were still with- in the range of sustainable speeds. Although the horizontal movement clearly differed, many features regarding vertical movement were the same for both residency and traveling phases. One parameter that did differ was that of temperature, where the difference between water and fish viscera was 1.0°C larger during the traveling phase than during the residency phase. Feeding causes an increase of visceral temperature in tuna (Carey et al., 1984; Gunn et al., 2001; Itoh et al., 2003). However, the slightly more frequent feeding in traveling phases was not enough to explain the large thermal excess. In addition, the larger temperature difference was observed not only at daytime when the visceral temperature was usually higher because of more frequent feeding, but also at night when the visceral temperature was usually lower (Itoh et al., 2003). The visceral temperature seemed to be raised by high muscle temperature during traveling phases. If this is indeed the case, this would lead to an increase in the delivery rate of oxygen to muscle, which would make the fish less tired and more able to travel (Stevens and Carey, 1981; Brill, 1996). The distinct traveling phase might be one of the tactics adopted by young Pacific bluefin tuna to use energy most efficiently for long distance travel. During the traveling phase, the frequency of feeding in- creased slightly and the fish dived to water deeper than 150 m depth more frequently. The fish would feed and seek food at least as aggressively as in the residency phase. Ambient water temperature is one of the most important environmental factors for young Pacific bluefin tuna (Sund et al., 1981; Koido and Mizuno, 1989; Ogawa and Ishida, 1989; Itoh et al., 2003). The onset of most traveling phases were preceded by specific water temperature changes that reached the upper or lower limit of the preferable water temperature of 14— 20°C for young Pacific bluefin tuna (Itoh et al., 2003). Changes in ambient water temperature ap- pears to be a possible trigger for a fish to move. Because no remarkable change in frequency of feeding was observed within several days before or after traveling began, the possibility that shortage of prey is a trigger for migration does not seem to be plausible. If the impulse to travel in young Pacific bluefin tuna is regulated only by the search for the preferred water tem- perature range and the aim of traveling is to reach the pre- ferred water temperature range, the ambient temperature would be expected to be out of the preferred range at the onset of travel and within that range at the end. However this did not occur in the fish studied. In addition, half of the observed traveling phases were continued after the fish encountered along the way the same temperature that was present at the end of traveling. According to these results, it appears that the preferred water temperature is neither the sole regulator of traveling in young Pacific bluefin tuna nor is it the sole aim of traveling. Data from tagged fish released around Japan in the 1980s revealed that some fish released from Nagasaki prefecture migrated to the Sea of Japan and others to the Pacific Ocean (Bayliff et al., 1991). Archival tag data showed that fish released in the same season and same area migrated in various patterns involving different onset times and different destinations when traveling from the East China Sea. In addition, some fish continued to remain in the East China Sea. The migration scenario seems not to be fixed or limited for age 0-1 fish distributed around the East China Sea. A detailed examination of fish be- haviors relating to the water temperature has suggested that although young Pacific bluefin tuna prefer a specific temperature range, they can still tolerate temperatures outside of this range (Itoh et al., 2003). This temperature tolerance would contribute to the diversity of migration scenarios for the species. The trans-Pacific migration The trans-Pacific migration of Pacific bluefin tuna was originally validated by tagging tuna both from the west- ern Pacific Ocean to the eastern Pacific Ocean and from the eastern Pacific Ocean to the western Pacific Ocean (Orange and Fink, 1963; Clemens and Flittner, 1969). The duration required for trans-Pacific migration was estimated as 215 days from the shortest interval between the release offish from one side of the Ocean to the recovery of fish on the other side (Bayliff et al., 1991). The present study obtained a full record of daily locations during a trans-Pacific migra- tion of one fish. The fish took two months to traverse the whole Pacific Ocean, which was much shorter than expected from previous records. The starting time for trans-Pacific migration was estimated by Yamanaka (1982) as May- 532 Fishery Bulletin 101 (3) August based on fishery information and as autumn and winter by Bayliff et al. (1991) based on tagging data. The fish observed in our study started its trans-Pacific migra- tion in late autumn. Although data concerning distribution of young Pacific bluefin tuna in the central Pacific Ocean is limited, a record of young Pacific bluefin tuna catch (age 1-3, age 1 mainly) in the area of 35^5°N, 150°E-140°W from April to November has been reported (Saito et al.'^). Moreover, two tagged fish were recaptured in the central Pacific Ocean at 38°N, 172°E in June and 39°N, 162°W in June, respectively (Bayliff et al., 1991). Although the sea- sons differ, the path of the fish tagged with an archival tag passed near these locations. The limited data available at present suggest that the trans-Pacific migration route lies in this area. Together with data which would be obtained from future recovery of additional fish tagged with archival tags, we expect that the overall features of trans-Pacific migration to be revealed in the near future. Acknowledgments We thank the staff of the Marino Forum 21 and the Kagoshima Fisheries Experimental Station for their coop- eration in the pen-held fish experiment. We also thank troll fishermen, staff in the Kamiagata Fisheries Cooperative Association, the Tsushima Fisheries Extension Service, and the Nagasaki Fisheries Experimental Station for their cooperation with tag and release procedures. We acknowl- edge fishermen, consumers, and stafFof the Inter-American Tropical Tuna Commission for their kindness in returning archival tags along with pertinent information about the recapture. We especially thank J. Gunn in CSIRO for giving us valuable information about implanting the archival tag in fish. We are grateful to Northwest Marine Technology Inc. and Tanaka Sanjiro Co., Ltd., for providing us with archi- val tags. We would like to thank P. Ekstrom of Northwest Marine Technology Inc. for his critical review and help with the English text. We thank our staff in Japan NUS Co., Ltd., the Suidosya Co., Ltd., and the National Research Institute of Far Seas Fisheries, and also T Kitagawa in the Ocean Research Institute of the University of Tokyo, for their help in implanting the tags in fish. We gratefully acknowledge S. Kume of Japan NUS Co., Ltd., N. Baba of Fishery Research Agency, and Z. Suzuki and Y. Uozumi of National Research Institute of Far Seas Fisheries for their critical review. Literature cited Aikawa, H. 1949. Fisheries population ecology. SangyoToshoCo. Ltd., Tokyo, 545 p. |In Japanese.] Saito, S., K. Shimazaki, and T. Sato. 198L Distribution of bluefin tuna around the area of polar front in the north Pacific Ocean. In Report of the Maguro Gyogyo Kcnkyu Kyogikai, in 1980, p. 247-252. Nat. Res. Inst. Far Seas Fish. Pelagic Fish Resource Division, 5-7-1 Shimizu-Orido, Shizuoka, Shizuoka, 424-8633, Japan. |In Japanese, the title was translated by authors.) Aikawa, H., and M. Kato. 1938. Age determination of fish. I. Bull. Japan. Soc. Sci. Fish. 7:79-88. [In Japanese.] Anonymous. 1994. Archival tags 1994: present and future. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-357, 42 p. Arnold. G., and H. Dewar. 2001. Electronic tags in marine fisheries research: a 30- year perspective. In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 7-64. Kluwer Academic Publisher, Netherlands. j Bayliff, W. H. \ 1980. Synopsis of biological data on the northern bluefin tuna, Thunnus thynnus (Linnaeus, 1758), in the Pacific Ocean. Spec. Rep. lATTC 2:261-293. 1993. Growth and age composition of northern bluefin tuna, Thunnus thynnus, caught in the eastern Pacific Ocean, as estimated fi-om length-ft'equency data, with comments on trans-pacific migrations. Bull. LATTC, 20:503-540. 1994. A review of the biology and fisheries for northern bluefin tuna, Thunnus thynnus, in the Pacific Ocean. FAO Fish. Tech. Pap. 336/2:244-295. Bayliff, W. H., Y. Ishizuka, and R. B. Deriso. 1991. Growth, movement, and attrition of northern bluefin tuna, Thunnus thynnus, in the Pacific Ocean, as determined by tagging. Bull. LATTC 20:1-94. Bell, R. R. 1963a. Synopsis of biological data on California bluefin tuna Thunnus saliens Jordan and Everman 1962. FAO Fish. Rep. 6(21:380-421. 1963b. Preliminary age determination of bluefin tuna, Thun- nus thynnus. Calif Fish. Game 49:307. Block, B. A., H. Dewar, S. B. Blackwell, T. D. Williams, E. D. Prince, C. J. Farwell, A. Boustany S. L. H. Teo. A. Seitz, A. Walli, and D. Fudge. 2001. Migratory movements, depth preferences, and thermal biology of Atlantic bluefin tuna. Science 293:1310-1314. Block, B. A., J. E. Keen, B. Castillo. H. Dewar, E. V. Freund, D. J. Marcinek, R. W. Brill, and C. Farwell. 1997. Environmental preferences of yellowfin tuna {Thun- nus albacares) at the northern extent of its range. Mar Biol. 130:119-132. Brill, R. W. 1996. Selective advantages conferred by the high perfor- mance physiology of tunas, billfishes, and dolphin fish. Comp. Biochem. Physiol. 113A:3-15. Calkins, T. P 1982. Observations on the purse-seine fishery for bluefin tunaiThunnus thynnus) in the eastern Pacific Ocean. Bull. LATTC 18:123-225. Carey, F. G., J. W. Kan wisher, and E. D. Stevens. 1984. Bluefin tuna warm their viscera during digestion. J Exp. Biol. 109:1-20. Carey, F. G., and R. J. Olson. 1982. Sonic tracking experiments with tunas. ICCAT Coll. Vol. Sci. Papers XVII. 2:458-466. Cayre, P. 1991. Behavior of yellowfin tuna iThunnus albacares) and skipjack tuna iKatsuwonus pelamis) around fish aggregat- ing devices ( FADs ) in the Comoros Islands as determined by ultrasonic tagging. Aquat. Living. Resour 4:1-12. Cayre, P., and F. Marsac 1993. Modeling the yellowfin tuna iThunnus albacares) ver- tical distribution using sonic tagging results and local envi- ronmental parameters. Aquat. Living Resours. 6:1-14. Itoh et al.: Migration patterns of Thunnus onentalis determined with archival tags 533 Clemens, H. B., and G. A. Flittner. 1969. Bluefin tuna migrate across the Pacific Ocean. Calif. Fish Game 55:132-135. CoUette, B. B. 1999. Mackerels, molecules, and morphology. Soc. Fr Ich- thyol. 25:149-164. Foreman, T. 1996. Estimates of age and growth, and an assessment of ageing techniques for northern bluefin tuna, Thunnus thyn- nus. in the Pacific Ocean, Bull. lATTC 21:75-123. Gunn, J., and B. Block. 2001. Advances in acoustic, archival, and satellite tagging of tunas. In Tuna: physiology, ecology, and evolution (B. A. Block and E. D. Stevens, eds.), p.167-224. Academic Press, San Diego, CA. Gunn, J., J. Hartog, and K. Rough 2001. The relationship between food intake and visceral warming in southern bluefin tuna iThunnus maccoyii). Can we predict from archival tag data how much a tuna has eaten? In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L, Nielsen, eds.), p. 109-130. Kluwer Academic Publisher, Netherlands. Hill, R. D. 1994. Theory of geolocation by light levels. In Elephant seals, population ecology, behavior and physiology (B. J. LeBoeuf and R. M. Laws, eds. ), p. 227-236. Univ Califor- nia Press, Berkeley, CA. Hill, R. D.,and M. J. Braun 2001. Geolocation by light level. The next step: latitude. In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 315-330. Kluwer Aca- demic Publisher, Netherlands. Holland, K. N., R. W. Brill, and R. K. C. Chang. 1990. Horizontal and vertical movements of yellowfin and bigeye tuna associated with fish aggregating devices. Fish. Bull. 88:493-507. Hunter, J. R., A. W. Argue, W. H. Bayliff, A. E. Dizon, A. Fonteneau, D. Goodman, and G. R. Sekel. 1986. The dynamics of tuna movements: an evaluation of past and future research. FAO, Fish. Tech. Pap. 277:vi, 78 p. Itoh, T, S. Tsuji, and A. Nitta. 2003. Swimming depth, ambient water temperature prefer- ence, and feeding frequency of young Pacific bluefin tuna ( Thunnus orienlalis) determined with archival tags. Fish. Bull. 101:535-544. Kitagawa, Y., Y. Nishikawa, T. Kubota, and M. Okiyama. 1995. Distribution of ichthyoplanktons in the Japan Sea during summer, 1984, with special reference to scombroid fishes. Bull. Jpn. Soc. Fish. Oceanogr. 59(2):107-114. (In Japanese.] Kitagawa, T, H. Nakata, S. Kimura, T Itoh, S. Tsuji, and A. Nitta. 2000. Effect of ambient temperature on the vertical distribu- tion and movement of Pacific bluefin tuna Thunnus thyn- nus onentalis. Mar Ecol. Prog. Ser. 206:251-260. Koido, T, and K. Mizuno. 1989. Fluctuation of catch for bluefin tuna {Thunnus thyn- nus) by trap nets in .Sanriku coast with reference to hydro- graphic condition. Bull. Jpn. Soc. Fish. Oceanogr. 53: 138-152. [In Japanese.! Korsmeyer, K. E., H. Dewar, N. C. Lai, and J. B. Graham. 1996. The aerobic capacity of tunas: adaptation for multiple metabolic demands. Comp. Biochem. Physiol. 113A:17-24. Laurs, R. M., H. S. H. Yuen, and J. H. Johnson. 1977. Small-scale movements of albacore, Thunnus ala- lunga, in relation to ocean features as indicated by ultra- sonic tracking and oceanographic sampling. Fish. Bull. 75:347-355. Marcinek, D. J., S. B. Blackwell, H. Dewar, E. V. Freund, C. Farwell, D. Dau, A. C. Seitz, and B. A. Block. 2001. Depth and muscle temperature of Pacific bluefin tuna examined with acoustic and pop-up satellite archival tags. Mar Biol. 138:869-885. Metcalfe, J. D. and G. P Arnold 1997. Tracking fish with electronic tags. Nature 387:665- 666. Musyl, M. K., R. W. Brill, D. S. Curran, J. Gunn, J. R. Hartog, R. D. Hill, D. W. Welch, J, P Eveson, C. H. Boggs, and R. E. Brainard. 2001. Ability of archival tags to provide estimates of geo- graphical position based on light intensity. In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds. ), p. 343-367. Kluwer Academic Pubhsher, Netherlands. Nakamura, H. 1965. Tuna resources of the world (II), 52 p. Japan Fisher- ies Resources Conservation Association, Tokyo, Japan. [In Japanese.] Ogawa, Y., and T. Ishida 1989. Hydrographic conditions governing fluctuations in the catch of Thunnus thynnus by set-nets along the Sanriku coast. Bull. Tohoku Reg. Fish. Res. Lab. 51:23-39. [In Japanese.] Okachi, I. 1963. Studies on the distribution and structure of the fish fauna in the Japan Sea by — catch statistics. II. Supple- ment of seasonal distribution and fishing condition of the bluefin tuna. Bull. Jap. Sea Reg. Fish. Res. Lab. 11:9-21. [In Japanese.] Okiyama, M. 1974. Occurrence of the postlarvae of bluefin tuna, Thun- nus thynnus, in the Japan Sea. Bull. Japan Sea Reg. Fish. Res. Lab. 25:89-97. [In Japanese.] Orange, C. J, and B. D. Fink. 1963. Migration of a tagged bluefin tuna across the Pacific Ocean. Calif. Fish Game, 49:307-309. Radovich, J. 1961. Relationships of some marine organisms of the north- west Pacific to water temperatures particularly during 1957 through 1959. Fish Bull. Calif Dep. Fish Game 112: 1-62. Shingu, C, I. Warashina, and N. Matsuzaki. 1974. Distribution of bluefin tuna exploited by longline fish- ery in the western Pacific Ocean. Bull. Far Seas Fish. Res. Lab. 10:109-140. [In Japanese.] Smith, P. and D. Goodman. 1986. Determining fish movements from an "Archival" tag: precision of geographical positions made from a time series of swimming temperature and depth. U.S. Dep. Commer., NOAA-TM-NMFS-SWFC-60, 13 p. Stevens, E. D., and F. G. Carey 1981. One why of the warmth of warm-bodied fish. Am. J. Physiol. 240:R151-R155. Sund, P N., M. Blackburn, and F Williams. 1981. Tunas and their environment in the Pacific Ocean: a review. Oceanogr. Mar. Biol. 19:443-512. Ueyanagi, S. 1969. Observations on the distribution of tuna larvae in the Indo-Pacific Ocean with emphasis on the delineation of the spawning areas of albacore, Thunnus atalunga. Bull. Far Seas Fish. Res. Lab. 2:177-256. [In Japanese.] 534 Fishery Bulletin 101(3) Welch, D. W., and J. P. Eveson 1999. An assessment of light-based geoposition estimates from archival tags. Can. J. Fish. Aquat. Sci. 56:1317-1327. Yabe, H., S. Ueyanagi, and H. Watanabe. 1966. Studies on the early life history of bluefin tuna Thun- nus thynnus and on the larvae of the southern bluefin tuna T. maccoyii. Rep. Nankai Reg. Fish. Res. Lab. 23:95-129. (In Japanese.) Yamanaka, H., 1982. Fishery biology of the bluefin tuna resource in the Pacific Ocean, 140 p. Japan Fisheries Resources Conser- vation Association, Tokyo, Japan. |In Japanese.] Yokota, T., M. Toriyama, F. Kanai, and S. Nomura. 1961. Studies on feeding habit of fishes. Rep. Nankai Reg. Fish. Res. Lab. 14:1-234. (In Japanese.] Yonemori, T. 1989. To increase the stock level of the highly migrated pelagic fish. In Marine ranching (Agriculture, Forestry and Fisheries Research Council secretariat, eds.), p. 9-59. Koseisha-Koseikaku, Tokyo, Japan. Jin Japanese, the title was translated by authors.] Yorita, T. 1976. Bluefin tuna in the Sea of Japan in Hokkaido. Monthly report of Hokkaido Fishery Experimental Station, 33(3): 2-11. [In Japanese.] Yukinawa, M., and Y. Yabuta. 1967. Age and growth of bluefin tuna, Thunnus thynnus (Linnaeus), in the north Pacific Ocean. Rep. Nankai Reg. Fish. Res. Lab. 25:1-18. [In Japanese.] 535 Abstract — We investigated the migra- tion and behavior of young Pacific blue- fin tuna {Thunnus orientalis) using archival tags. The archival tag mea- sures environmental variables, records them in its memory, and estimates daily geographical locations based on measured light levels. Of 166 archival tags implanted in Pacific bluefin tuna that were released at the northeastern end of the East China Sea from 1995 to 1997, 30 tags were recovered, including one from a fish that migrated across the Pacific. This article describes swimming depth, ambient water temperature, and feeding frequency of young Pacific blue- fin tuna based on retrieved data. Tag performance, effect of the tag on the fish, and horizontal movements of the species are described in another paper. Young Pacific bluefin tuna swim mainly in the mixed layer, usually near the sea surface, and swim in deeper wa- ter in daytime than at nighttime. They also exhibit a pattern of depth changes, corresponding to sunrise and sunset, apparently to avoid a specific low light level. The archival tags recorded temperature changes in viscera that appear to be caused by feeding, and those changes indicate that young Pacific bluefin tuna commonly feed at dawn and in the daytime, but rarely at dusk or at night. Water temperature restricts their distribution, as indicated by changes in their vertical distribution with the seasonal change in depth of the thermocline and by the fact that their horizontal distribution is in most cases confined to water in the temperature range of 14-20°C. Swimming depth, ambient water temperature preference, and feeding frequency of young Pacific bluefin tuna (Thunnus orientalis) determined with archival tags Tomoyuki Itoh Sachiko Tsuji National Research Institute ol Far Seas Fisheries 5-7-1 Shlmizu-Ondo, Shizuoka Shizuoka, 424-8633, Japan Email address (for T Itoh): ltou(3'fra.affrc,go.|p Akira Nitta Japan NUS Co, Ltd. Loop-X BIdg. 3-9-15 Kaigan, Minato Tokyo, 108-0022, Japan Manuscript approved for publication 22 October 2002 by Scientific Editor Manuscript received 3 January 2003 at NMFS Scientific Publications Office. Fish Bull. 101:535-544 (2003). Swimming behavior oi Thunnus species and its relation to various environmen- tal factors have been examined mainly by acoustic tracking (e.g. Carey and Lawson, 1973; Laurs et al., 1977; Carey and Olson, 1982; Holland et al., 1990b; Cayre, 1991; Cayre and Marsac, 1993; Block et al., 1997). Acoustic tracking has also been applied to young Pacific bluefin tuna iT. orientalis): to one fish tracked for three hours around Japan (Hisada et al.'), and to six fish tracked for several days each in the eastern Pacific Ocean (Marcinek et al., 2001). Acoustic tracking can monitor fish move- ment, behavior, and even physiological conditions on a time scale of seconds. However, the duration of monitoring any one fish is generally limited to several days at most because of the short life of the tracking transmitter This limitation, together with the high cost of adequate ship-time, generally makes it difficult to monitor the behavior of a large number offish over a long period of time. An archival tag is an electronic device that measures environmental variables and records raw or processed data in its memory. The archival tag can monitor animal behavior, its physiological con- ditions, and the several environmental factors that the animal is actually ex- periencing, simultaneously. Data can be collected for a much longer period than with acoustic tracking, if the tags are suc- cessfiilly retrieved. Recently, a type of ar- chival tag that can estimate geographical locations has been developed. This type of tag has been applied to southern blue- fin tima T. maccoyii (Gunn and Block, 2001) and Atlantic bluefin tuna T. thyn- nus (Block et al., 1998a, 1998b). These reports show the remarkable value of the archival tag data for investigating fish migration and behavior. We have implanted archival tags in young Pacific bluefin tuna since 1994 to investigate their migration and be- havior This article describes the results obtained from recovered tags and places special emphasis on vertical swimming behavior, preferred water temperature, and feeding frequency. Some insights regarding vertical swimming depth have already been reported in Kitagawa et al. (2000) who used some of the same data. We have described in another pa- per (Itoh et al., 2003) the performance of the archival tag used in the present study and the characteristics of young Pacific bluefin tuna migration based on data from these same tags. 1 Hisada, K., H. Kono, and T. Nagai. 1984. Behavior of young bluefin tuna during migration. In Progress report of the marine ranching project 4, p. 1-7. Nat. Res. Inst. Far Seas Fish. Pelagic Fish Resource Division, 5-7-1 Shimizu-Orido, Shizuoka, 424-8633, Japan. (In Japanese, the title is translated by the authors.] 536 Fishery Bulletin 101 (3) Materials and methods The archival tag used in the present study (Northwest Marine Technology, Inc., Shaw Island, WA) had four sen- sors— for external temperature, internal temperature, pressure, and light intensity. The external and internal temperature sensors had a 0.2°C resolution and response times of three seconds and 20 seconds, respectively. Resolu- tion of the pressure sensor was 1 m of depth between the surface and 126 m, then 3 m down to the scale limit of 510 m. The tags measured data every 128 seconds. Two types of data files were created within the tag memory. One data file stored daily records containing date, estimated times of sunrise and sunset, water temperatures at 0 m, plus two other selectable depths (we selected 60 m and 120 m), and other information required or produced in the course of location estimates for each day. This file is re- ferred to as the "summary file" in this article, and it stored daily data from the time the memory was last cleared. The second data file contained unprocessed time series data records taken at 128-second intervals. The tag could record at any integer multiple of its measurement interval and a multiple of one was chosen. Each record consisted of external temperature, internal temperature, pressure, and light intensity, and corresponded to a known time. This file is referred to as the "detail file" in this article. It can hold about 54,000 records, or about 80 days of steady data at the high rate chosen — a small fraction of the tags lifetime. The time-series memory was divided into two sections, and the size allocation between the two sections was determined by the user The first section filled first and did not change thereafter. The second section filled next, but once full, it is then continually overwrote old data. Thus the first section always contained the earliest data seen in a tag deploy- ment and the second always contained the latest data. We divided the detail file into two 40-day sections for releases in 1995 and 1996, and into 20- and 60-day sections for releases in 1997. Most of the analyses in this study were conducted with the detail file. Prior to experiments on wild fish, we applied archival tags to three pen-held Pacific bluefin tuna from 93 to 97 cm in fork length ( FL I at Kasasa in Kagoshima Prefecture (31°25'N, 130°11'E) in November 1994 to observe the effect of archival tag attachment and implantation on fish. One of the fish that had an archival tag inserted in its abdominal cavity was recovered 453 days after tag implantation, when the fish was caught for sale in the market. Experiments on wild young bluefin tuna were conducted near Tsushima at the northeastern end of the East China Sea every November and December from 1995 to 1997. A to- tal of 166 fish, ranging from 43 to 78 cm FL (age 0 or 1 ), were caught by chartered trolling vessels, tagged on the vessel by inserting archival tags into their abdominal cavities, and re- leased immediately. Details of the tag, its performance, and the manner of tagging are described in Itoh et al. (2003). Thirty of the 166 archival tags (18.1%) were recovered. The durations of the tags at sea were 50 days or less for 13 fish, 96-211 days for 13 fish, and 359-375 days for three fish, all recaptured around Japan. One additional fish was recaptured off the west coast of Mexico, on the east side of the Pacific Ocean, 610 days after release. Data could not be downloaded from one archival tag released in 1995 and re- covered 30 days after release; all other tags yielded data. Results Sample records of swimming depth, water temperature, and temperature offish viscera as recorded in the detail file are shown in Figure 1 for three days in winter and three days in summer The fish changed swimming depth fre- quently with rapid dives and ascents. The water tempera- ture changed little in winter, but it changed frequently and substantially corresponding to dives in summer. Visceral temperature changed gradually. Diurnal and seasonal change of swimming depth Differences between daytime and nighttime swimming depth, water temperature, visceral temperature, and the temperature difference between water and fish viscera (thermal excess) were examined by using the Mann- Whit- ney Utest (P=0.05) with data in the detail file. Data taken during the hour before and the hour after both dawn and dusk (four hours in all) were excluded from the test in order to distinguish clearly between daytime and night- time. For this purpose, dawn and dusk were taken as the times the tag sensed the first or final light of the day. The average swimming depth was significantly deeper during daytime for 70% of all recorded days, which accounts for the additional observation that the water temperature was significantly lower during daytime for 66% of the days. The visceral temperature was significantly higher during day- time for 71% of the days, and thermal excess was signifi- cantly larger during daytime for 85% of the days. Fish spent about 40% of their time within a 0-9 m depth range and the time spent within each depth interval de- creased as depth increased. This concentration in the 0-9 m depth range was observed at both daytime and nighttime, but was more pronounced at night (Fig. 2). The vertical thermal profiles (Fig. 2) show the change of depth range of the surface mixed layer, the ocean layer that lies above the seasonal thermocline, for one year Although young Pacific bluefin tuna aggregated in the 0-9 m depth range for almost all months, swimming depth was more broadly distributed in winter when the depth range of the mixed layer was greater (e.g. January and March). When the depth range of the mixed layer became less in summer (e.g. May and July), fish tended to concentrate near the sea surface. Then as the depth range of the mixed layer became greater in autumn (e.g. September and November), the vertical distribution of the tuna gradually expanded toward deeper water. Vertical swimming behavior at dawn and dusk The fish commonly showed a distinctive vertical movement at dawn and dusk. At dawn, after a slow and steady descent for about 40 minutes to reach to a maximum depth of 82 ±28 m (average ±SD), fish suddenly and rapidly ascended Itoh et al.: Swimming depth, water temperature preference, and feeding of Thunnus onentalis 537 12 16 Time of day (h) Figure 1 Sample records of swimming depth (thick solid line), water temperature ( lower thin solid line), and temperature of viscera (upper thin solid line) for Pacific bluefin tuna from the detail files of two archival tags. Upper panel: records of a fish from 23 to 25 December 1995 in the East China Sea. Lower panel: records of another fish from 15 to 17 June 1997 in the East China Sea. Shadows indicate nighttime. to near the sea surface (upper panel in Fig. 1, and Fig. 3). At dusk, after a rapid descent from near the sea surface to the maximum depth of 89 ±34 m, fish slowly and steadily ascended for about 40 minutes. The time that maximum depth was recorded was most frequently at four minutes before (or after) the time when the archival tag sensed the first (or final) light at dawn (dusk) for 29% (37%) of cases where this behavior was observed. The end (onset) of rapid ascents (descents ) occurred most frequently at the time when the tag sensed the first (final) light, that is to say for 68% (49%) of all cases. The light detection threshold for the tags used in our study corresponded to an intensity in the blue- green transmission window of seawater (470 nanometers (nm) center, 60 nm width of the filter) of about 3 x 10"'' times the surface noon solar intensity on a clear day in that same spectral band. Because the light-sensitive region of the measurement stalk was located below the animal's body, we assumed that the eyes encountered a somewhat higher intensity than did the light sensor. Quick examination of tags in air in the early October revealed that the time when the archival tag sensed the first light at dawn or the final light at dusk was about 40 minutes before sunrise or about 40 minutes after sunset, respectively. Some combination of these vertical swimming behaviors was observed in 1081 of 1452 days (74%) during which the detail file data were available for both dawn and dusk. They occurred at both dawn and dusk in 679 days (47%), only at dawn in 77 days (5%), and only at dusk in 325 days (22%). Occurrences of these behaviors differed by area and sea- son (Table 1). The area was determined by using fish loca- tions estimated as described in Itoh et al. ( 2003 ). The average occurrence of these depth excursions at dawn and dusk was as high as 87-88% in the East China Sea from November to January but decreased to 15-49% from February to June and October. In the Sea of Japan, the average occurrence was 4-39% in April, May, and September to November. In the Pacific Ocean, the behaviors were observed in 75-90% of days from May to July, in contrast to the relatively low occur- rences within the East China Sea in the same season. Water temperature Water temperatures recorded in archival tags ranged from 8.3° to 28.4°C at 0 m depth, and 1.4° to 28.4°C when all depths were combined. A range of water temperatures that appeared to be preferred by young Pacific bluefin tuna was estimated. A simple frequency distribution of recorded water temperatures was inappropriate because fish could be forced to tolerate water out of their preferred tempera- ture range because of a lack of water with a more favorable temperature within the geographical range accessible to the fish. Instead we compared the range of water temperature within the geographical area accessible to the fish with the frequency distribution of actual recorded water tempera- tures, grouped by year, month, and sea. We assumed that the accessible areas in each sea were those areas that a tag reported as its position. These include areas along the Japanese coast between 35° and 45°N in the Sea of Japan, 30-45°N in the western Pacific Ocean west of 160°E, and 25-45°N in the eastern Pacific Ocean east of 160°E. In 538 Fishery Bulletin 101(3) B 10 a. a 90% 9C % 0% 90 (1 1 |J 1 January 50 100 1 15 20 25 10 0% 90% 90% 0 1 ■ >» 1 50 / July 00 ^ 50 100 9 0% 0% 9C n 1 t 1 1 i ■ / March 50 ' 1 1 * 00 f 15 20 25 90% September 10 15 20 25 10 15 20 25 Temperature (°C) 90% 50 100 10 90% 10 0% 90% May 25 90% 0 1 1 1 1 1 1 1' 1 \ ^ i 50 1 9 November 00 ' 1 ' ' 1 15 20 25 Figure 2 Vertical occurrence frequencies of young Pacific bluefin tuna in daytime (clear barl and nighttime (shaded bar) and vertical thermal profile (line with dots) by month. 120 -> Time (minutes) 100 120 Figure 3 Average swimming depth of young Pacific bluefin tuna recorded by archival tags at dawn (left panel) and dusk (right panel). Times marked as 0 are the time just after the end of rapid ascent at dawn and just before the onset of rapid descent at dusk. All of records in which a rapid ascent or rapid descent occurred were averaged. Bar shows standard deviation. the East China Sea, the area considered was enclosed by four points of 29°N-126°E, 29°N-128°E, 35°N-130°E, and 33°N-126°E, where the majority of estimated tag locations occurred. The temperature range in each area was derived from the sea-surface temperature maps published by the Japan Fisheries Information Service Center. Itoh et al.: Swimming depth, water tempeiature preference, and feeding of Thunnus orientalis 539 Table 1 The number of days during which swimming behaviors with rapid ascent at dawn and rapid descent at dusk were observed by area and month. Number Dawn Dusk Average Total Observed Observed Total Observed Observed Area Month offish days days rate days days rate observed rate East China Sea 1 7 87 73 84% 86 77 90% 87% 2 1 23 6 26% 22 16 73% 49% 3 2 36 11 31% 37 12 32% 31% 4 7 66 24 36% 67 41 61% 49% 5 10 220 51 23% 221 128 58% 41% 6 8 182 17 9% 176 37 21% 15% 10 1 24 4 17% 24 6 25% 21% 11 23 132 109 83% 153 142 93% 88% 12 29 487 388 80% 489 457 93% 87% Sea of Japan 4 1 13 0% 12 1 8% 4% 5 1 9 1 11% 9 1 11% 11% 9 1 12 0% 12 4 33% 17% 10 2 61 10 16% 61 22 36% 26% 11 2 43 15 35% 42 18 43% 39% Pacific Ocean 5 1 16 11 69% 16 13 81% 75% 6 2 26 19 73% 26 25 96% 85% 7 1 31 25 81% 31 31 100% 90% Total 1468 764 52%- 1484 1031 69%' 61% The tag temperature used in our comparison was the temperature at 0 m depth recorded in the summary file because that file contained a much larger number of days than that of the detail file. In support of this, we confirmed that the average water temperature over all depths for a day in the detail file had only slight differences of -0.1 ±0.7°C in average (range: -4.0-+4.3°C) from the tem- perature at 0 m recorded for the day in the summary file. Because the fish swam near the surface, the temperature recorded to represent 0 m depth also represented well the temperature at all depths where the fish actually swam. Frequencies of days were summed from the data for all individuals in one-degree temperature bins separated by year, by month, and by area, such as the East China Sea, the Sea of Japan, and the Pacific Ocean. The water temperature recorded by archival tags com- monly ranged from 14° to 20°C (Fig. 4). When water of this temperature range was located within an accessible area (e.g. many months in the East China Sea and the Pacific Ocean, and November in the Sea of Japan), almost all fish were found in such water. Where water temperature was higher (e.g. June 1996, June 1997, and between June and October 1998 in the East China Sea) or lower (e.g. May 1996 and April 1997 in the Sea of Japan), fish tended to choose water that was close to this range. However there were a few cases in which fish stayed in water with a temperature outside of this range (e.g. between July and September 1998 in the Sea of Japan) even when water of the 14-20°C range was accessible to them. Visceral temperature and feeding events Temperature of the fish viscera ranged from 13.0° to 30.9°C. It was usually higher than the water temperature and the thermal excess for a given individual ranged from 1.3° to 4.6°C, and averaged 3.0° ±1.0°C. The visceral tem- perature changed in parallel with the water temperature all year (Fig. 5). In preliminary experiments, pen-held fish were fed com- pletely thawed mackerel of 20-30 cm FL twice a day at approximately 900 h and 1500 h. The following changes in visceral temperature before and after feeding were generally observed (Fig. 6). All figures given below regarding thermal excess and timing after feeding are average values. The vis- ceral temperature in a stable state just before feeding (n=5) was 3.7°C higher than ambient water temperature. This thermal excess decreased to 2.3°C at 22 minutes sifter feed- ing, and then increased to 7.6°C at 7.7 hours after feeding. Then it slowly decreased again and reached a stable thermal excess of 3.1°C at 21.0 hours after feeding. When the fish fed again before the visceral temperature stabilized (n=7), with the thermal excess still as high as 6.8°C, the thermal excess reached a high of 8.5°C but the increase after feeding was 1.7°C, much smaller than that observed in the previous case (3.9°C ). The time required to reach the highest visceral temperature and the time to change back to a stable state were similar to the previous case. When the fish were not fed, because of rough sea conditions, the visceral temperature stayed stable all day (e.g. 16 January on Fig. 6). 540 Fishery Bulletin 101(3) Area Year- Number Temperature (°C) month of tish 5 6 718 9 10 11 12 13 14 15 16 17 18 19 20 21 ?? 23 24 25 26 27 28 29 30 '95-12 7 3 68 49 1 •96-1 4 9 20 29 24 13 7 ■96-2 3 1 7 25 19 14 9 8 2 3 ■96-3 3 1 , 5 18 23 17 1 3 1 ■96-4 2 3 15 14 11 11 5 1 1 ■96-5 2 1 3 2 9 4 6 6 3 ■96-6 ' 1 ' ' 2 10 16 il ; 1 ■96-11 12 , ! 2i 1 7 14 ■96-12 12 1' 17 94 197 17 •97-1 1 8 2 5 17 48 86 67 1 •97-2 1 7 3 17 91 44 32 9 S ■97-3 7 2 82 59 35 35 4 "" ■97-4 7 63 73 22 37 1 1 i ■97-5 5 11 95 32 6 ■97-6 4 24 24 22 23 5 1 1 ^97-11 10 1 14 70 15 i" '97-12 10 1 12 122 122 18 i 1 ■98-1 B 1 21 40 SO 70 23 10 4 J ■98-2 7 10 77 57 31 11 4 \ \ 1 ■98-3 6 2 39 75 44 23 3 •98-4 6 1 11 37 65 64 2 '98-5 6 , 3 28 82 54 15 4 ■98-6 6 14 53 35 14 13 1 , : 1 ■98-7 2 1 3 7 15 15 2i ; ■98-8 1 3 3 8 9 3 3 2 ■98-9 1 4 12 11 3 ■98-10 1 2 3 8 13 5i : 1 ■96-11 1 2 2 6 3 4 7 1 3 11 i 1 ■96-5 1 . 1 • : 2l 2| 2 2| . '97-1 1 1 11 21 2 1 3 -2] 1 1 g. '98-6 ' 1 ^ 5 1 " ■98-7 2 1 1 I 4 12 18 8 4 4 1 o ■98-8 2 1 1 1 1 1 1 26 15 16 3 1 1 ■gs-g 2 1 4 15 9 1 1 13 15 1 ■98-10 2 1 5 8 15 10 6 1 13 2 1 ■98-11 2 1 4 5 2 3 11 6 6 1 2 ■96-3 1 1 ' . 1 2 3 8 4 2 5 ■96-4 1 Ill 2 4 5, 2 12 5 ■96-5 1 1 : , i 1 1 1 1 4: 17 9 1 ■96-6 1 'III 1 1 1 1 ! 1 15 1 1 ■97-5 1 ' 111 1 12 11 3 2 2 i! — •97-6 ! ! 1 ■ 1 7 12 10 •97-7 2 2 12 6 9 % '97-8 111 2 9 8 6 2 4 S '97-9 13 2 4 8 4 9 ; u '97-10 ^ '97-11 -^- 1 1 ' 11 9 8 5 6 1 1 -^ 1' 9 7 2l 2 1 1 7 S. '97-12 3 5 2 1 8 9 3 1 '98-1 2 7, 9 5, 61 2 ■98-2 1 14 9 1 i: 3i 1 ■98-3 ' 20i 10 r '• ■98-4 8 19 2 1 ■98-5 2 7 9 2 1 1 2 4 3 '98-6 1 1 1 1 ( 4 5 2 11 7 1 1 i i '98-7 1 1 1 1 I 1 1 4 10 12 3 III Figure 4 Frequencies of sea-surface water temperatures for young Pacific bluefin tuna with archival tags. The number given is the sum (over all individuals) of days in the category based on temperature at 0 m depth from the summary files. Gray, thick bars indicate the sea-surface temperature range that appeared to be accessible to the fish in each month. In the detail files obtained from wild fish, the following three types of visceral temperature changes were identified as indicators of feeding ( Fig. 7 1. Type A: a sharp decrease of 1-2°C followed by an increase above the initial temperature. Type B: an increase above the initial temperature, preceded by either no decrease or a slow decrease. Type C: a sharp decrease of 1-2°C followed by a return to the initial temper- ature. In addition, in only cases where the visceral tempera- ture changes could not be explained by water temperature changes were they counted as being caused by feeding. For example, a slow decrease of visceral temperature when the fish dived into cold water was not counted as a feeding event. Cases where the water temperature changed frequently were also excluded as too difficult to interpret. Feeding events were observed in 942 days out of 1494 to- tal recorded days (63% ). Because it was found that fish did not feed normally for approximately the first 30 days after release (Itoh et al., 2003), data from the first 60 days after release were excluded in the following description. After exclusion of data for the first 60 days, feeding events were observed in 726 out of the remaining 807 recorded days (90%). The number of daily feeding events ranged from zero to ten, with an average of 1.8 ±1.4. Feeding events oc- curred most frequently in the daytime (69% of all obser\'ed feeding events more than 60 days after release), followed by dawn with 27% (Fig. 8). Feeding events rarely occurred at dusk and at night, accounting for only 1% and 3%^ of the total feeding events, respectively. For this analysis, dawn Itoh et al.: Swimming depth, water temperature preference, and feeding of Thunnus onentahs 541 and dusk were defined as explained above, a one-hour pe- riod centered on the time of first or last detected daylight. Changes in visceral temperature of "type B" (increase only) were observed most frequently (51.4% of total observed feedings), followed by "type C" (decrease only, 45.2%). Bi- polar events (type A) were as few as 3.5%. When averaged over individual months, the number of feeding events per day per individual ranged from 0.9 in January to 2.2 in June and averaged 1.5. Feeding events were observed all year, although they were slightly more frequent in May and June than in other months (Fig. 9). Discussion Diurnal and seasonal change of swimming depth Young Pacific bluefin tuna were previously assumed to swim near the sea surface based on the fact that most of the catch was made by surface fishing gear and fish schools were observed at the sea surface (Yabe et al., 1953). However, details of their vertical swim- ming behavior and relationships between their behaviors and oceanic structures have not been well investigated. Recently, Marcinek et al. (2001) observed during an acoustic tracking experiment over several days that Pacific bluefin tuna in the eastern Pacific Ocean spent the majority of their time in the top portion of the water column. Our archival tag data showed that young bluefin tuna in the western Pacific Ocean also ordinarily stayed within the surface mixed layer and most frequently near sea surface, regardless of the time of day or the season. The vertical distribu- tion of fish changed according to the seasonal change in depth range of the surface mixed layer and appeared to be controlled by the depth of the thermocline. Restriction by the thermocline was also observed for yellowfin tuna (T. albacares) and bigeye tuna (T! obesus) (Carey and Olson, 1982; Holland et al., 1990b; Cayre and Marsac, 1993; Block et al., 1997). Occasionally young Pacific bluefin tuna dived through the thermocline into deep, cooler water, but they returned to the surface mixed layer after a short period. Diurnal change in swimming depth, i.e. deeper swimming depth during daytime, was reported by acoustic tracking studies not only for Thunnus species, such as yellowfin tuna (Carey and Olson, 1982;Hollandetal.,1990b;Cayre,1991;Yonemori-) and bigeye tuna (Holland et al., 1990b), but also for other large pelagic species, such as skipjack tuna, Katsuwonus pelamis (Yuen, 1970; Dizon et al., 1978); swordfish, Xiphias gladius (Carey and 30 o 25 20 15 10 3 4 5 6 9 10 11 12 Month Figure 5 Monthly change of average water temperature (O) and average visceral temperature (•) in young Pacific bluefin tuna with archival tags. Average values of each individual were averaged. ^Yonemori.T. 1982. Swimming behavior of tunas by the use of sonic tags — a study particularly of swim- ming depth. Far Seas Fish. Res. Lab. Newsletter 44:1-5. Pelagic Fish Resource Division, 5-7-1 Shi- mizu-Orido, Shizuoka, Shizuoka, 424-8633, Japan. [In Japanese.] 29 - 27 - 9 25- i 23- 1 21- CL £ 19 - ^ 17- 15 - 0 Viscera of a pe Triang l/n/YV [ k A h 12h Oh 12h Oh 12h Oh 12h Oh 12h Oh 12h Oh 12 Jan 13 Jan 14 Jan 15 Jan 16 Jan 17 Jan Figure 6 il temperature (thick line) and water temperature (thin line) n-hold young Pacific bluefin tuna recorded by an archival tag. les indicate the time of feeding. o 25 20 - E 15 - 10 C A C C C C T^'^^^^^rtT- Oh 2h 4h 6h 8h lOh 12h 14h 16h 18h 20h 22h Time of day (h) Figure 7 An example of visceral temperature change in a wild young Pacific bluefin tuna recorded by an archival tag. Visceral temperature (thick line) and water temperature (thin line) are shown. Shadows indicate nighttime. Data are from a fish in the Sea of Japan on 2 May 1996. A, B, and C indicate the types of visceral temperature changes described on page 540. 542 Fishery Bulletin 101(3) Robison, 1981); blue marlin, Makaira nigricans (Holland et al., 1990a); mako shark, Isiirus oxy- rinchus; and blue shark, Prionace glauca (Carey and Scharold, 1990). However some reports did not note any difference in swimming depth be- tween day and night, such as that of Block et al. (19971 for yellowfin tuna, Cayre (1991) for skip- jack tuna, and Brill et al. (1993) for striped mar- lin, Tetrapturus audax. The swimming depth of young bluefin tuna recorded by the archival tags in the present study was deeper during the day for lQ'7i of recorded days. This finding agrees with the speculation made by Carey and Olson (1982) that a deeper swimming depth in the daytime is a common feature for large pelagic fish. Vertical swimming behavior at dawn and dusk A characteristic vertical movement pattern was found in young Pacific bluefin tuna. They dived gradually and constantly for about 40 minutes and then rapidly ascended to the sea surface at dawn. Inverse behavior were observed at dusk. The same behavior was reported in larger size Pacific bluefin tuna in the eastern Pacific Ocean and in yellowfin tuna ( Block et al., 1997; Marcinek et al.. 2001). The percentage of days when this behavior was observed varied according to the season and area. This variation was commonly observed in individuals as well as in the group as a whole. Two potential reasons for this behavior were considered. The first one was avoidance of a spe- cific range of light intensities. The times of onset and end of the behavior are apparently related to the time of sunrise and sunset. Assuming that young Pacific bluefin tuna dislike a specific light intensity range, we describe their vertical move- ments at dawn and at dusk as follows. About 80 minutes before sunrise when the light intensity at the sea surface reaches a specific value near the lower boundary of the avoided range, fish begin descending into water with lower light intensity. About 40 minutes before sunrise when the light intensity in deep water reaches that avoided range, the fish rapidly ascend almost to sea sur- face, and as the light brightens further, gradually expand into their normal distribution pattern while staying within the range of water depths where light intensities exceed the avoided low-intensity range. A possible reason for avoiding a specific intensity range might be an increased risk of predation at intensities where tuna see less well than some predator that hunts visually. The other potential reason for the characteristic vertical movements is feeding. It is well known that some small animals show diurnal vertical migration, i.e. they descend gradually as the light level increases toward dawn and rise again at dusk. Young Pacific bluefin tuna following these species to feed on them would show similar behavior How- ever, young Pacific bluefin tuna were observed to feed only 0% 10% 20% Frequency 30% 40% 50% 60% 70% Dawn Daytime Dusk Nigtittime DA □ B ■ C Figure 8 Frequency of feeding events of young Pacific bluefin tuna by period within a day. Only data taken more than 60 days after release were used. A, B, and C indicate the types of temperature changes of viscera described on page 540. 2 - 1 - 1 2 8 12 10 1 3 3 () <> 1 C) () <> () <> " O 6 7 Month 10 11 12 Figure 9 Average frequency of feeding events by month for young Pacific blue- fin tuna. Only data taken more than 60 days after release were used. The number above each point indicates the number of individuals contributing to the average. Bars show standard deviations. at dawn, not at dusk, although the characteristic vertical behavior was observed at both dawn and dusk. In addition, rapid ascents and descents at a specific time with respect to sunrise and sunset could not be explained by the verti- cal migration behavior of bait species. Therefore, feeding seems not to be a primary cause of the vertical migration in young Pacific bluefin tuna. Generally speaking, fishermen consider dawn and dusk to be good times for catching Pacific bluefin tuna. The ar- chival tag records showed that young Pacific bluefin tuna did not usually feed at dusk, although tag records showed that fish aggregated very close to the sea surface after their rapid ascent at dawn and before their rapid descent at dusk. Judging from this behavior, good fishing seemed to be caused by a concentration offish near the sea surface rather than by the feeding activities Moreover, the low light level at these times would make it difficult for fish to distinguish between artificial bait with a hook and live prey. Itoh et al.: Swimming depth, water temperature preference, and feeding of Thunnus orientalis 543 Preferred water temperature Water temperature is thought to be one of the most impor- tant environmental factors controlhng the distribution of young Pacific bluefin tuna (Sund et al., 1981; Koido and Mizuno, 1989; Ogawa and Ishida, 1989). Kitagawa et al. (2000) attached importance to the gradient of water tem- perature; however Uda (1957) emphasized the absolute value of temperature, although his study was presumably for large-size fish. Data from the archival tags indicated that young Pacific bluefin tuna seemed to prefer to remain in water of 14-20°C. When there was no accessible water within this temperature range, the fish tended to stay in water of a temperature as close as possible to this range. In addition, archival tags showed that the vertical distri- bution of young Pacific bluefin tuna was restricted by the thermocline, even when the temperature below the ther- mocline was in the tunas preferred temperature range (14-20°C). These observations support the importance of water temperature as shown in previous studies and suggest that both the absolute value and the gradient of water temperature are important as environmental factors controlling the distribution of young Pacific bluefin tuna. Feeding Visceral temperature of pen-held Pacific bluefin tuna with archival tags changed in a specific way during feeding. Stomach temperature changes have also been observed in pen-held giant bluefin tuna in the Atlantic and in pen- held southern bluefin tuna (Carey et al., 1984; Gunn et al., 2001). The cycle of visceral temperature change for young Pacific bluefin tuna was completed in 21 hours (shorter than that observed in previous studies of 1.5 to 2 days) probably due to the smaller size of the fish. Similar vis- ceral temperature changes were also noted in the records of archival tags recovered from wild fish, ranging from a distinct pattern the same as that observed in pen-held fish (type A) to less distinct ones such as type B or type C, which were observed more frequently. All of these changes could be distinguished quite easily from gradual decreases of vis- ceral temperature when fish dived into cold water There- fore, we assumed that these three types of temperature changes were caused by feeding. Temperature changes of type A could be expected if fish consumed a large amount of food at one time as they do when fed in a pen. However, wild fish may seldom have an opportunity for such large meals, and the apparently more frequent small meals would be expected to cause the less dramatic visceral temperature changes of types B or C. In the present study, a visceral temperature change was taken to indicate feeding only when that change could not be explained by a change in water temperature. Also, when the water temperature changed very frequently, it was dif- ficult to decide whether water temperature could account for a feeding event and it was not counted as such. Finally it is possible that feeding might not cause a recognizable change in visceral temperature. As a result of these three factors the feeding frequency estimated in our study might have been underestimated. Frequencies of feeding events did not change much over the year, although there was a slightly higher frequency in early summer Growth in length of young Pacific bluefin tuna is known to become slow in winter (Yukinawa and Yabuta, 1967; Bayliff, 1993). Because fish weight at a length was constant throughout the year for wild young Pacific bluefin tuna (Itoh, 2001), food consumption in win- ter appears not to be used for increasing weight at a length. We did not reach a conclusion on this question and further investigation of seasonal change in food items and of the physiology of tuna is needed. Acknowledgments We thank the staff of Marino Forum 21 and the Kagoshima Fisheries Experimental Station for their cooperation in the pe-held fish experiment. We also thank troll fishermen, staff in the Kamiagata Fisheries Cooperative Association, the Tsushima Fisheries Extension Service, and the Naga- saki Fisheries Experimental Station for their cooperation with the experiment on wild fish. We greatly acknowledge fishermen, consumers, and staff at the Inter-American Tropical Tuna Commission for their kindness in returning recovered archival tags bearing the information necessary for our study. We especially thank J. Gunn at CSIRO for giving us valuable information about implanting the archi- val tag in fish. We are also grateful to the staff of Northwest Marine Technology Inc. and Tanaka Sanjiro Co., Ltd., for providing us with tags. We would like to thank P. Ekstrom of Northwest Marine Technology Inc. for his critical review and help with the English text. We thank our staff in Japan NUS Co., Ltd., the Suidosya Co., Ltd., and the National Research Institute of Far Seas Fisheries, and also T. Kitagawa in the Ocean Research Institute of the Univer- sity of Tokyo, for their efforts in regard to implantation of the tags in fish. We gratefully acknowledge S. Kume of Japan NUS Co., Ltd., N. Baba of Fishery Research Agency, Z. Suzuki, and Y. Uozumi of National Research Institute of Far Seas Fisheries for their critical review. Literature cited Bayliff, W. H. 1993. Growth and age composition of northern bluefin tuna, Thunnus thynnus, caught in the eastern Pacific Ocean, as estimated from length-frequency data, with comments on trans-pacific migrations. Bull. lATTC 20:503-540. Block, B. A., H. Dewar, E. V. Freund, C. Farwell, and E. D. Prince. 1998a. A new satellite technology for tracking the move- ments of Atlantic bluefin tuna. Proc. Natl. Acad. Sci. USA 95:9384-9389. Block, B. A., H. Dewar, T. Williams, E. D. Prince, C. Farwell, and D. Fudge. 1998b. Archival tagging of Atlantic bluefin tima {Thunnus thynnus thynnus). Mar. Tech. Soc. J. 32: 37-46. Block, B. A., J. E. Keen, B. Castillo, H. Dewar, E. V. Freund, D. J. Marcinek, R. W. Brill, and C. Farwell. 1997. Environmental preferences of yellowfin tuna {Thun- nus albacares) at the northern extent of its range. Mar. Biol. 130:119-132. 544 Fishery Bulletin 101(3) Brill, R. W., D. B. Holts, R. K. C. Chang, S. Sullivan, H. Dewar, and F. G. Carey. 1993. Vertical and horizontal movements of striped marlin (Telrapturus audax) near the Hawaiian Islands, determined by ultrasonic telemetry, with simultaneous measurement of oceanic cuiTents. Mar. Biol. 117:567-574. Carey, F. G., J. W. Kanwisher, and E. D. Stevens. 1984. Bluefin tuna warm their viscera during digestion. J. Exp. Biol. 109:1-20. Carey, F. G., and K. D. Lawson. 1973. Temperature regulation in free-swimming bluefin tuna. Comp. Biochem. Physiol. 44A:375-392. Carey, F G., and R. J. Olson 1982. Sonic tracking experiments with tunas. ICCAT Col- lective Volume of Scientific Papers XVII. 2:458-466. Carey, F. G., and B. H. Robison 1981. Daily patterns in the activities of swordfish, JVT/p/i /as gladius, observed by acoustic telemetry. Fish. Bull. 79: 277-292. Carey, F. G., and J. V. Scharold 1990. Movements of blue sharks (Prionace glauca ) in depth and course. Mar. Biol. 106:329-342. Cayre, P. 1991. Behavior of yellowfin tuna (Thunnus albacares) and skipjack tuna (Katsuwonus pelamis) around fish aggregat- ing devices ( FADs ) in the Comoros Islands as determined by ultrasonic tagging. Aquat. Living. Resour. 4:1-12. Cayre, P. and F. Marsac. 1993. Modeling the yellowfin tuna (Thunnus albacares) ver- tical distribution using sonic tagging results and local envi- ronmental parameters. Aquat. Living Resours. 6:1-14. Dizon, A. E., R. W. Brill, and H. S. H. Yuen. 1978. Correlations between environment, physiology, and activ- ity and the effects on thermoregulation in skipjack tuna. In The pysiological ecology of tunas (G. D. Sharp and A. E. Dizon, eds. ), p. 233-259. Academic Press, New York, NY. Gunn, J., and B. Block. 2001. Advances in acoustic, archival, and satellite tagging of tunas. In Tuna: physiology, ecology, and evolution (B. A. Block and E. D. Stevens, eds.), p. 167-224. Academic Press, San Diego, CA. Gunn, J., J. Hartog, and K. Rough. 2001. The relationship between food intake and visceral warming in southern bluefin tuna (Thunnus maccoyii). Can we predict from archival tag data how much a tuna has eaten? hi Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 109-130. Kluwer Academic Publisher, Netherlands. Holland, K. N., R. W. Brill, and R. K. C. Chang. 1990a. Horizontal and vertical movements of Pacific blue marlin captured and released using sportfishing gear Fish. Bull. 88:397-402. 1990b. Horizontal and vertical movements of yellowfin and bigeye tuna associated with fish aggregating devices. Fish. Bull. 88:493-507. Itoh, T 2001. Estimation of total catch in weight and catch-at-age in number of bluefin tuna Thunnus orientalis in the whole Pacific Ocean. Bull. Nat. Res. Inst. Far Seas Fish. 38: 83- 111. |In Japanese.] Itoh, T, S. Tsuji, and A. Nitta. 2003. Migration patterns of young Pacific bluefin tuna (Thunnus orientalis) determined with archival tags. Fish. Bull. 101: 514-534. Kitagawa, T, H. Nakata, S. Kimura, T. Itoh. S. Tsuji, and A. Nitta. 2000. Effect of ambient temperature on the vertical distribu- tion and movement of Pacific bluefin tuna Thunnus thynnus orientalis. Mar Ecol. Prog. Ser. 206:251-260. Koido, T, and K. Mizuno. 1989. Fluctuation of catch for bluefin tuna (Thunnus thynnus) by trap nets in Sanriku coast with reference to hydrographic condition. Bull. Jpn. Soc. Fish. Oceanogr. 53: 138-152. [In Japanese.] Laurs, R. M., H. S. H. Yuen, and J. H. Johnson 1977. Small-scale movements of albacore, Thunnus ala- lunga, in relation to ocean features as indicated by ultra- sonic tracking and oceanographic sampling. Fish. Bull. 75: 347-355. Marcinek, D. J., S. B. Blackwell, H. Dewar, E. V. Freund, C. Fai-well, D. Dau, A. C. Seitz and B. A. Block. 2001. Depth and muscle temperature of Pacific bluefin tuna examined with acoustic and pop-up satellite archival tags. Mar Biol. 138:869-885. Ogawa, Y, and T. Ishida. 1989. Hydrogi"aphic conditions governing fluctuations in the catch ofThunnus thynnus by set-nets along the Sanriku coast. Bull. Tohoku Reg. Fish. Res. Lab. 51:23-39. [In Japanese.) Sund, P. N., M. Blackburn and F. Williams. 1981. Tunas and their environment in the Pacific Ocean: a review. Oceanogr. Mar Biol. 19:443-512. Uda, M. 1957. A consideration of the long years trend of the fisheries fluctuation in relation to sea condition. Bull. Jap. Soc. Sci. Fish. 23:368-372 Yabe, H., N. Anraku, and T Mori. 1953. Scombroid youngs found in the coastal seas of Abu- ratu, Kyusyu. in summer. Contribution of Nankai Reg. Fish. Res. Lab. 1:1-10. [In Japanese.] Yuen, H. S. H. 1970. Behavior of skipjack tuna, Katsuwonus pelaniis, as determined by tracking with ultrasonic devices. J. Fish. Res. Board Canada 27:2071-2079. Yukinawa, M., and Y. Yabuta. 1967. Age and growth of bluefin tuna, Thunnus thynnus (Linnaeus), in the north Pacific Ocean. Rep. Nankai Reg. Fish. Res. Lab. 25:1-18. ]In Japanese.] 545 Abstract — Demersal groundfish den- sities were estimated by conducting a visual strip-transect survey via manned submersible on the continental shelf off Cape Flattery, Washington. The purpose of this study was to evaluate the statis- tical sampling power of the submersible survey as a tool to discriminate density differences between trawlable and untrawlable habitats. A geophysical map of the study area was prepared with side-scan sonar imagery, multibeam bathymetry data, and known locations of historical NMFS trawl survey events. Submers- ible transects were completed at ran- domly selected dive sites located in each habitat type. Significant differences in density between habitats were observed for lingcod (Ophiodon elongatus), yel- loweye rockfish (Sebastes ruberrimus), and tiger rockfish (S. nigrocinctus) individually, and for "all rockfish" and "all flatfish" in the aggregate. Flatfish were more than ten times as abundant in the trawlable habitat samples than in the untrawlable samples, whereas rockfish as a group were over three times as abundant in the untrawlable habitat samples. Guidelines for sample sizes and implications for the estimation of the continental shelf trawl-survey habitat- bias are considered. We demonstrate an approach that can be used to establish sample size guidelines for future work by illustrating the interplay between statistical sampling power and 1) habitat specific-density differences, 2) variance of density differences, and 3) the proportion of untrawlable area in a habitat. Manuscript approved for publication 12 February 2003 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:545-565 (2003). Demersal groundfish densities in trawlable and untrawlable habitats off Washington: implications for the estimation of habitat bias in trawl surveys Thomas Jagielo Annette Hoffmann Jack Tagart Washington Department of Fish and Wildlife 600 Capitol Way N. Olympia, Washington 98501-1091 E-mail address (for T Jagielo) lagiethim'dfw wa gov Mark Zimmermann National Marine Fisheries Service 7600 Sandpoint Way NE Seattle, Washington 98115-0070 Despite their utility, trawl surveys C£ui- not obtain quantitative samples from rough, rocky habitats, and thus have a limited ability to sample all habitats representatively (Uzmann et al., 1977; Kulbicki and Wantiez, 1990; Krieger, 1993; Gregory et al., 1997). Since 1977, triennial bottom trawl surveys have been used to estimate the abundance of commercially and recreationally exploited groundfish species in the continental shelf waters off Washing- ton, Oregon, and California (Shaw et al., 2000). The data generated from these NMFS surveys are often a key component of groundfish stock assess- ments which are used to set levels of acceptable biological catch (ABC) for selected species (PFMC, 2001). Clearly, proper interpretation of these survey data with respect to fish habitat prefer- ences is an important part of developing unbiased stock assessments for fisher- ies management. In trawl survey methodology, popula- tion biomass is related to CPUE by the following equation (Dark and Wilkins, 1994): S, =-L CPUE.x- a. \ q where i B. area-depth stratum; estimated biomass in the ith area-depth stratum; A, = total area in the ith stratum; a, = total area swept during a standard trawl haul in stra- tum i; CPUE, = mean catch per unit of effort in the ith stratum; and q = the catchability coefficient of the sampling trawl. For this model to be an unbiased esti- mator of abundance, it is necessary to assume that the area sampled by the trawl is representative of the entire area-depth stratum of interest (i.e. a, is representative of A,). Validating this assumption becomes particularly important where untrawlable habitat comprises a significant proportion of the total area assessed, and where species composition and density vary between habitats. We shall refer to error in trawl survey estimates of abun- dance due to differences in groundfish density between habitat types as the trawl-survey habitat-bias. The trawl-survey habitat-bias may be substantial on the west coast conti- nental shelf because of the considerable spatial extent of untrawlable habitat in some management regions (Shaw et al., 2000). It is also widely recognized that demersal groundfish species com- position and density can vary consider- ably by bottom type (Richards, 1986; 546 Fishery Bulletin 101(3) Matthews and Richards, 1991; Stein et al.,1992; O'Connell and Carlile, 1993; Gregory et al., 1997; Krieger and Ito, 1999; Nasby, 2000; Yoklavich et al., 2000). Thus, there is considerable interest in evalu- ating alternative survey tools. 48°30' One alternative to trawl surveys that has gained increased attention in recent years is the method of direct observation of the seafloor, typically conducted with a remotely operated vehicle (ROV) or with an occupied submersible (Auster et al., 1989; Krieger, 48°00' 1993; Caimi et al., 1993; Adams et al., 1995; Gregory et al., 1997; Nasby, 2000). We evaluated the sampling power of the benthic video-strip transect method, us- ing videotapes of the sea floor collected in situ with an occupied submersible. Our goal was to judge the feasibility of using this approach to provide mean- ingful comparisons of demersal groundfish densities between trawlable and untrawlable habitats on spa- tial scales large enough to be useful for west coast fisheries management. We prepared a geophysical map of the bottom and conducted a submersible survey at a study site located on the continental shelf off Cape Flattery, Washington. Our objective was to provide guidelines on sample sizes (num- ber of submersible transects) that would be needed to characterize differences in density between the two habitat types, and specifically, sample sizes that would be needed to estimate the trawl survey habitat bias in subsequent studies designed to cover wider geographic areas. The study was structured to answer the following questions: 1) what species occupy trawlable and untrawlable habitats off Washington; 2) what magnitude of density differences can be expected between trawlable and untrawlable habitats; 3) what is the variability offish density within each habitat type; and 4) what sample sizes are required to estimate density differences between habitats, and the trawl survey habitat bias, in a statistically reliable manner. Our focus was on the benthic species and species groups that could be assessed reliably with our submersible survey method; primarily rockfish (Sebastes spp.), lingcod (Ophiodon elon- gatus), and flatfish (Pleuronectiformes). Materials and methods Study site Selection of the study site was aided by examination of historical NMFS trawl survey records and Washington Department of Fish and Wildlife (WDFW) trawl fishery logbook data. We chose a rectangular area west of the Point of Arches, Washington, which extends from the Juan de Fuca Canyon in the east (125°17'W) to Nitinat Canyon in the west ( 125°37'W) and ranges from 48°13' in the south to48°16' in the north (Fig. 1). We selected this area because 1) this portion of the Washington coast has been the site of a productive groundfish fishery since the 1940s (Alverson 1951), 2) this location has been surveyed tri-annually since 1977 as part of the NMFS west coast shelf survey, 3) the area has demersal groundfish species of interest, and 4 ) 126°00' 125°00' Stu dy jSit e ;l / Figure 1 Location of the study area (marked "study site" on map) on the continental shelf off Washington State. the area contains both trawlable and untrawlable habitats. The seafloor of this area was sculpted and shaped by ice movements during the late Pleistocene period (approxi- mately 18-20 thousand years ago) and is characterized by boulder fields resulting from glacial deposition that cover substantial portions of the area (GoldfingerM. Planning for the submersible survey required geodetically precise knowledge of the seafloor characteristics of the study area. This was facilitated by conducting geophysical surveys and by preparing a detailed map, which was instrumental to the submersible survey design. Geophysical surveys and map preparation Geophysical surveys of the study site were conducted by collecting side-scan sonar and multibeam bathymetry data simultaneously during a five-day effort on board the USN Agate Passage (YP-697) in May 1998. Slant-range- corrected side scan sonar data were collected by using a Waverly widescan 100-kHz system, with a swath width of 800 m. Eighteen parallel track lines were conducted with 100% overlap. The resulting imagery was assembled into a mosaic map of the bottom relief for a rectangular area measuring approximately 5.6 by 24.8 km (13,888 hect- ares). Bathymetric data, with resolution on the order of ±0.4 m were collected with a Reson Model 8101 multibeam bathymetry system. The multibeam bathymetry data were processed to produce a detailed map of the bottom topogra- phy with 1-m depth contour intervals. Map overlays were prepared that showed the locations of trawl survey events and trawl fishery tows. Detailed NMFS records were used to identify the location of various events associated with historical surveys of the area. The NMFS survey event types included good hauls, bad hauls, short hauls (tows ended early because of rough bottom), ' Goldfinger, C. 2001. Personal commun. Department of Geol- ogy, Oregon State University, Corvalis, OR 97331. Jagielo et al.: Demersal groundfish densities in trawlable and untrawlable habitats off Washington 547 •*< J • Jt^' V ■^ 4 ^ i ( i 1 V ) I ^t -^ s '^hsn Figure 2 Geophysical map of the study area with associated overlays. Area outlined in bold is the submersible survey area. The grid squares are 800 by 800 m sample units. (Top): Mosaic of side-scan imagery of the seafloor (Middle): Depth contours (1-m isobaths) obtained from multibeam bathymetry. (Bottom): Locations of historical NMFS shelf trawl survey events. Hatched lines = chain drags, stars = chain snags, unhatched lines = successful tows. skipped hauls, chain drags, and chain snags. Interviews with knowledgeable fishermen were also conducted to establish the locations of known trawling sites within the area. The resulting geophysical map, with overlays, provided a geographically accurate reference of the study area that allowed a priori classification of the bottom into trawlable and untrawlable habitat types (Fig. 2). The final map consisted of the following layers: 1) a mosaic of side- scan imagery of the bottom (Fig. 2, top); high-resolution depth contours (1-m isobaths) obtained from multibeam bathymetry (Fig. 2, middle); and 3) locations of historical NMFS trawl survey events (Fig. 2, bottom). Experimental design Our survey design process made use of the detailed map of the study area for 1) definition of the sampling unit, 2) classification of all sampling units as trawlable or untraw- lable habitats, and 3) specification of the in situ survey area. A sample of units to be surveyed by submersible was selected from each habitat type by using computer-gener- ated pseudo-random numbers. In defining the size of the sampling unit, we sought to strike a balance between a spatial scale that was small enough to have homogeneity but large enough to have meaning as a trawlable or untrawlable space. We chose square sample units of 800 by 800 m in size. This size was smaller than the standard NMFS tow length of about 3,000 m and was well within the order of resolution of the multibeam bathymetry and side-scan imagery used for discerning rock outcrops. A grid consisting of the 800 by 800 m sampling units was prepared and overlaid onto the map of the study area (Fig. 2). Classification of the sampling units into "trawlable" and "untrawlable" habitats was facilitated by examination of the geophysical map of habitat features, together with an evaluation of historical NMFS trawl survey records. The survey map layer helped us to interpret the appearance of trawlable and untrawlable habitat on the bathymetric and side-scan geophysical map layers. Trawlable bottom 548 Fishery Bulletin 101(3) 4 5 3 2 1 14 13 9 11 15 12 10 8 6 7 16 Figure 3 Schematic diagram of the study area divided into 800 m by 800 m sample units and classified a priori as untrawlable (shaded) and trawlable (unshaded) habitat types. The numbered sites represent the eight sample units selected at random from each habitat type, which were numbered sequentially for the cruise plan. was inferred from locations with good hauls and unevent- ful chain drags; untrawlable bottom was inferred from bad hauls, short hauls, skipped hauls, and chain snags. On the side-scan mosaic layer, untrawlable locations were typi- cally darker than surrounding areas, indicating boulder fields or hard, rocky bottom. Such areas often showed high bottom relief as evidenced by shadows on the mosaic, and bathymetric contours that indicated abrupt topographic features, such as sharp ridges or pinnacles. A sample unit was classified as untrawlable habitat when 1 ) NMFS sur- vey events within the unit indicated rough bottom, or 2) the mosaic or bathymetric layers of the unit resembled other units that were classified as untrawlable, or 3) a sample unit of unknown habitat type was completely surrounded by untrawlable habitat. A sample unit was classified as trawlable habitat when 1 ) NMFS survey events indicated successful trawl tows in the unit or 2) when the mosaic or bathymetric layers of the unit resembled other units that were classified as trawlable. Our trawlable and untraw- lable habitat assignments agreed well with information obtained from knowledgeable fishermen. Each sampling unit in the entire mapped area was examined visually in detail according to the above procedure and was classified accordingly as trawlable or untrawlable habitat. We selected the eastern portion of the mapped area for the submersible survey (Figs. 2 and 3t. Our focus was restricted to this section to minimize the difference in bot- tom depths between the trawlable and untrawlable areas as a factor, and for logistical convenience to complete the most submersible dives possible within our survey budget. Because the 800 m by 800 m sampling units were too large to be surveyed in their entirety, we sampled using the strip transect method at each location. Logistically, this was accomplished by conducting 2-3 ncmoverlapping passes across the sampling unit and by pooling these segments together to form a single transect for analysis. Submersible survey We used the submersible Delta to conduct the fish survey with the support vessel FV Auriga in July of 1998. The Delta is 4.7 m long, accommodates one observer and one pilot, and has a maximum operating depth of 365 m. An acoustic Trak- Point system was used with differential GPS and WinFrog navigational software (Thales GeoSolutions (Pacific), San Diego, CA) to track and log the position of the submersible from the support vessel. The Delta was equipped with halogen lights, external video cameras, an external Photosea 35-mm camera with strobe, and a Pisces Box data-logging system that recorded 1 ) the time of day, 2) depth of the submersible, 3) its distance from the bottom, and 4) sea temperature at 5-second intervals. Strip tran- sects were conducted 1-2 m off bottom at a cruising speed of approximately 2.5 km/h. All dives were made during daylight hours. To quantify fish density, each strip transect was docu- mented with a high 8-mm video camera mounted exter- nally on the bow of the Delta, and pointed forward. The camera was equipped with two parallel lasers, spaced 20 cm apart, which were used for estimating the area that was swept. The scientific observer onboard the Delta verbally annotated the videotape record with observations taken through the submersible viewing ports, to help identify fish and interpret the videotapes during subsequent analysis. The high 8-mm tapes were copied to S-VHS format to facilitate videotape analysis. The transect area that was swept (m-) was estimated as the product of average area swept per second (m-/min) and the total transect duration in minutes (see Appendix I for details). The average area that was swept per second (m'^/min) was determined from a set of 30-second samples randomly selected from the transect. On average, approximately 29'7r of each transect was subsarapled in this manner. Bottom habitat type was also visually characterized for the transect subsamples. Following the method of Stein et al. (1992) and using the classification criteria developed by Greene et al. ( 19991, we categorized bottom microhabitat type (mud, pebble, cobble, boulders, and rock ridge) as primary (at least 50% of the area viewed) or as secondary (>20% of the area viewed). The bottom-type measurements observed directly in the transect subsamples were expanded to estimate microhabi- tat coverage for each transect. Jagielo et al.: Demersal groundflsh densities in trawlable and untrawlable habitats off Washington 549 Fish were enumerated by identifying and counting only those fish observed in the lower portion of the video moni- tor screen (counting area), below the imaginary line con- necting the laser spots. Lighting and visibility was greatest in this zone, and we assumed that the probability of observ- ing and counting fish in this portion of the video image was 1009f (i.e. g=l). Afish was counted if any portion of the fish was visible in the counting area. The distance obsei-ved be- tween the two laser spots was used as a reference to classify fish into two size categories; large (>20 cm) and small (<20 cm). Fish were identified to the lowest taxonomic level pos- sible. We recognized that fish detection and identification were subject to observer error The variability describing that error was obtained by conducting a repeat counting of a sample of transects by the same observer. Additional validation checks were made between multiple observers. where Zj^ = the percentile of the unit normal which gives power; Zj^ = the percentile of the unit normal for the significance criterion; for a two-tailed test, a = a^2/2; d = the standardized effect size index for the two-tailed <-test calculated as L. ^ CD ^ 0) a H II (N JJ TJ 41 Oh s 01 -a 3 CO e H c S tn 3 II W C S a +j c ^ OJ CO ^ 'C '> ■ 2 CO i~ a) « c- "S " b aj Si 5- c 3 M = g CO S V, '5 o -S c •2 « bo PQ 01 o rn r^ 00 i> cx> CO CD o t^ O r- CD o -^ -^ 1-H lo ic -^ CD -^ in lo t~ O) CO [> in r^ 000(NOOC0005t> OOOOpOlNO'^O cbooocio-HO'H C^OOO'^i-HC-) i-H(N oooooooooooot-ooio pOptDOOOOOOOOOO(MO oooooocDooodocJooo OOOOOOOOOOOOlO-^t>lO ooooooooopoooopp OOOCJcdocDOOOOOOcDOO OOOOOOOOr-i ppOOOOOOC>5 OCJOOOOOO O'HOOO— lOO 0(N OOOOOOOOlOCOOOO-^CDt-H pppppOOOOCDOOOOOOIO OOC?OOOOOOOOOCDOCD»-< ooooooooooooooooo pooooooooooooooo OCDOOcicDOOOc o o O 00 o o o o o o o O -H 00 in O 'H o ,-i <3 CO ooooooooiO'^cooinincoco pppooooooo'^coiO'^a^mt>j CDCDOOOOOO'-ioC OOOC^OOCDOOOO oooooo-^oooo o o d c> CD d c lOlO'^COTflOCO'^COCSIlO-^CO'-ICO'* OOOOOOOO'-HOOOOOOO pooooooooooooooo ^ CD d T-i o o o o o oooooooooooooooo pOppppppppTOppOOp oinoooooo-^ooooooo pcopppooooocoooooo d'^ddSciddicidcidicicicidi IM O O O O ■* o o o o •^ d <6 d d CO O ■* CO p p ddcoc^Tfi-icD^co"^ O O O O O O "-I o o o o o o in d d d d d Oi-lOOOOOOOOO -HO o d o o d o d o d o d c? o d OOOOOOOO OO i-l 05 o O O £^ O CO o d d d CO in in in ^ ^ in o o § 8 00 CO o CD o o o o o o o o o 00 o o o o o o o o o in 00 in in o >* CD Tf in (M ^ o o o CO o o o o CO CO 00 HHHHHHHHDDDDDDDD h^ h TriotCiO'-i(>icom.-HOjcct^ 552 Fishery Bulletin 101(3) A summary of counts for large (>20 cm) and small (<20 cm) fish is shown in Table 4. Small flatfish and rockfish were very difficult to count, often becoming indistinguishable from the background when the videotape was paused, and their counts are most likely underestimated. Among the large fish, "total rockfish" as a group was the most abundant numerically followed by "total flatfish" as a group. Of the large rockfish identified to species (Table 5), rosethorn rockfish were the most abundant followed in order by yellowtail, greenstriped, yelloweye, tiger, and redstripe rockfish. Unidentified rockfish represented 30% of the total large rockfish enumerated. Of the large flatfish identified to species (Table 6), Dover sole were most abun- dant followed in order by arrowtooth flounder and Pacific halibut. Unidentified flatfish represented 78% of the total large flatfish counted. Other individual fish species and groups identified below the generic classification level were dominated by eelpout (Zoarcidae), raffish , skates and rays (Raja), and greenling {Hexagramrnos spp.) (Table 7). Species composition differed considerably between habi- tats. The number of individually identified species was 15 in the trawlable habitat, and 18 in the untrawlable habitat (Table 8). Flatfish dominated in the trawlable habitat, and rockfish in the untrawlable habitat. Yelloweye, redstripe, silvergray, and quillback rockfish, as well as greenling and wolf-eel were observed in the untrawlable habitat but not in the trawlable habitat. Spiny dogfish (Squalus acanthias), Pacific cod (Gadus macrocephalus), and salmon (Oncorhynchus spp) were observed in the trawlable habitat but not in the untrawlable habitat. Comparisons of fish densities and variances between habitat types were made only for fish >20 cm in length and in taxonomic units where reliable identification and enumeration could be assured (Table 9). Thus, density com- parisons were performed at the species level for distinctive species (i.e. lingcod, yelloweye rockfish, and tiger rockfish), but were made at the group level for "all rockfish" and "all flatfish" bwcause of the presence of fish that could not be identified to individual species within each of these groups. For all comparisions, tests of homogeneity of variance of fish density between habitats (//(,: s'^i=s^^) were rejected using Cochran's test (Winer, 1971 ) ( «=0.05, k=2, df=7), indi- cating heteroscedastisity (Table 9). Significant differences in densities between habitats were found for each of the species and group comparisons using the Mann-Whitney two-sample test on ranks (Winer, 1971) (a=0.05, 2) (Table 9). Densities were higher in the untrawlable habitat for the "all rockfish" group, tiger rockfish, yelloweye rockfish, and lingcod; densities were higher in the trawlable habitat for the "all flatfish" group. Statistical power analysis The validity of our approach for analyzing the statistical sampling power of the submersible survey depends upon, among other things, fidelity to the assumptions of the two-sample t-test of means. The <-test requires that 1) the two sample means are estimated from random samples drawn from normally distributed populations, and that 2) the variance of the two populations are equal. Because Table 3 Common and scientific names of fishes observed at 16 sub- mersible dive sites off Cape Flattery, Washington. Common name Scientific name Canary rockfish Sebastes pinniger Greenstriped rockfish Sebastes elongatus Quillback rockfish Sebastes maliger Redstripe rockfish Sebastes proriger Rosethorn rockfish Sebastes helvomaculatus Silvergray rockfish Sebastes brevispinis Tiger rockfish Sebastes mgrocinctus Yelloweye rockfish Sebastes ruberrimus Yellowtail rockfish Sebastes flavidus Greenling Hexagramrnos spp. Lingcod Ophiodon elongatus Pacific cod Gadus macrocephalus Arrowtooth flounder Atheresthes stomias Dover sole Microstomus pacificus Pacific halibut Hippoglossus stenolepis Spotted ratfish Hydrolagus colliei Spiny dogfish Squalus acanthias Longnose skate Raja rhina Big skate Raja binoculata Salmon Oncorhynchus spp. Wolf-eel Anarrhichthys ocellatus Eelpout Zoarcidae Poacher Agonidae Generic group classifications Unidentified rockfish Sebastes spp. Unidentified flatfish Pleuronectiformes Unidentified roundfish Osteichthyes our estimates of variance differed considerably between habitats (Table 9), we examined the properties of our data in more detail to confirm the reliability of using the t-test for our statistical power analysis. We conducted a bootstrap simulation experiment, in which we compared estimates of empirical power derived from our study (n=8) with the estimates of power obtained with Equation 1, under the assumption of asymptotic normality. The results of this comparison indicated that estimates of statistical power obtained from Equation 1 were generally conservative (indicated lower power) in relation to the empirical esti- mates of power for simulated known differences in density (Fig. 4). Given this result, we proceeded with our power analysis based on the <-test, under the assumption that, based on our observations, this approach will tend to err in the conservative direction; that is, it will tend to understate statistical power It is evident that, as it becomes necessary to detect small- er effect sizes, the required sample size increases accord- ingly. The relationship between sample size (;!=the number of sample units Isubmersible dive sites] in each habitat t3TJe) and the effect size-index (d) for density comparisons Jagielo et al.; Demersal groundfish densities in trawlable and untrawlable habitats off Washington 553 Table 4 Summary offish counts for large (>20 cm) and small (<20 cm) fish for major fish groups. Site type T = trawlable, U = untrawlable. Site Site type Number of large fish (> 20 cm) Number of small fish (< 20 cm Rockfish Lingcod Flatfish Other Total Rockfish Flatfish Other Total 4 T 0 1 77 8 86 0 95 48 143 5 T 8 0 54 17 79 0 94 15 109 6 T 2 0 76 12 90 0 68 63 131 10 T 7 3 29 5 44 0 26 107 133 11 T 0 1 35 10 46 0 8 101 109 12 T 0 0 46 5 51 0 6 63 69 13 T 39 1 119 19 178 0 77 37 114 15 T 0 0 31 2 33 0 70 64 134 1 U 115 1 6 16 138 43 0 10 53 2 U 128 14 12 28 182 348 3 52 403 3 U 9 2 28 10 49 41 9 58 108 7 U 43 9 13 22 87 0 21 46 67 8 U 32 3 4 9 48 40 2 12 54 9 U 206 5 6 14 231 339 0 11 350 14 u 30 3 8 11 52 38 7 27 72 16 u 111 5 30 9 155 28 4 17 49 Totals T 56 6 467 78 607 0 444 498 942 u 674 42 107 119 942 877 46 233 1156 All 730 48 574 197 1549 877 490 731 2098 Table 5 Summary offish counts by site for large rockfish (>20 cm). Site type: T = = trawlable, U = untrawlable. Site Number offish (>20 cm) Rose- Yellow- Silver- Green- Quill- Red- Yellow Site type thorn tail gray striped Canary back stripe Tiger eye Unidentified Total 4 T 0 0 0 0 0 0 0 0 0 0 0 5 T 0 0 0 8 0 0 0 0 0 0 8 6 T 0 0 0 2 0 0 0 0 0 0 2 10 T 0 0 0 0 2 0 0 0 0 5 7 11 T 0 0 0 0 0 0 0 0 0 0 0 12 T 0 0 0 0 0 0 0 0 0 0 0 13 T 2 1 0 14 0 0 0 1 0 21 39 15 T 0 0 0 0 0 0 0 0 0 0 0 1 U 31 1 0 9 0 1 0 0 8 65 115 2 U 88 3 0 0 0 0 0 7 12 18 128 3 U 8 0 0 0 0 0 0 0 1 0 9 7 U 16 14 0 1 0 0 0 2 3 7 43 8 U 25 2 0 1 2 0 0 1 0 1 32 9 u 121 1 1 3 0 0 16 6 5 53 206 14 u 15 10 0 1 0 0 0 1 0 3 30 16 u 34 17 3 0 0 0 0 2 7 48 111 Totals T 2 1 0 24 2 0 0 1 0 26 56 U 338 48 4 15 2 1 16 19 36 195 674 All 340 49 4 39 4 1 16 20 36 221 730 554 Fishery Bulletin 101(3) Table 6 Summary of fish counts by site for large flatfish (>20 cm) Site type. T = trawlable, U = = untrawlable. Site Site type Number of fish ( >20 cm ) Arrowtooth flounder Dover sole Pacific halibut Unidentified Total 4 T 3 6 2 66 77 5 T 3 8 1 42 54 6 T 15 2 6 53 76 10 T 0 3 3 23 29 11 T 1 3 3 28 35 12 T 5 2 5 34 46 13 T 10 13 7 89 119 15 T 0 2 1 28 31 1 U 0 4 0 2 6 2 u 0 2 0 10 12 3 u 0 4 2 22 28 7 u 0 0 5 8 13 8 u 0 0 0 4 4 9 u 0 1 1 4 6 14 u 0 1 0 7 8 16 u 1 0 0 29 30 Totals T 37 39 28 363 467 All 38 51 36 449 574 Table 7 Summary offish counts by site for other large (>20 cm) fish. Site type: T = trawlable, U = untrawlable. Number offish (>20 cm) Site Site type Greenling Pacific cod Ratfish Spiny dogfish Skates/Rays Eelpout Salmon Unidentified Total 4 T 0 0 0 0 0 8 0 0 8 5 T 0 0 6 1 0 10 0 0 17 6 T 0 0 0 0 1 11 0 0 12 10 T 0 0 0 0 4 1 0 0 5 11 T 0 0 0 0 1 8 1 0 10 12 T 0 2 0 0 1 2 0 0 5 13 T 0 1 1 6 5 5 0 1 19 15 T 0 0 0 0 0 1 0 1 2 1 U 2 0 1 0 0 12 0 1 16 2 U 1 0 1 0 0 26 0 0 28 3 u 1 0 1 0 1 6 0 1 10 7 u 3 0 0 0 4 15 0 0 22 8 u 2 0 2 0 0 5 0 0 9 9 u 2 0 6 0 0 5 0 1 14 14 u 0 0 2 0 0 9 0 0 11 16 u 1 0 5 0 0 3 0 0 9 Totals T 0 3 7 7 12 46 1 2 78 U 12 0 18 0 5 81 0 3 119 All 12 3 25 7 17 127 1 5 197 between trawlable and untrawlable habitats is shown in Figure 5. To achieve power of 80% (r<=0.05), the required number of dives ranges from n = 5 (.d=2.0) to n = 17 (d=1.0); similarly, to obtain 90'/r power would require 8 to 27 dives. Empirical estimates of c/ from our study ranged from 1.1 for tiger rockfish to 2.0 for flatfish. This result suggests Jagielo et al.: Demersal groundfish densities in trawlable and untrawlable habitats off Washington 555 Table 8 Composition offish densities in trawl ab e and untrawlable sites by species (>20 cm), ranked in descending order of observed abun- dance (avg. no./hectare). Italicized species were not found in the other habitat type. Tra wlable sites Untrawlable sites Species or group Avg. no./hectare Species or group Avg. no./hectare Eelpout 11.46 Rosethom rockfish 77.78 Dover sole 9.33 Eelpout 19.26 Arrowtooth flounder 9.25 Yellowtail rockfish 10.70 Pacific halibut 6.88 Lingcod 9.78 Greenstriped rockfish 5.65 Yelloweye rockfish 8.65 Skate 2.81 Tiger rockfish 4.40 Spiny dogfish 1.67 Spotted ratfish 3.90 Spotted ratfish 1.54 Greenstriped rockfish 3.76 Lingcod 1.39 Redstripe rockfish 3.39 Pacific cod 0.70 Dover sole 3.00 Rosethom rockfish 0.48 Greenling 2.67 Canary rockfish 0.41 Pacific halibut 1.77 Salmon 0.28 Skate 1.11 Yellowtail rockfish 0.24 Silvergray rockfish 0.79 Tiger rockfish 0.24 Wolf-eel Canary rockfish Quillback rockfish Arrowtooth flounder 0.49 0.36 0.27 0.19 Generic group All flatfish 114.29 All flatfish 23.90 All rockfish 13.14 All rockfish 155.63 All fish 146.65 All fish 211.70 that it is relatively more difficult (i.e. more dive sites are required) to detect density differences between habitats for tiger rockfish, as compared to flatfish. The associated power curves for these two values of cf are illustrated in Figure 6. Figure 6 suggests that, given our observations (for values of c? as low as 1.1), a sample size guideline of approximately 15 submersible dive sites in each habitat type would yield approximately an 80% chance of detecting a difference in mean density at least as large as the random noise esti- mated in the data for a similarly designed study. Our statistical power analysis also indicated that, when the relative proportions of untrawlable and trawlable habi- tat, as well as the variability in the trawl survey estimates of abundance, are taken into consideration, the problem of estimating the trawl survey habitat bias can require sub- stantially more samples than would be required simply to compare the density differences between two habitat types. Values of the trawl-survey habitat-bias effect size- index (c?^), calculated for a range of untrawlable habitat proportions with empirical trawl and submersible survey data, are given in Table 10 and are plotted for rockfish and flatfish in Figure 7. Using the calculated values of d^ from Table 10, we derived sample size guidelines for rockfish and fiatfish (at power=0.80, a=0.05). The resulting relationship between the sample size required to estimate the trawl survey habitat bias (the «=number of submersible dive sites in each habitat type) and the proportion of untraw- lable habitat in a management area (A^/A) is illustrated in Figure 8. If, for example, the area of untrawlable habitat represented 20% of a management unit. Figure 8 indicates that the sample size required to estimate the trawl survey habitat bias would be « = 31 for rockfish (d^=0.73), and n = 9 for flatfish (c/^=1.41). Sample sizes for lingcod were much higher (n>100), owing to the comparatively small detectible effect size required (d(,=0.13). Discussion Our study successfully obtained a first look at the variabil- ity in groundfish densities in trawlable and untrawlable habitats for a study area off Washington. We also developed a framework to use these types of observations to derive sample size guidelines for designing larger-scale studies to estimate the trawl survey habitat bias. The limited geo- graphic scope of our study precludes extrapolation of our specific results to the west coast at large. However, we dem- onstrated an approach that can be used to establish sample size guidelines for future work by illustrating the interplay between statistical sampling power and 1) habitat-specific density differences, 2 ) variance of density estimates, and 3) the proportion of untrawlable area in a habitat. In our study area, we observed striking differences in species composition and fish density between the traw- 556 Fishery Bulletin 101(3) Table 9 Summary of estimated fish densities (no./hectare) and summary statistics for selected fish U = untrawlable. groups (>20 cm). Site type: T = trawlable, Site Site type Estimated fish density ( number/10^ m^) Rockfish Flatfish Lingcod Yelloweye rockfish Tiger rockfish Mean SE Mean SE Mean SE Mean SE Mean SE 4 T 0.00 0.00 15.16 8.55 0.20 0.28 0.00 0.00 0.00 0.00 5 T 1.39 0.10 9.36 1.26 0.00 0.00 0.00 0.00 0.00 0.00 6 T 0.43 0.63 16.26 3.70 0.00 0.00 0.00 0.00 0.00 0.00 10 T 1.15 1.96 4.78 0.45 0.49 0.02 0.00 0.00 0.00 0.00 11 T 0.00 0.00 7.84 3.54 0.22 0.36 0.00 0.00 0.00 0.00 12 T 0.00 0.00 8.52 1.12 0.00 0.00 0.00 0.00 0.00 0.00 13 T 7.54 9.23 23.01 8.97 0.19 0.40 0.00 0.00 0.19 0.40 15 T 0.00 0.00 6.49 3.13 0.00 0.00 0.00 0.00 0.00 0.00 1 U 25.07 29.36 1.31 0.54 0.22 0.34 1.74 1.79 0.00 0.00 2 U 27.05 19.23 2.54 1.61 2.96 2.49 2.54 2.00 1.48 1.24 3 U 1.61 2.68 5.02 4.63 0.36 0.28 0.18 0.30 0.00 0.00 7 U 7.60 8.83 2.30 1.29 1.59 2.65 0.53 0.88 0.35 0.59 8 U 4.62 0.97 0,58 0.55 0.43 0.06 0.00 0.00 0.14 0.27 9 U 34.92 15.63 1.02 0.12 0.85 0.47 0.85 0.93 1.02 1.19 14 U 6.43 7.41 1.71 1.62 0.64 0.80 0.00 0.00 0.21 0.27 16 U 17.20 7.37 4.65 2.58 0.77 0.51 1.08 0.93 0.31 0.28 Summary statistics 1.31 11.43 0.14 0.00 0.02 s2, 6.64 37.92 0.03 0.00 0.00 "J,, 15.56 2.39 0.98 0.87 0.44 ^\ 151.58 2.69 0.82 0.81 0.28 Cochran s test for homogeneity of variance (Winer 1971); C^^„ = 0.83 C 0.96 0.93 0.96 1.00 0.98 Mann Whitney test for equality offish densities (Winer 1971); [/,,^„ = U 61 63 51 60 56 51 Statistics to calculate effect size index id) for submersible survey power analysis |m,-mj 14.25 9.04 0.84 0.87 0.42 Sp 8.894 4.51 0.65 0.64 0.38 d 1.6 2.0 1.3 1.4 1.1 labia and untrawlable habitats. Flatfish were more than ten times as abundant in the trawlable habitat samples, whereas rockfish as a group were over three times as abundant in the untrawlable habitat samples. Silvergray, quillback, redstripe, and yelloweye rockfish were observed in the untrawlable habitat but not in any of the trawlable habitat samples. We know of no visual-transect data comparable to that presented here for fish abundances off Washington. However, previous habitat specific studies in other areas have also reported differences in species composition and fish densities between low relief (trawlable) and highly rugose (untrawlable) habitats. Richards (1986) conducted a submersible study in the Strait of Georgia, British Co- lumbia (21-140 m), and observed that the distribution of greenstriped, quillback, and yelloweye rockfish varied by depth and bottom type. Greenstriped rockfish were most abundant in fine sediment habitats, such as mud and cobble terrain. Quillback rockfish were most abundant in complex habitats, and yelloweye rockfish had higher densities in wall and complex habitats than in fine sedi- ment habitats. In the coastal fjord of Saanich Inlet, British Columbia (21-150 m), Murie et al. ( 1994) also reported that quillback rockfish density was higher in areas of complex or wall habitat, compared to areas of sand-mud habitat. Additionally, tiger, copper (S. caurinus), yellowtail, and yelloweye rockfish were observed only over complex or wall habitats, and gi-eenstriped rockfish were seen mostly over sand-mud habitat. Using sunken gill nets to sample trawlable and untrawlable habitats off Vancouver Island, B.C. (198-311 m in depth), Matthews and Richards (1991) reported difTerences in species composition between traw- lable and untrawlable areas and higher species diversity in trawlable habitat. Major species on trawlable bottom Jagielo et al ; Demersal groundfish densities in trawlable and untrawlable habitats off Washington 557 were Pacific ocean perch (S. alutus), splitnose rockfish (S. diploproa), greenstriped rockfish, and bocaccio [S. paucispinis). Major species on untrawlable bottom were sharpchin (S. zacentrus) and redbanded rockfish (S. bab- cocki). In a submersible study conducted off Southeast- ern Alaska (188-290 m), Krieger (1993) compared the fish densities of 4 untrawlable sites with 16 trawlable or marginally trawlable sites, and reported that densities of large (>25 cm) rockfish (a category that included Pa- cific ocean perch, sharpchin rockfish, redstripe rockfish, and harlequin rockfish (S. variegatus) were highest at trawlable sites. In a study of shortraker (S. boreal is) and rougheye (S. aleutianus ) rockfish conducted on the upper continental slope off southeastern Alaska (262-365 m), Krieger and Ito (1999) reported that soft substrates of sand or mud usually had the greatest densities; hard substrates of bedrock, cobble, or pebble had the least densities; and habitats containing steep slopes and numerous boulders had greater densities of rockfish than habitats with gradual slopes and few boulders. O'Connell and Carlile (1993) conducted a submersible survey off southeastern Alaska in two depth strata; shallow {<108 m) and deep (a 108 m). Yelloweye rock- fish were observed in cobble, continuous rock, broken rock and boulder habitats but were most abundant in broken rock and boulder habitats of the deep stratum. Habitat-specific studies in Oregon and California have used finer scales of habitat classification to characterize fish-habitat associations than our comparatively coarse trawlable or untrawlable classification. In Oregon wa- ters. Stein et al. ( 1992) reported estimates offish density by habitat-type from a submersible study of six stations at Heceta Bank in waters ranging from 60 to 340 m in depth. Rockfishes, particularly pygmy (S. wilsoni), sharpchin, rosethorn, and yellowtail, dominated all substrates except mud, where Dover sole and black- belly eelpouts (Lycodes pacificus) were most abundant. In California waters, Yoklavich et al. (2000) conducted a submersible study at Soquel canyon (94-305 m) in Monterey Bay. Cluster analysis grouped fish densities into six habitat guilds; most distinct were 1 ) guild I (fish associated with uniform mud bottom of flat or low relief, dominated by stripetail rockfish (S. saxicola)) and guild VI (fish associated with rock-boulder habitat of low to high relief dominated by pygmy rockfish). To contrast our results in Washington with findings from Oregon and California, we summarized the fish density estimates reported by Stein et al. ( 1992 ) and Yokla- vich et al. (2000) into a format roughly comparable to our data. Differences in the objectives and methods of their studies precluded a rigorous quantitative comparison with our results, particularly because of differences in habitat classification and survey design (random sampling in our study, purposive sampling in the other two studies). How- ever, some interesting similarities are apparent if the most highly rugose habitats of these two studies are treated as a proxy for untrawlable habitat and if the low bottom relief habitats are treated as a proxy for trawlable habitat (Table 11). Seven species (italicized in Table 1 1 ) co-occurred in all three studies. For all three studies, greenstriped rockfish Roc)^ • - - ■Lingcod ^i^ S 125- S 100- ^^^^ f 75- S 50- ^^^-^^^ 25 • ^ 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Proportion untrawlable Figure 8 Sample size guidelines for estimating the trawl survey habitat bias (A!=the number of submersible dives in each habitat type) as a function of the proportion of untrawlable habitat (A^/A) in a management area. for power = BO'/f and o=0.05 for three categories offish. Fig. 8). However, because our analysis aggregated flatfish as a group, these results do not address the estimation of a bias correction factor for individual species, which is a requirement for any correction factor to be useful for stock assessment purposes. As for any survey method, the visual transect survey method has an array of advantages and disadvantages, which have been well chronicled elsewhere (Uzmann et al., 1977; Ralston et al., 1986; Butler et al., 1991; Adams et al.,1995; Starr et al., 1996; CaiUiet et al., 1999). Some of the disadvantages include 1) difficulties in fish identification, particularly for small fish or fish with subtle coloration, 2) the potential for attraction or repulsion of fish from the 560 Fishery Bulletin 101(3) Table 11 Comparison offish density estimates (average number offish/hectare) in trawlable (D,) and untrawlable (D„) habitats from sub- mersible studies in Washington, Oregon, and California. Densities for italicized species were reported in all three studies. Washington (present study) Oregon ' Species D, D„ D. D.. California- D, D.. Rockfish Bank rockfish Bocaccio Canary rockfish 0.41 Cowcod Darkblotched rockfish Greenblotched rockfish Greenspotted rockfish Greenspotted and greenblotched rockfish Greenstriped rockfish 5.65 Halfbanded rockfish Pygmy rockfish Quillback rockfish Redstripe rockfish Rosethorn rockfish 0.48 Sharpchin rockfish Shortspine thomyhead Stripetail rockfish Tiger rockfish 0.24 Widow rockfish Yelloweye rockfish Yetlowtail rockfish 0.24 Flatfish Arrowtooth flounder 9.25 Dover sole 9-33 Pacific halibut 6.88 Other Fish Eelpout 11-46 Greenling Lingcod '■■'?-9 Pacific cod 0.70 Pacific hagfish Pacific hake Poachers Spotted ratfish 1-54 Salmon 0.28 Skate 2.81 Spiny dogfish 167 Wolf-eel 0.36 3.76 0.27 3.39 77.78 4.40 3.90 1.11 0.49 165.00 510.00 479.50 119.50 8.65 10.70 33.50 0.19 3.00 249.50 1.77 19.26 2.67 9.78 33.50 93.00 0.00 105.00 6.33 586.00 120.00 0.00 148.00 4.33 152.67 86.33 52.00 1.33 36.33 162.33 237.67 1.67 16.33 39.50 218.67 46.00 220.00 85.67 892.50 126.33 734.33 574.50 40.33 175.33 96.50 138.50 41.33 5.33 304.67 63.67 0.33 33.67 13.50 0.67 78.67 95.50 2.67 28.00 7.50 15.00 9.00 58.00 43.67 3.00 91.67 25.67 4.00 14.67 14.00 138.00 22.67 ' Oregon data source: Table ,3 of Stein et al. (1992). Categories "mud" and mud-cobble" were averaged and used as a proxy for trawlable habitat; categories "flat rock" and "rock ridge" were averaged and used as a proxy for untrawlable habitat. -' California data source: Table 2 of Yoklavich et al. (2000) Categories "mud," "cobble-mud" and "mud-pebble" were averaged and used as a proxy for trawlable habitat; categories "rock-mud," "rock ridge," and "rock boulder" were averaged and used as a proxy for untrawlable habitat. submersible, 3) variation in countability due to habitat type; for example, due to reduced visibility when the submersible maneuvered ofTbottom to avoid large boulders, or the failure to detect fish hiding behind boulders, and 4) the limitation of the technique to quantifying the density of benthic spe- cies found in close proximity to the bottom. The advantages of the visual transect survey method include the ability to 1) sample in habitats that are inaccessible to other survey methods, 2) observe /;; situ fish behavior, and 3) observe the distribution of fish and fish-habitat associations on a fine Jagielo et al.: Demersal groundfish densities in trawlable and untrawlable habitats off Washington 561 scale. Although our study was subject to the limitations of the visual transect method, we assumed that the method could reliably estimate (with a catchability of 9=1.0) the true density of selected demersal bottomfish in both trawlable and untrawlable habitats for evaluation of the habitat bias present in the trawl-survey approach (which does not allow for sampling in untrawlable habitat). We do not feel that this assumption was severely violated, although we have no objective measure of the potential biases of the method, and thus we cannot estimate the consequences of assump- tion failure. We did recognize clearly that difficulties in fish identification limited the number of species that we could quantitatively sample with this technique. Technological improvements in underwater videography and image rec- ognition software are likely to enhance the capabilities of the visual transect survey technique in the future. In conclusion, it is clear that relatively large-scale sur- veys are needed to assess bottomfish densities in habitats that are not accessible to trawl survey gear. For some spe- cies, it may be possible to derive an area-specific trawl-sur- vey bias correction factor, but for many other species it is likely that there will be no substitute for direct estimation of densities in untrawlable habitat on a routine and synop- tic basis. In either case, stratified random sampling designs should be employed with sample sizes sufficient to ensure acceptable levels of statistical power At present, the in situ visual transect submersible survey method appears to be a useful tool for this purpose, and the utility of this method will likely improve further with technological advances. Acknowledgments We would like to thank Farron Wallace and Brian Culver for help during the submersible dive survey and with fish identification on the videotapes; Cindy Knudsen for video- tape area-swept data collection; Kevin Redman and Colin Stewart (Williamson and Associates) for geophysical data analysis and mapping, and Mike Famam, and Brian Bunge (USN) for geophysical data acquisition; the captains and crews of the USN Agate Passage and FW Auriga for excel- lent support vessel services; D. Slater, C. Ijames, and J. Lilly of Delta Oceanographies for safe and efficient use of the Delta submersible; Victoria O'Connell and Waldo Wakefield for advice on field logistics and data collection; and Marion Larkin (FV Larkin), for his insights regarding trawlable and untrawlable habitat obtained from many years of fish- ing experience in the study area. This study was supported by the NOAA National Undersea Research Program, West Coast and Polar Regions Undersea Research Center, Uni- versity of Alaska Fairbanks (grant no. UAF 98-0045), the Washington Department of Fish and Wildlife, and the National Marine Fisheries Service. Literature cited 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. Alverson, D. L. 1951. Deep water trawling survey off the coast of Washing- ton (August 27-October 19, 1951) Commercial Fisheries Review 13:11. U.S. Dep. Fish. Wild. Serv., Sep. 292. Auster, P. J., L. L. Stewart, and H. Sprunk. 1989. Scientific imaging with ROVs: tools and techniques. Mar. Technol. Soc. J. 23(3):16-20. Butler, J, L., W. W Wakefield, P B. Adams, B. H. Robison, and C. H. Baxter 1991. Application of line transect methods to surveying demersal communities with ROVs and manned sub- mersibles. Proceedings of the Oceans 91 Conference, Honolulu, Hawaii, 1-3 October 1991, p. 689-696. Marine Technology Soc, Columbia, MD. Cailliet, G. M., A. H. Andrews, W W. Wakefield, G. Moreno, and K. L. Rhodes. 1999. Fish faunal and habitat analyses using trawls, camera sleds and submersibles in benthic deep-sea habitats off cen- tral California. Oceanol. Acta 22(6):579-592. Caimi, F. M., J. H. Blatt, B. G. Grossman, D. Smith, J. Hooker, D. M. Kocak, and F. Gonzalez. 1993. Advanced underwater laser systems for ranging, size estimation, and profiling. Mar. Technol. Soc. J. 27(1):31^1. Cohen, J. 1988. Statistical power analysis for the behavioral sciences, 2"'' ed., 567 p. L. Erlbaum Associates, Hillsdale, NJ. Dark, T A., and M. E. Wilkins. 1994. Distribution, abundance, and biological characteris- tics of groundfish off the coast of Washington, Oregon, and California, 1977-1986. U.S. Dep. Commer Nat. Mar. Fish. Serv., NOAA Tech. Rep. NMFS 117, 73 p. Davis, D. L., and R. F. Tusting. 1991. Quantitative benthic photography using laser calibra- tions, 5 p. Undersea World, San Diego, CA. Dixon, W. F., and F. J. Massey. 1957. Introduction to statistical analysis, 2"'' ed., p. 244—255. McGraw-Hill, New York, NY. Greene, H. G., M. M. Yoklavich, R. M. Starr, V. M. O'Connell, W. W. Wakefield, D. E. Sullivan, J. E. McRea, Jr, and G. M. Cailliet. 1999. A classification scheme for deep seafloor habitats. Oceanol. Acta 22(6):663-678. Gregory, R. S., J. T. Anderson, and E.L. Dalley. 1997. Distribution of juvenile Atlantic cod Gadus mor- hua relative to available habitat in Placentia Bay, New- foundland. Northwest Atl. Fish. Organ. Sci. Counc. Stud. 29:3-12. Krieger, K. J. 1993. Distribution and abundance of rockfish determined from a submersible and by bottom trawling. Fish. Bull. 91:87-96. Krieger, K. J., and D. H. Ito. 1999. Distribution and abundance of shortraker rockfish, Sebastes borealis, and rougheye rockfish, S. aleutianus, determined from a manned submersible. Fish. Bull. 97: 264-272. Kulbicki, M., and L. Wantiez. 1990. Comparison between fish bycatch from shrimp trawl- net and visual censuses in St.Vincent Bay, New Caledonia. Fish. Bull, 88:667-675. Matthews, K. R., and L. J. Richards. 1991. Rockfish (Scorpaenidae) assemblages of trawlable and untrawlable habitats off Vancouver Island, British Columbia. N. Am. J. Fish. Manage. 11:312-318. 562 Fishery Bulletin 101 (3) Murie, D. J., D. C. Parkyn, B. G. Clapp, and G. G. Krause. 1994. Observations on the distribution and activities of rock- fish, Se6as;es spp., in Sannich Inlet, British Columbia, from the Pisces IV submersible. Fish. Bull. 92:313-323. Nasby, N. M. 2000. Integration of submersible transect data and high- resolution sonar imagery for a habitat-based groundfish assessment of Heceta Bank, Oregon. M.S. thesis, 50 p. Marine Resource Management Program, College of Oceanic and Atmospheric Science, Oregon State Univ., Corvallis, OR. O'Connell, V. M., and D. W. Cariile. 1993. Habitat-specific density of adult yelloweye rockfish Sebastes ruberrimus in the eastern Gulf of Alaska. Fish. Bull. 91:304-309. PFMC (Pacific Fishery Management Council). 2001. Status of the Pacific Coast groundfish fishery through 2001 and recommended biological catches for 2002: stock assessment and fishery evaluation, 26 p. [Document pre- pared for the Council and its advisory entities). Pacific Fishery Management Council, Portland, OR. Peterman, R. M. 1990. Statistical power analysis can improve fisheries re- search and management. Can. J. Fish. Aquat. Sci. 47:2-15. Ralston, S., R. M. Gooding, and G. M. Ludwig. 1986. An ecological survey and comparison of bottom fish resources assessments (submersible versus handline fish- ing) at Johnston Atoll. Fish. Bull. 84:141-155. Richards, L. J. 1986. Depth and habitat distributions of three species of rockfish (Sebastes) in British Columbia: observations from the submersible Pisces IV. Environ. Biol. Fish. 17(1): 13-21. Shaw, F. R., M. E. Wilkins, K. L. Weinberg, M. Zimmermann, and R. R. Lauth. 2000. The 1998 Pacific West Coast bottom trawl survey of groundfish resources: estimates of distribution, abundance, and length and age composition. NOAA Technical Memo- randum NMFS-AFSC-114, 138 p. Starr, R. M., D. S. Fox, M. A. Hixon, B. N. Tissot, G. E. Johnson, and W. H. Barss. 1996. Comparison of submersible-survey and hydroacous- tic-survey estimates offish density on a rocky bank. Fish. Bull. 94:113-123. Stein, D. L., B. N. Tissot, M. A. Hixon, and W. Barss. 1992. Fish-habitat associations on a deep reef at the edge of the Oregon continental shelf Fish. Bull. 90:540-551. Uzmann, J. R., R. A. Cooper, R. B. Theroux, and R. L. Wigley. 1977. Synoptic comparison of three sampling techniques for estimating abundance and distribution of selected mega- fauna: submersible vs. camera sled vs. otter trawl. Mar. Fish. Rev 39(12): 11-19. Winer, B. J. 1971. Statistical principles in experimental design, 2"'' ed., 907 p. McGraw Hill, New York, NY. Yoklavich, M. M., H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fish. Bull. 98:625-641. Appendix I: Procedure used for estimating the swept transect area At each sample unit (submersible (iive site), we estimated the total swept transect area, where the swept area (m^) = (average area swept per second [m-^/sec]) x (total elapsed time [seconds] ). The average area swept per second ( m'^/sec) was computed for a set of randomly selected thirty second portions of each transect. Conceptually, we determined the average area swept per second for the subsampled areas from a series of adjacent trapezoids (Fig. 1). For each trapezoid, we determined swept area (A,) by measuring the width that was swept (/, ) and distance that was swept (T,), where A =-(/,+/,.: )?;• 2 The process involved a frame-by-frame analysis of the video image, which required tracking an object from the center of the video monitor display to the bottom edge of the video display for a known time interval ( Fig. 2 ). The elapsed time for this interval was obtained from the video frame count, and was used to calculate area swept per second. Width-swept estimates (/,) were calculated from 1 ) the distance between the laser spots on the video monitor display (w,), 2) the width of the video monitor display (V), and 3) the known distance between the lasers (W) (20 cm), where Submersible path Area 3 Area 2 Area Figure 1 Schematic representation of adjacent trapezoids. /; = VW (1) Because the width that was swept varied as the submers- ible distance olT bottom varied, it was measured for each block. The following procedure was performed in sequence: Jagielo et aL: Demersal groundfish densities in trawlable and untrawlable habitats off Washiington 563 ^ideo monitor Object 1 \ Object 2 \ r9i eft Object 1 (9r 1 Screen 1: Distance between laser spots, m', , is Tieasured and object adjacent to lasers is identified. ' > Screen 2: Frame by frame advance until object idjacent to lasers in screen 1 rests at bottom of /ideo screen; then, distance between laser spot; s 1 kV,, is measured, and the process repeats. Figure 2 Illustration of the video-monitor display showing the frame-by-frame screen advance procedure used to determine the area swept per second. 1) «', was measured to the nearest milli- meter, 2 ) an object on the seafioor adjacent to the laser spots was identified, 3) the videotape was advanced until the object appeared at the bottom of the video moni- tor display, and 4) w^ was measured again (Fig. 2). The distance that was swept during this interval (T) is calculated trigonometri- cally by using the angle of the camera and constants estimated with the following pro- cedures of Davis and Tusting (1991). The process is illustrated in Figures 3 and 4. The variables of interest are T = the geodetic distance between the lo- cation of the laser spots on the sea- floor and the bottom edge of the cam- era's field of view (distance swept); the height of the video camera above the sea floor; o = the angle of the camera lens; 8 = the tilt angle of the camera; D = the distance between the focal point of the camera and the reflection of the laser spots on the seafioor; D] = the horizontal distance from the camera to a point on the sea floor at the center of the camera's field of view; Dg = the horizontal distance from the camera to a point on the sea floor at the bottom edge of the field of view; and D' = the distance from the camera lens to the refiection of the laser spots on the seafioor; H = Figure 3 Schematic representation of the submarine and camera orientation to the bottom (the line labeled D' represents the center line of the camera). Defini- tions for the variables are provided in the left column of text on this page. w = the distance measured between the laser spots as they appear on the video monitor display; W = the known distance (20 cm) between the lasers mounted in parallel on the camera housing. Note the following relationships: 564 Fishery Bulletin 101(3) Laser spots on seafloor Figure 4 Schematic representation of the relationship between the camera lens, image plane, and laser spots on the seafloor. Di = D' cos 0, and // = D' sin 0 H D'sine A=- tan 6 + tan e + (2) (3) In Equation 3, estimation of D^ requires the height of the camera above the seafloor (H); however, the need for a direct measurement of H can be eliminated by using camera parameters that provide an independent estimate ofDMFig. 4). Figure 4 shows the relationships between the camera lens, image plane, and laser spots, where d is a constant representing the distance from the focal point to the image plane, and c is a constant representing the distance from the camera lens to the image plane (note that c may be positive or negative). Note that both d and c are specific to the video display monitor employed, W, 0, and « are fixed, and w is observed. Therefore, W D' = D-d-c. ■dndD = d w D' = il\—-\\-( (4) (5) Underwater tests were conducted and the constants c and d were estimated for Delta s video camera and laser set-up by following the procedures of Davis and Tusting (1991). The distance traveled (T) for each area-swept trapezoid (from the center of the image to the lower edge of camera field of view), then, is T=a-D, = D' cos 180 sin 6 1 180 tani 1 e + 180l T (6) Appendix II: Derivation of the trawl-survey habitat-bias estimator, and the trawl-survey habitat-bias effect size-index (t/^,) To estimate the trawl survey habitat bias, we contrasted 1 ) the traditional abundance estimator, which does not dis- criminate between fish density differences in trawlable and untrawlable habitats (habitat-biased), with 2) an unbiased abundance estimator that explicitly allows for density dif- ferences between trawlable and untrawlable habitats. Let D, = the true density in the trawlable habitat; A, = the area of trawlable habitat; D = the true density in the trawlable habitat; A = the area of untrawlable habitat; A = the total area = A^ + A„; A^ = total abundance; and A = the difference in true densities = D^-D^. Then, for the unbiased estimator. Ar = D,A, + £>„Ar and for the biased estimator, N = D^ = D,A, + D,A^. The habitat bias, then, is the difference of the two estima- tors, or Bias = (D/i, + D/iJ - (1) The total error in the abundance estimator is a function of both the bias and the variance ¥(/),) of the fish density estimator MSE = Bias'^ -t- (A2)V(D,), (2) Jagielo et al : Demersal groundfish densities in trawlable and untrawlable habitats off Washington 565 where V(D,) describes the uncertainty in the abundance estimator. If the bias is much less than this uncertainty, then its impact will be minimal. Therefore, we arbitrarily set Bias- = (A- mo,), (3) and substituting A A^ for bias from Equation 1 into Equa- tion 3 gives A^2^2 = (A2)V(Dp. Solving for A gives A = — 5£)(D,). A. (4) (5) where SD{ D, ) = the standard deviation of the trawl survey density estimator in the trawlable habitat. Thus, the effect size threshold used for detecting differ- ences in mean density in the power analysis is a product of the arbitrary decision for the bias in the abundance estimator to be equal to its standard error. For the statistical power analysis, we expressed A (the difference in densities between habitats) as the standard- ized effect size index (c/^) for a two-sample f-test (Cohen, 1988); dividing (Eq. 5) lay an estimate of the population standard deviation, which yields — SD(D,)/s^ A, 566 Abstract — Understanding the onto- genetic relationship between juvenile Steller sea lions iEiimetopiaa jubatus) and their foraging habitat is key to understanding their relationship to available prey and ultimately their sur\nval. We summarize dive and move- ment data from 13 young-of-the-year (YOY) and 12 yearling Steller sea lions equipped with satellite dive recorders in the Gulf of Alaska and Aleutian Islands (n=18), and Washington (/i=7) from 1994 to 2000. A total of 1413 d of transmission (.v=56.5 d, range: 14.5- 104.1 d) were received. We recorded 222,073 dives, which had a mean depth of 18.4 m (range of means: 5.8-67.9 m: SD= 16.4 ). Alaska YOY dived for shorter periods and at shallower depths (mean depth=7.7 m, mean duration=0.8 min, mean maximum depth=25.7 m, and maximum depth=252 m) than Alaska yearlings (.?=16.6 m, 0=1.1 min, .v = 63.4 m, 288 m), whereas Washing- ton yearlings dived the longest and deepest (mean depth=39.4 m, mean duration=1.8 min, mean maximum depth=144.5 m, and maximum depth= 328 m ). Mean distance for 564 measured trips was 16.6 km; for sea lions slO months of age, trip distance (7.0 km) was significantly less than for those >10 months of age (24.6 km). Mean trip duration for 10 of the 25 sea lions was 12.1 h; for sea lions slO months of age, trip duration was 7.5 h and 18.1 h for those >10 months of age. We identified three movements types: long-range trips (>15 km and >20 h), short-range trips (<15 km and <20 h) during which the animals left and returned to the same site, and transits to other haul-out sites. Long-range trips started around 9 months of age and occurred most frequently around the assumed time of weaning, whereas short-range trips happened almost daily (0.9 trips/day, n=426 trips). Tran- sits began as early as 7 months of age, occurred more often after 9 months of age, and ranged between 6.5 and 454 km. The change in dive characteristics coincided with the assumed onset of weaning. These yearling sea lion move- ment patterns and dive characteristics suggest that immature Steller sea lions are as capable of making the same types of movements as adults. Manuscript approved for publication 29 October 2002 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull 101:566-582 (2003). Diving behavior of immature Steller sea lions iEumetopias jubatus) Thomas R. Loughlin Jeremy T. Sterling National Marine Mammal Laboratory Alaska Fishenes Science Center, NMFS 7600 Sand Point Way, NE Seattle, Washington 98115 E-mail address (for T R Loughlin) torn loughlin®noaa gov Richard L. Merrick, Northeast Fishenes Science Center, NMFS 166 Water Street Woods Hole, Massachusetts 02543 John L. Sease Anne E. York National Marine Mammal Laboratory Alaska Fishenes Science Center, NMFS 7600 Sand Point Way, NE Seattle, Washington 98115 Steller sea lions range throughout the North Pacific Ocean rim and are declin- ing in numbers in most of Alaska and Russia (Loughlin et al., 1992; Loughlin and York 2000). Studies of mitochon- drial DNA suggest that at least two stocks exist: an eastern stock (Califor- nia through southeast Alaska) and a western stock (Prince William Sound and areas west) (Bickham et al., 1996; Loughlin, 1997). For the western U.S. stock (west of 144°W), counts of adults and juveniles have fallen from about 110,000 individuals in the late 1970s to about 25,000 individuals in 2000— a decline of almost 80%. Although the numbers of sea lions that died were greater from the late 1970s to the early 1990s than at present, the rate of decline has remained high. As a result of this decline the U.S. government desig- nated the western stock as "endangered" in 1997 under the U.S. Endangered Spe- cies Act; the eastern stock is designated as "threatened." Reasons for the decline in numbers are unknown but may be hnkcd to reduced availability of prey caused indirectly by environmental changes or commercial fishing activi- ties, or both (Loughlin and Merrick, 1989; Merrick, 1995). Severe environ- mental perturbations and commercial fishing, both resulting in changes in the abundance or availability of prey, have been implicated in the alteration of pin- niped foraging behavior and declines in pinniped abundance (e.g. Trillmich and Ono, 1991; Melin, 2002). One method for studying the effect of reduced prey avail- ability on pinnipeds is to measure diving behavior and foraging ecology by using either a time-depth-recorder ( Kooyman et al., 1983; Gentry and Kooyman, 1986) from which dive data are retrieved after the animal returns from a feeding trip (e.g. Goebel et al., 1991; Boyd et al., 1994; Werner and Campagna, 1995), or by using a satellite-linked time-depth recorder (SLTDR; the newer version is called a "satellite dive recorder" SDR), which transmits dive and transmitter- status information to orbiting satel- lites and thus eliminates the need to recapture the animal (e.g. Merrick et al., 1994). Few data are available concerning the foraging ecology of Steller sea lions. Merrick et al. (1994) and Merrick and Loughlin (1997) presented information on the dive characteristics and foraging Loughlin et al.: Diving behavior of immature Eumetopias jubatus 567 Table 1 Satelhte transmitter number (PTT number), deployment location, age, sex, and morphometric measures of 25 Steller sea lions studied for diving behavior in Alaska and Washington. 1994-2000. The ten ST- 10 and ST- 16 SDRs we deployed that transmitted time-line messages are shown with **. 1 = Washington State area; 2 = Kodiak area; 3 = Shumagin Islands; 4 = Unimak Pass area; 5 = Sequam area, n/d = no data obtained. PTT number is the satellite transmitter identification number. Est. = estimated. PTT number Location code (regional) Age (months) Sex Deployment date Length of transmission (d) Mass (kg) Girth (cm) Length (cm) 14073 12 M 6/8/95 51.67 86.26 n/d 151 14084 11 F 5/5/99 50.22 77.18 102 159 14085 19 M 1/5/00 14.50 154.36 133 193 14087 16 F 10/3/97 68.99 122.45 113 171 14089 16 F 10/3/97 55.58 111.11 122 173 21103** 22 M 3/30/00 63.80 139.23 128 192 21106** 22 M 3/30/00 83.28 143.31 119 192 14071 2 6 M 12/2/94 19.61 92.00 n/d n/d 14074 2 6 M 12/6/94 53.13 79.80 n/d n/d 14077 2 18 M 12/9/94 39.63 n/d n/d n/d 14078 2 18 F 12/7/94 57.99 n/d n/d n/d 14079 2 7 F 1/14/96 27.11 94.90 115 155 14080 2 7 M 1/16/96 79.04 106.20 115 156 14170** 2 21 M 3/12/00 93.99 Est. 95-105 n/d 150 21094** 2 9 M 3/12/00 66.40 62.20 90 145 14076 3 21 F 3/1/96 52.21 103.70 111 n/d 14081 3 9 M 3/2/96 45.02 n/d 144 n/d 14072 4 22 F 4/13/95 56.35 116.10 n/d n/d 14075 4 8 F 2/25/96 31.76 104.00 123 140 14164** 4 9 M 3/8/00 97.61 79.60 n/d n/d 14167** 4 9 F 3/9/00 29.67 100.20 n/d 155 14111** 5 9 F 2/29/00 61.71 87.00 108.5 151 14114** 5 9 F 2/29/00 52.53 85.80 108 157 14116** 5 9 F 2/29/00 56.61 76.20 102.5 148 14163** 5 9 M 2/29/00 104.14 109.00 113 156 behavior of a small sample of Steller sea lions in Alaska; Loughlin et al. (19981 provided similar information for Steller sea lions off the Kuril Islands, Russia. Merrick et al. (1990) and Brandon (2000) presented information on female pup-attendance behavior of sea lions with VHF radio-transmitters off the Kuril Islands and Alaska, re- spectively. These studies showed that during the breed- ing season, adult female Steller sea lions generally spent about half their time at sea on relatively brief (18-20 h) foraging trips. Dives tended to be shallow (,v=21 m), brief (.v=1.4 min), and frequent (about 13/h). Observations dur- ing winter showed that females with suckling yearlings (17-22 months of age) had feeding trips of about 2.3 days, whereas those with young-of-the-year (5-10 months of age) had trips lasting 0.9 of a day; time on shore for lactating females of both groups averaged 14.2 hours (Porter, 1997). Baba et al. (2000) were able to follow a yearling Steller sea lion for 5 months using two location-only satellite trans- mitters; one was attached to the top of the head and the other on the back. This animal traveled from Hokkaido to Sakhalin Island and throughout the southern Okhotsk Sea. No dive data were obtained. Our objective is to present a description of the diving behavior of juvenile Steller sea lions for the western stock of Steller sea lions in Alaska and the eastern stock in Wash- ington state. We deployed SDRs on juvenile Steller sea li- ons over a broader geographical range in Alaska and over a wider range of dates, providing a more comprehensive picture of the diving behavior of young Steller sea lions. Additionally, SDRs are now smaller and of higher quality, so that more detailed information on diving behavior is available. We then provide in the "Discussion" section a comparison of the accounts in the present study to those we published earlier on adult female diving behavior (e.g. Merrick and Loughlin, 1997). Materials and methods We captured 25 free-ranging Steller sea lions of both sexes from approximately 6-22 months of age at rookeries and haul-out sites in the Aleutian Islands and Gulf of Alaska (Table 1, Fig. 1 ) throughout the year from 1994 to 2000, and during 1995-2000 at Shilshole Marina in Puget Sound, near Seattle, Washington. Animal age was estimated by using 568 Fishery Bulletin 101(3) 50°0'N- -50'0'N XL I I. ll 12 3 4 5 6 7 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Deployment Age in Months I r- 16°0'W 13O0'W Figure 1 Locations where satellite dive recorders (SDRs) were deployed on 25 Steller sea lions in Alaska and Washington between 1994-2000. "Deployment age" is the age of the sea lions when satellite transmitters were attached. mid-June as the presumed birth date (Pitcher and Calkins, 1981) and pubUshed accounts of mass, standard length, and girth at age (Calkins et al., 1998). Some juveniles before 1996 were chemically immobilized with Telazol'" injected intramuscularly by a dart fired from a pneumatic gun (Loughlin and Spraker, 1989). Those animals were not weighed; therefore exact dosage levels were not determined. However, dosages were most likely between 1.5 and 2.5 mg/ kg. Once a sea lion was immobilized, intramuscular injec- tion of 3-10 cc of Dopram was administered to stimulate respiration and facilitate recovery. After 1996, young sea lions were captured on land with a hoop net and physically restrained. During all years a SLTDR or SDR was glued to the pelage on the animal's back with fast-setting cpoxy resin (Loughlin ot al., 1987), and two plastic cattle ear tags with the same identification numbers were attached, one to each front flipper The instruments were not recovered and were expected to be shed during or before molt. Instrument description and programming We used 0.5-watt ST-6 SLTDRs (packaged by Wildlife Computers, Redmond, WA), which provide dive depth, dive duration, and transmitter status. Further develop- ment by Wildlife Computers resulted in 0.25-watt ST- 10 and ST- 16 SDRs which could provide five messages: 1 ) dive depth, 2 ) dive duration, 3 ) transmitter status, 4) proportion of time at depth, and 5) a time line. Messages are sent at prescribed intei-vals; transmission interval at sea is every 43 sec and on land it is every 1 min 28 sec. The number of transmissions (and thus messages received) while the sea lion is at sea depends on the length of exposure of the instrument's salt-water switch at the surface. Location data are not sent by the transmitter but are calculated by Service-Argos, Inc. from the received message. Additional information on these instruments and their capabilities can be found in Merrick et al. (1994). The satellite track- ing system (Argos) is described in detail in Fancy et al. (1988) and Stewart et al. (1989). Additional information can be obtained from the manufacturer at their web site (www.wildlifecomputers.com). The ST-6 SLTDR stored, summarized, and transmitted dive data as histograms. Individual dives and surface in- tervals were not provided; therefore sampling frequency for measuring dive behavior was not a consideration (e.g. Boyd, 1993). Software programming of the SLTDR subdivided each day into four 6-h periods (2100-0300 h, 0300-0900 h, 0900-1500 h, and 1500-2100 h local time). Frequency histo- Loughlin et al.: Diving behavior of immature Eumetopias /ubatus 569 grams were summarized separately for dive depth and dive duration for each of the four time periods. The SLTDRs re- corded dive depth information in six separate "bins": 4—10 m, 10-20 m, 20-50 m, 50-100 m, 100-250 m, and >250 m. We used 4 m as the minimum depth for a dive based on ear- her studies in Alaska (Merrick et al., 1994). Dive-duration bins were 0-60 sec, 60-120 sec, 120-180 sec, 180-240 sec, 240-360 sec, and >360 sec. The ST-10 and ST-16 units used the same 6-h periods as the ST-6. However, the ST-10 and ST-16 SDKs subdivided dive depth information into 14 bins: 4 m; 4-6 m, 6-10 m, 10-20 m, 20-34 m, 34-50 m, 50-74 m, 74-100 m, 100- 124 m, 124-150 m, 150-174 m, 174-200 m, 200-250 m, and >250 m. Dive duration also contained 14 bins at one- minute intervals (e.g. 1-2 min, 2-3 min, 3-4 min, etc.). The 14 time-at-depth bins coincided with dive-depth bins (e.g. 0, 4, 4-6, 6-10, etc. and the last was >200). However, the first bin was set to zero to determine if an animal was on land based on the proportion of dry readings of the salt-water switch during a 6-hour period. Time-at-depth was calcu- lated as the proportion of time that dives occurred within a particular depth bin of a 6-h period while the sea lion was at sea (e.g. if an animal was at sea for 3 hours during a 6-h period and spent half its dive time in bin 50-74, the value in bin 50-74 would be 25'7f ). We deployed ten ST-10 and ST-16 SDKs (Table 1) which transmitted time-line messages in bins of 20-min periods (there are 72 periods of 20-min each in a 24-h day). These messages provide information on whether the instrument was wet or dry >10 min of a 20-min period for each of the 72 periods. Time-line messages thus allow calculation of time spent at sea and on land. Maximum dive depth in a 24-h period, from midnight GMT to midnight GMT. was provided in the status mes- sage. This is a separate message that provides information on transmitter status, including a pressure offset, battery status, number of transmissions to date, at-surface data, date, time, ID of message, and a saltwater conductivity reading. All 25 transmitters that we deployed transmitted a status message. The ST-6 SLTDRs were on 24 h/day and transmitted a maximum of 400 transmissions/day. To save battery power the instrument had a 6-h haul-out period; that is, it would turn off only if the transmitter was "dry" for 6 hours, indi- cating that the animal was on land. The ST-10 and ST-16 SDRs had 3-h haul-out periods; the ST-10 had a maximum of 250 transmissions/day, and the ST-16 had a maximum of 325/day Both the ST-10 and ST-16 had duty cycles of 4 h on and 2 h off during a 24-h period to distribute transmissions during different times of the day and to ensure recording of information in all bins. All duty cycles started at midnight, with an offset of + 13 h from GMT for Alaska. Location data Locations were estimated by the Service-Argos, Inc. clas- sification scheme, where location class (LC ) 3 is accurate to <150 m, LC 2 is accurate 150 m-s350 m, LC 1 is accurate 350 m-1000 m, and LC 0 is accurate >1000 m. LCs A and B have no assigned accuracy range (Service-Argos, 1984; Keating, 1994). However, after our analysis, Vincent et al. (2002) used an algorithm published by McConnell et al. (1992) to filter satellite locations and found that both filtered and unfiltered LC A locations were of a similar accuracy to LC 1 locations for four gray seals iHalichoerus grypus). Because of the large variance in our samples asso- ciated with LC A locations, we excluded them (and the LC Bs) from our analyses. We sorted location data by date and time line to determine the locations for each trip. Data analysis Data analysis followed that of Merrick et al. (1994) and Merrick and Loughlin (1997). Analysis of the number of dives was prepared by summing counts of dives from the histograms. Median depths and durations of dives were calculated by using the range midpoint of a bin (e.g. 7 m for a 4—10 m bin) as the depth for all dives in the bin. We rec- ognize that this approach invokes a possible error for dive profiles in large increment bins (e.g. 50-100 m) where the mean dive depth is the same, 75 m, regardless of whether the animal made most of its dives between 51 and 60 m or if it made most of its dives between 90 and 100 m. This error is inherent in the data collection process and could not be eliminated with the instruments used in our study. We also recognize that more locations will be recorded when the animals are at the surface for long periods or when transit- ing to different locations. However, because of the repetitive transmission of the histogram data and the usual short duration of short-range trips, there should be no inherent behavior-based bias in the dive data reported. Differences in dive depth and duration between locations were tested by using the Pearson chi-square tests or analysis of vari- ance (ANOVA) (F-statistic), and P-value differences less than 0.05 were considered significant. Analysis of trip distance and duration were analyzed by using a repeated measures ANOVA, and because the distances were skewed, they were log-trEinsformed to examine differences among groups. Trips were defined and measured for distance by us- ing an integrated process of the SDR data. For animals deployed with ST6 SDRs, which did not contain time line data sets (/! = 15), trip distances were extracted by using a combination of the dive histogram, duration histogram, and land or sea data sets to estimate arrival and departure times as well as locations calculated at sea or on land. Once arrival and departure times were estimated, the location data were examined to confirm that all locations calculated during that trip were wet locations. We then had all locations for an individual trip and from those locations we filtered out all A and B locations and im- posed a swim speed filter (3 m/s). Finally, we reported the maximum straight-line distance from the departure site. For animals with STlO/16 SDRs, we were able to extract arrival and departure times from the time-line messages. However, if a day of time-line data was not received, we referenced the time-at-depth data, depth, and duration histograms to reconstruct the missing day of data. Once arrival and departure times were calculated we then fol- lowed the protocol stated above. 570 Fishery Bulletin 101(3) ■ Alaska YOY(n= 13) D Alaska yearlings (n=5) ■ Washington yearlings (n=7) ^ £ ji n 4-10 10-20 20-50 50-100 100-150 >150 Depth bins in meters Figure 2 Percentage of dives occurring in each dive-depth bin for Alaska young- of-the-year (YOY), Alaska yearling, and Washington yearling Steller sea lions. Twenty-five animals are represented from 1994 to 2000 and a total of 222,073 dives. Results We report on SDR data obtained from 25 (13 male, 12 female) young-of-the-year and juvenile (estimated ages of <2 yr) Steller sea lions from Washington state. Gulf of Alaska, and Aleutian Islands, Alaska (Table 1). Most (22 of 25) were caught during October-March 1995-2000 and the remainder during May-July (Table 1). Mean number of days of transmission received from the SDRs was 56.8 d (range 14.5-104.1 d). Dive characteristics We recorded over 222,073 dives for young-of-the-year and juvenile Steller sea lions which had a mean dive depth of 18.4 m (range of means: 6.1-67.0 m; SD=16.23). Alaska young-of-the-year dived to shallower depths and for shorter periods (mean depth=7.7 m, SD=1.7; mean duration=0.8 min, SD=0.1; mean maximum depth=25.7 m, SD=16.9; and maximum depth=252 m) than did Alaska yearlings (mean depth=16.6 m, SD=10.9; mean duration=l.l min, SD=0.4; mean maximum depth=63.4 m, SD=37.7; and maximum dcpth=288 m), whereas Washington yearlings dived the deepest and the longest (mean depth=39.4 m, SD=14.9; mean duration=1.8 min, SD=0.6; mean maxi- mum depth=144.5 m, SD=32.6; and maximum depth=328 m ). Alaska animals dived to much shallower depths ( mean depth= 10.3 ml than animals from Shilshole,WA. There was no significant difference in the mean dive depths among locations in Alaska (P=0.8). Alaska animals, in compari- son to the Washington animals, had a significantly greater proportion of dives in the 4-10 m depth bin ( 707f , P<0.001) than in the deeper depth bins. We compared the proportion of dives in the shallowest bin (depth bin 4— 10 m ) for animals captured in Washington state versus Alaska using a generalized linear model with a binomial link function (McCuUagh and Nelder, 1989). The proportion of shallow dives was significantly greater (P<0.00 1 ) among the Alaskan animals ( 8 1 .47f ) than among the Washington state animals (43.8%). Among the Wash- ington state animals, the proportions of dives in the 1020 m depth bin (20.4%) and the 20-50 m depth bin (19.4%) were similar; proportions of dives in the deeper depth bins were progressively fewer (Fig. 2). Maximum and mean- maximum dive depth were also greater for young sea lions from Washington that dived to 141.5 m (SE=11.4) mean- maximum depth versus 33.8 m (SE=7.2) for Alaska sea lions (F=63.4, 23 and 24 df; P<0.001) (Table 2). We plotted the maximum depth for each 24-h period by the number of days in which the Argos satellite received a status message (which contains maximum depth for 24 hours) and found that with one exception ( PTT 14078 ), Washington yearlings consistently dived deeper than their Alaska counterparts (Fig. 3, A and B). Two of three Alaska young-of-the-year were shallow divers and the third dived to 250 m once and beyond 100 m on numerous occasions late in the track- ing period (Fig. 3C). The maximum depth for all sea lions that we studied was 328 m for a juvenile sea lion that was equipped with a SDR at Shilshole, WA (PTT 21106); the deepest dive for a yearling Alaska sea lion was 288 m (PTT 14078) (Fig. 3). Mean dive duration was 1.1 min for all young sea lions (n =226,497 dives). Dive duration was significantly longer for Shilshole sea lions (x=1.75 min; range: 0.95-3.10) com- pared to Alaska sea lions (v=0.85 min; range: 0.61-1.86; F=24.5, 23 and 24 df; P<0.001). Few dives were greater Loughlin et al.: Diving behavior of immature Eumetopias lubatus 571 Table 2 Summary of d ive parameters from satellite dive recorders (SDRs) deployed on Stellar sea lions in Washington and Alaska, 1994- | •2000. "PTT" is the satellite transmitter identification number. Mean max. Mean max. Max. Mean Mean Mean Mean dive depth dive depth depth depth depth duration duration PTT (m) in) (m) (m) M (min) (n) Washington 14073 77.68 38 168 31.99 10,746 0.96 11,241 14084 154.12 34 288 47.29 5183 1.69 5047 14085 187.67 12 280 67.94 1991 3.10 2025 14087 164.37 53 256 33.21 14,287 1.69 14,572 14089 144.00 46 200 44.76 9682 1.82 9659 21103 124.22 18 256 23.59 6920 1.40 5431 21106 159.09 44 328 26.92 11,839 1.61 11,647 Alaska 14071 10.44 18 12 7.20 1541 0.61 1732 14074 11.07 30 20 7.24 3044 0.72 2954 14077 28.57 21 144 9.67 4745 0.81 5146 14078 125.74 23 288 35.00 4186 1.82 3657 14079 12.80 20 44 7.13 7546 0.71 8482 14080 15.80 59 24 7.60 17,236 0.96 18,056 14170 41.87 47 180 10.23 17,741 0.91 19,447 21094 48.90 31 152 11.67 9745 0.79 9388 14076 51.69 26 144 9.29 11,593 0.86 12,739 14081 11.76 34 20 7.25 8717 0.70 9013 14072 68.98 49 100 18.69 12,597 1.06 11,931 14075 8.00 15 12 7.01 5424 0.67 4869 14164 26.09 44 60 9.46 16,352 0.96 16,426 14167 20.73 11 60 7.01 2985 0.81 3022 14111 17.76 25 40 6.44 10,919 0.72 10,903 14114 13.75 16 16 5.84 5846 0.68 6256 14116 24.24 17 40 6.70 7359 0.73 7739 14163 65.55 49 252 10.94 13,849 0.82 15,204 Mean 62.42 135.36 18.42 1.10 SE 11.50 3.23 3.23 0.11 than 6 min (Fig. 4). There was a significant positive Un- ear relationship between dive duration and dive depth (r2=0.89, F=7.06, 1 and 23 df, P<0.001), and a significant positive relationship between sea lion mass at the time of capture and mean dive duration (r^=0.46, F=3.86, 1 and 20 df,P<0,001) but not girth (r2=0.10,F=1.62, land 14 df, P=0.22). The relationship between dive duration and dive depth for males was not different from that for females (F=1.16, 2 and 21 df, P=0.33). The positive relationship between dive duration and mass was likely driven by the greater mass of the male sea lions because the relation- ship was not statistically significant when the analysis was restricted to females. Dive depth and duration showed an interesting ontoge- netic trend. Alaska animals 7-10 months old typically had a mean dive duration of <1 min and a mean dive depth of about 10 m; by 11-12 months of age both increased, almost doubling in most cases (Fig. 5). Although sample size was small, this ontogeny of diving to deeper depths for longer periods at about 11-12 months of age was evident in the percentage of time at depth (Fig. 6). There was a higher pro- portion of time spent in the deeper depth bins during May and June (at age 11 and 12 months, respectively) than when younger, and the proportion of time hauled out was reduced for the older animals. Interestingly, the decrease in dive depth and dive duration for two Washington animals at 23 months of age (Fig. 5) corresponded with movement from inside Puget Sound to deeper waters off the Washington coast. The greatest proportion of all diving (37%) occurred dur- ing 2100-0300 h; the least (about 16%) during 0900-1500 h (Fig. 7). There were no periods when young-of-the-year or juvenile sea lions from any location did not dive. The fre- quency distribution of dives was similar in all time periods for all age groups from Alaska and Washington (Fig. 7). Distance and duration of trips at sea Mean distance of trips at sea for 564 measured trips of the 25 study animals was 16.6 km (SD=44.9 km; range: 572 Fishery Bulletin 101(3) Washington yearlings 21106 max depth 14085 max depth 350 300 250 200- 150- 100 50 • 11 21 31 Days with maximum depth (m) per 24-h reception 21103 max depth Days with maximum depth (m) per 24-h reception 14087 max depth 350 300 250- 200 150 100 50- Days with maximum depth (m) per 24-h reception 14089 max depth 11 21 31 41 51 Days with maximum depth (m) per 24-h reception 350 300 250 200 150 100 50 1 11 21 31 ■ll Days with maximum depth (m) per 24-h reception Figure 3 Summaries of the maximum depth (m) for each 24-h period (O) m which a status message containing maximum depth informa- tion was received by the Argos satellite for 13 of the 25 SDR-cquipped Steller sea lions. For example, the Argos satellite received a status message from SDR 21 106 for about 44 days, yet the SDR was operational for a total of 83 days (A; see Table 1 ). These 13 were chosen to compare yearlings of comparable age in Washington (A), and Alaska (B), and to provide typical examples of Alaska young-of-the-year (C). 10 months of age, the mean distance of all tripswas24.6km(«=307;range: '°° /\ 50 0 y.^,ee«ee««ee«/V^^V^^^^ 1 11 21 31 41 1 11 21 Days with maximum depth (m) per 24-h reception Days with maximum depth (m) per 24-h reception ?, 14076 max depth 14078 max depth a. 350- 360-. 4 300 300- 250 200 2S0' 200 aa a E^ 150 a 150 n / V ^^^"~9^ ^\ / \ y\ E E 100 50 0 -VV-vV--''^-*"--""""\j : f\i V 1/ 1 11 21 1 11 21 0} 5 350-1 300 250 200 150 ■ 100 ■ 50 Days with maximum depth (m) per 24-h reception 14170 max depth Days with maximum depth (m) per 24-h reception 1 11 21 31 41 Days with maximum depth (m) per 24-h reception Figure 3 (continued) cant gender (P = 0.6) nor gender x age interaction effects (P=0.19). Trip distance increased with age. For example, we cap- tured a 9-nionth-old male sea lion (bearing transmitter identification number PTT 21094, Table 1) near Kodiak Island in March 200. It had short trip distances (<10 km) which tended to concentrate near the capture site and nearshore (Fig. 8). As the animal matured through April and May, trip distance progressively increased until the sea lion was swimming over 50 km offshore beyond the 100- m depth contour and had a maximtim dive depth >150 m (Table 2; Fig. 8). Trip duration was measured for 10 of 25 animals with SDRs containing time-line data (it was not possible to cal- culate trip duration for 15 SDRs with the earlier SLTDRs that did not transmit time-line data). Mean trip duration for these 10 animals was 12.1 hours (n=544; SD=23.83 h; range: 1—344 h; median=7.3 h). For animals slO months 574 Fishery Bulletin 101(3) 350t 300 250 200- 150 100 50 0 Alaskan young of the year 14163 max depth 11 21 31 41 Days witti maxmum depth (m) per 24-h reception 14164 max deptti 1 11 21 31 Days with maxmum depth (m) per 24-h reception 21094 max depth 350 SOO- TS o £ 250 a. i 200 ■T f 150 a 100 m 50 S 0 -e-^-e-^ 11 21 31 Days with maxmum depth (m) per 24-h reception Figure 3 (continued) ■ Alasl10 months of age, the mean duration of all trips was 18.1 hours («=237; SD=34.2 h; range: 1-344 h; median=10.3 h). Averaged across individual animals, the mean duration of trips at sea ranged between 6.2 to 21.4 hours; this range was 6.2 to 17.2 hours for the younger animals and 10.3 and 21.4 hours for the older animals. The analysis of the re- peated-measures ANOVA on the logarithm of trip duration showed that the older sea lions had longer trip durations (P<0.001). We could not test for gender and gender x age effects because there were no measured trip durations for females >10 months. Among the younger animals, there was no gender difference in mean trip duration (P=0.11). Types of movement We identified three types of movements for the sea lions at sea: long-range trips (>15 km and >20 h), short-range trips (<15 km and <20 h), and transits to other haul-out sites (Fig. 9). Long-range trips most likely were foraging trips and began around 9 months of age. These trips had a mean of 48.7 km (SD=55.7 km; max=240.8 km) and may coincide with the assumed onset of weaning; they represented 6% of all trips to sea. The most numerous trips (S&'Yc ) were short- range foraging trips (v=3.6 km; SD=0.4; max=21.0 km), which happened almost daily (0.9 trips/d, n=A2& trips). Transits were movements from one haul-out site to another haul-out site; these trips were characterized as the straight line distance from one haul-out site to another and began as early as 7 months of age but occurred more often after 9 months of age. Transit trips represented 6% of all trips at sea and had a mean distance of 66.6 km (SD=83.7 km; range: 6.5-341.9 km). Discussion The differences in diving behavior between young Steller sea lions in Washington and those off Alaska are intriguing. Possible reasons for these differences include variable habi- tat type, prey resources, or morphological or genetic differ- ences. However, there is no evidence, based on morphology or genetics, to either support or refute differences in the diving behavior that we observed. The evidence of genetic differences between the western and eastern stock of Steller sea lions is based on mtDNA haplotype differences for a seg- ment of the mitochondrial D-loop which does not code for any structural proteins (Bickham et al., 1996; Loughlin, 1997). One morphological difference between the two stocks is a progressive increase in mass of Steller sea lion pups from east to west (Merrick et al., 1995), but whether this difference in mass continues with increasing age is un- known. Large animals typically dive deeper and longer than smaller (and younger) animals (Schreer and Kovacs, 1997). Larger animals have less drag per unit of mass and generally have more blood than smaller ones and thus are able to store more oxygen. Larger animals also have lower 576 Fishery Bulletin 101(3) 80% 70% - 60% 50% - 40% 30% 20% 10% 0% a t c d c ' g h 21 00-2 59 ■ a; Haul out n b 4 meters 0 c 4-6 meters Hd: 6-10 meters lae 10-20 meters Hf: 20-34 meters Hg: 34-50 meters Eh: 50-74 meters H i: 74-100 meters ??aTTi>-.^ a 0 c 6 (■ I g t\ \ atcdefgh 3:00-8:59 900-1459 Period Figure 6 Percentage of time spent at depth for seven young-of-the-year Steller sea lions approximately (A) 7-10 months of age in Alaska, and percentage of time spent at depth for three young-of-the-year Steller sea lions approximately (B) 11-12 months of age (during May and June) in Alaska. This figure suggests that as young sea lions approach one year of age they tend to spend less time hauled out and that a greater proportion of their dives are deeper. mass-specific metabolic rates than their smaller counter- parts and thus expend less energy and use less oxygen stores (Schreer and Kovacs, 1997). Our sample size of sea lions of comparable age is small; however, we compared the mean mass of three Washington sea lions to the mean mass of three Alaska sea lions of approximately the same age (Table 1) and found that the Alaska animals had less mass than those in Washington (108 kg vs. 145 kg). Whether or not this difference in mass can account for the differences we saw in diving characteristics for animals of similar age (Fig. 3, A and B) is unknown. The differences in diving characteristics between animals tracked in coastal waters of Puget Sound, Washington, and those tracked in Alaska waters are most likely linked to localized differences in prey habitat. The primary prey of Steller sea lions across their range are fish and ccphalo- pods, both of which have a broad but predictable range in temporal, spatial, and seasonal nearshore availability. Typically, each species makes predictable migrations sea- sonally from pelagic to nearshore waters where they form large spawning concentrations. The prey are then further concentrated by local transition boundaries such as frontal zones and bathymetric features such as submarine chan- nels (Sinclair et al., 1994). Steller sea lions appear to have the foraging flexibility to take advantage of both the pre- dictable behavioral traits of these prey species, as well as the localized oceanographic conditions that enhance prey concentrations (Sinclair and Zeppelin, 2002). The primary prey of Steller sea lions in Alaska waters is walleye pollock (Theragra chalcogramma ), which is con- sumed year-round (Sinclair and Zeppelin, 2002). Walleye pollock is replaced as a dominant year-round prey item by Pacific whiting (Mer/(/cc/(/.s productus ) in Pacific Northwest waters (Gcarin et al., 1999). Both species are semidemersal and can be found from near surface waters to depths >1200 m, depending on localized conditions (Hart, 1973; Esch- meyer et al., 1983). The greatest abundances of both species are available to Steller sea lions in nearshore waters over the continental shelf and perhaps as the prey become more available during nighttime diurnal vertical movements. The physical features of Puget Sound, along with its com- plex bathymetry and the extensive channels and canyons, provides extensive microhabitat for both predator and prey species to express the full extent of their depth range. In this respect, Puget Sound is comparable to the Gulf of Alaska where Pacific cod (Gadus macrocephalus) is the predominant winter prey item for Steller sea lions. Pacific cod is thought to be consumed during spawning when it ap- Loughlin et al : Diving behavior of immature Eumetoplas jubatus 577 80% 70% 60% S 50% Z 40% g 30% 20% 10% 0% B I- c e f g h 21 00-259 3:00-8:59 ■ a Haul out Db 4 meters Be: 4-6 meters Sd 6-10 meters De 10-20 meters Bf: 20-34 meters Hg 34-50 meters Dh 50-74 meters S i 74-100 meters ^iMt^ a b c rt o ' g 9:00-14:59 a D c d p f g 15 00-20:59 Period Figure 6 (continued) pears to concentrate in the deep nearshore channels and gulhes of the Gulf of Alaska (Sinclair and Zeppelin, 2002). The differences in dive depths that we re- port also could be typical of the variability among individuals. Boveng et al. ( 1996 ) ana- lyzed TDR data for six dive-related variables and found that dive duration was the least variable and vertical distance (dive depth) was the most variable among individual Antarctic fur seals (Arctocephalus gazella ). In our study, there was high individual vari- ability in both dive depth and maximum depth and little variability in dive dura- tion— results similar to those of Boveng et al.'s (1996) study A female Steller sea lion nurses her pups during the day, stays with the pup for the first week, then goes to sea on foraging trips. Maternal pup-attendance patterns seem to vary over the sea lion's geographic range; the average range of time for foraging trips during lactation are from about 24 h to 2 d at the southernmost rookery at Aho Nuevo Is- land, California (Higgins et al., 1988; Hood and Ono, 1997; but note that some of this variability may have been the result of El Nifio conditions during part of the Higgins et ■ Alaska YOV DAIaska yearling ■ Washington yearling Period (local time) Figure 7 Percentage of dives occurring in each time bin for Alaskan young-of-the-year ( YOY), Alaska yearling, and Washington yearling Steller sea lions. Twenty- five animals are represented from 1994 to 2000. al. study period), about 25 h at Lowrie Island, 19 h at Fish Island, 11 h for Chirikof Island, and 7 h in the Aleutian Islands (Brandon, 2000). 578 Fishery Bulletin 101(3) 153°0'W 152''0'W 58°0'N- 57°0'N' •58°0'N ■57°0'N 152°0'W isro'w Figure 8 Figure showing the progressive increase in distance of locations from shore for a B-nionth-oId sea lion over time. This animal (identification number: PTT 21094) was equipped with a satellite transmitter near Kodiak Island in March 2000. Early trips were <15 km from shore in water <50 m in depth near the capture site. Trips became progi"essively greater as the animal matured through May 2000 when it was venturing over 50 km from shore in water >100 m in depth. Ontogeny of diving ability has been studied in two other otariids. Baker and Donohue (2000) used data loggers (which they termed "time wet recorders") to measure time spent in the water and diving behavior of northern fur seal {Callorhinus iiniiriiis) pup.s on St. Paul Island, Alaska. These pups began spending substantial time in the water at approximately 40-50 d of age that coincided with growth of the under fur and increases in sea surface temperature. Time spent in the water increased up to about 100 d of age; diving to depth did not occur until they were much older and about to migrate. Horning and Trillmich (1997) conducted an extensive study on the ontogeny of diving behavior in Galapagos fur seals (Arctoccphaliis f^alapagnensis), a spe- cies that weans no sooner than 2 years of age. They found that in young the development of diving behavior was close- ly linked to dependence on the mother and that substantial diving activity did not occur until one year of age; but even then the young fur seals were still nutritionally dependent on their mothers and did not dive as deep, or for as long, as mature females. The weaning date for Steller sea lions is unknown but is assumed to be between 4 and 12 months, and most pups are weaned just before the next breeding season ( 11-12 months) (Porter, 1997). The change in diving characteristics that we report is interesting in that it coin- cides with this period. Prior to weaning these pups forage in the company of their mother and learn to forage on their own; the need to dive deep for long periods to acquire food is compensated by nursing from the mother. Once weaning Loughlin et al.: Diving behavior of immature Eumetopias lubatus 579 500'N 54<'0'N 74 O'W Figure 9 The three types of movement exhibited by two immature Steller sea Hons captured at Turf Point, Seguam Island, Alaska, in 2000. A long-range trip (solid circles) >200 km is shown for PTT 14163 as it left and returned to Turf Point. A transit trip (open triangles) for PTT 14111 is shown as it left Turf Point and remained at the east end of Amlia Island where it went on numerous short-range trips (shaded squares). occurs, the yearlings are forced to explore more areas to acquire food for needed energy. Dives become deeper and longer as these yearlings forage at different depths within the water column. Just before their first birthday, many of these young sea lions are capable of diving to the same depths and for the same duration as those of many adults; they also begin to forage at greater distances and for longer periods. Juveniles that we studied had a mean dive depth of 18.4 m and dive duration of 1.1 min compared to adult females in Alaska that had a mean dive depth of 21 m and dive duration of 1.4 min (Merrick and Loughlin, 1997). Maximum depth in our study was 328 m for a Washington juvenile and 288 m for an Alaska juvenile. Maximum depth information for adult females in Alaska was not provided by the instruments used by Merrick and Loughlin ( 1997); their maximum depths were characterized by bin data only. They showed that about 5% of dives by adult females in winter were greater than 250 m. In another study, adult females in Alaska were equipped with early-style SLTDRs that had features that recorded time-depth information and these SLTDRs showed that the females frequently dived to 200 m or more (Merrick et al., 1994). Schreer and Kovacs (1997) summarized maximum dive depth and dive duration for air-breathing vertebrates and developed predictive allometric equations for both param- eters based on body mass. We fitted our Steller sea lion body mass data to these equations to estimate maximum dive depth (27.33M^'''*^), where Afj represents body mass in kilograms, and maximum dive duration (6.22^/^"'"). We found that the maximum dive depth equation provided rea- sonably close estimates but that dive durations were typi- cally overestimated (Table 3). In some cases measured and 580 Fishery Bulletin 101(3) Table 3 The recorded mass, recorded maximum dive depth, and recorded maximum dive duration for individual young Steller sea lions in Alaska and Washington from this study and the estimated maximum dive depth (27.33Mj'''"') and estimate maximum dive dura- tion 1 6.22 M^" ' 1 based on allometric equations in Schreer and Kovacs ( 1997 ). PTT number is the satellite transmitter identification number. Est.= estimated, n/a = no data obtained. PTT number Mass (kg) Maximum depth (m) Est. maximum depth (m) Maximum duration (min) Est. maximum duration (min) 14073 86.26 168 212.38 >6 9.71 14084 77.18 288 201.79 >6 9.61 14085 154.36 280 277.57 >6 10.30 14087 122.45 256 249.52 >6 10.06 14089 111.11 200 238.61 4-6 9.96 21103 139.23 256 264.70 13 10.19 21106 143.31 328 268.24 >14 10.22 14071 92.00 12 218.77 2-3 9.78 14074 79.80 20 204.91 >6 9.64 14077 n/a 144 n/a >6 n/a 14078 n/a 288 n/a >6 n/a 14079 94.90 44 221.91 >6 9.81 14080 106.20 24 233.70 >6 9.92 14170 Est 95-105 180 Est. 222.02-232.48 >14 Est 9.81-9.91 1 21094 62.20 152 182.72 >14 9.40 14076 103.70 144 231.15 >6 9.89 14081 n/a 20 n/a >6 n/a 14072 116.10 100 243.48 >6 10.01 14075 104.00 12 231.46 >6 9.90 14164 79.60 60 204.67 >14 9.64 14167 100.20 60 227.53 3-4 9.86 14111 87.00 40 213.22 8-9 9.72 14114 85.80 16 211.86 7-8 9.71 14116 76.20 40 200.60 5-6 9.59 14163 109.00 252 236.51 >14 9.94 estimated maximum dive depth values differed by large amounts (e.g. sea lion PTT 14071), perhaps because the deployment period was brief, before deep dives occurred. For others (e.g. PTT 14074) the difference may have been due to the young animal's continued dependence on the female for nourishment; deeper dives do not occur until weaning. In addition, we note that our dive duration data were stored in bins of 1-min intervals (from 1 to 6 min in the early instruments and from 1 to 14 min in the recent ones); the exact duration of each dive is unknown. Movement patterns also suggest that the swimming ability of juvenile sea lions is comparable to that of adults. It is not unusual for young sea lions to travel distances as great as 1784 km from the natal rookery; as they approach adulthood they generally remain within 500 km of their natal rookery (Raum-Suryan et al., 2002). In our study some young sea lions traveled several hundred kilometers between sites while presumably searching for food or ven- turing from the natal rookery site. Further analysis of our SDR data is warranted to more fully understand sea lion diving behavior and its relation- ships with oceanographic parameters, daily and season change, and behavioral features as discussed by Fedak et al. (2001). The time allocation at depth (TAD) index de- scribed by them will be a useful method for interpretation of our SDR (and TDR) data. Further analysis of our SDR data is needed to determine if such a study is possible. Acknowledgments Field assistance was provided by numerous NMML staff; logistical support was provided by Alaska Helicopters and the captain and crew of the U.S. Fish and Wildlife Service research vessel Tiglax. The manuscript was improved by comments from D. DeMaster, G. Duker.T. Gelatt, R. Hobbs, M. Lander, J. Lee, R. Ream, E. Sinclair, and two anonymous reviewers. Literature cited Baba. N., H. Nitto, and A. Nitta. 2000. Satellite tracking of young Steller sea lion off the coast of northern Hokkaido. Fisheries Sci. 66:180-181. Loughlin et al.: Diving behavior of immature Eumetopias jubatus 581 Baker, J. D., and M. J. Donohue. 2000. Ontogeny of swimming and diving in northern fur seal iCallorhiniis iirsin us) pups. Can. J. Zool. 78:100-109. Bickham, J. W., J. C. Patton, and T. R. Loughlin. 1996. High variability for control-region sequences in a marine mammal: implications for conservation and bio- geography of Steller sea lions (Eumetopias jubatus). J. Mammal. 77:95-108. Boveng, P. L., B. G. Walker, and J. L. Bengtson. 1996. Variability in Antarctic fur seal dive data: implications for TDK studies. Mar Mamm. Sci. 12:543-554. Boyd. I. L. 1993. Selecting sampling frequency for measuring diving behavior. Mar Mamm. Sci. 9:424-430. Boyd, I. L., J. P Y. Arnould. T. Barton, and J. P. Croxall. 1994. Foraging behaviour of Antarctic fur seals during periods of contrasting prey abundance. J. Animal Ecol. 63:703-713. Brandon, E. A. A. 2000. Maternal investment in Steller sea lions in Alaska. Ph.D. diss., 137 p. Texas A&M University, Galveston, TX. Calkins, D. G., E. Becker, and K. W. Pitcher 1998. Reduced body size of female Steller sea lions from a declining population in the Gulf of Alaska. Mar. Mamm. Sci. 14:232-244. Eschmeyer, W. N., E. S. Herald, and H. Hammann. 1983. A field guide to Pacific coast fishes of North America, 336 p. Boston Houghton Mifflin Company, Boston, MA. Fancy, S. G., L. F Pank, D. C. Douglas, C. H. Curby G. W. Gamer, S. C. Amstrup, and W. L. Regelin. 1988. Satellite telemetry: a new tool for wildlife research and management. U.S. Fish and Wildl. Serv. Resour Publ. 172:154. Fedak, M. A., P. Lovell, and S. M. Grant. 2001. Two approaches to compressing and interpreting time-depth information as collected by time-depth record- ers and satellite-linked data recorders. Mar. Mammal Sci. 17:94-110. Gearin, P., S. Jeffries, S. Riemer. L. Lehman, K. Hughes, and L. Cooke. 1999. Prey of Steller's sea lions, Eumetopiasjubatus, in Wash- ington state. In Abstracts of the 13th biennial conference on the biology of marine mammals, Wailea, Hawaii November 28 December 3, p. 65. Soc. Marine Mammalogy, Wailea, HL Gentry, R. L., and G. L. Kooyman. 1986. Methods and dive analysis. In Fur seals, maternal strategies on land and at sea (R. L. Gentry and G. L. Kooy- man eds. ), p 28-40. Princeton Univ. Press, Princeton, NJ. Goebel, M. E., J. L. Bengtson, R. L. DeLong, R. L. Gentry, and T. R. Loughlin. 1991. Diving patterns and foraging locations of female northern fur seals. Fish. Bull. 89:171-179. Hart, J. L. 1973. Pacific fishes of Canada. Bull. Fish. Res. Board Can. 180, 740 p. Higgins, L. V., D. P. Costa, A. C. Huntley, and B. J. LeBoeuf 1988. Behavioural and physiological measurements of maternal investment in the Steller sea lion, Eunietopias jubatus. Mar Mamm. Sci. 4:44-58. Hood, W. R., and K. A. Ono. 1997. Variation in maternal attendance patterns and pup behaviour in a declining population of Steller sea lions [Eumetopiasjubatus). Can. J. Zool. 75:1241-1246. Homing, M., and F. Trillmich. 1997. Ontogeny of diving behavior in the Galapagos fiir seal. Behaviour 134:1211-1257. Keating, K. A. 1994. An alternative index of satellite telemetry location error J. Wildl. Manage. 58:414-421. Kooyman, G. L., J. O. Billups, and D. W. Farwell. 1983. Two recently developed recorders for monitoring diving activity of marine birds and mammals. In Experi- mental biology at sea (A. G. Macdonald and L G. Priede, eds. ) p. 187-214. Academic Press, New York, NY. Loughlin, T. R. 1997. Using the phylogeographic method to identify Steller sea lion stocks. In Molecular genetics of marine mammals (A. Dizon, S. J. Chivers, and W. F. Perrin, eds.), p. 159-171. Spec. Publ. 3 of the Soc. Mar Mammal. Loughlin, T. R, J. L. Bengtson, and R. L. Merrick. 1987. Characteristics of feeding trips of female northern fur seals. Can. J. Zool. 65:2079-2084. Loughlin, T. R., and R. L. Merrick. 1989. Comparison of commercial harvest of walleye pollock and northern sea lion abundance in the Bering Sea and Gulf of Alaska. In Proceedings of the international symposium on the biology and management of walleye pollock, Nov. 14-16, 1988, Anchorage, Alaska, p. 679-700. Alaska Sea Grant Rep. 89-01, Univ. Alaska, Fairbanks. Loughlin, T R., A. S. Pedov, J. D. Baker, S. A. Blokhin, and A. G. Makhnyr. 1998. Diving behavior of adult female Steller sea lions in the Kuril Islands, Russia. Biosph. Cons. 1:21-31. Loughlin, T. R., A. S. Perlov, and V. A. Vladimirov. 1992. Range-wide survey and estimation of total number of Steller sea lions in 1989. Man Mamm. Sci. 83:220-239. Loughlin, T. R., and T Spraken 1989. Use of Telazol to immobilize female northern sea lion (Eunietopias jubatus) in Alaska. J. Wildl. Dis. 25: 353-358. Loughlin, T R., and A. E. York. 2000. An accounting of the sources of Steller sea lion mortality Mar. Fish. Rev 62(4):40-45. McConnell, B. J., C. Chambers, and M. A. Fedak. 1992. Foraging ecology of southern elephant seals in relation to the bathymetry and productivity of the southern ocean. Antarctic Science 4:393-398. McCullagh, P, and J. A. Nelder 1989. Generalized linear models, 2nd ed., 261 p. Chapman and Hall, London. Melin, S. R. 2002. The foraging ecology and reproduction of the Califor- nia sea lion (Zalophus californianus californianus). Ph.D. diss., 150 p. Univ. Minnesota, St. Paul, MN. Merrick, R. L. 1995. The relationship of the foraging ecology of Steller sea lions (Eunietopias jubatus) to their population decline in Alaska. Ph.D. diss., 171 p. Univ Washington, Seattle, WA. Merrick, R. L., R. Brown, D. G. Calkins, and T. R. Loughlin. 1995. A comparison of Steller sea lion, Eunietopias jubatus, pup masses between rookeries with increasing and decreas- ing populations. Fish. Bull. 94:753-758. Merrick, R. L., and T R. Loughlin. 1997. Foraging behavior of adult female and young-of-the- year Steller sea lions in Alaskan waters. Can. J. Zool. 75: 776-786. Merrick, R. L., T. R. Loughlin, G. A. Antonelis, and R. Hill. 1994. Use of satellite-linked telemetry to study Steller sea lion and northern fur seal foraging. Polar Res. 13: 105-114. Merrick, R. L., M. K. Maminov, J. D. Baker, and A. G. Makhnyr. 1990. Results of US.-U.S.S.R. joint marine mammal re- 582 Fishery Bulletin 101(3) search cruise in the Kuril and Aleutian Islands 6 June-24 July 1989. U.S. Dep. Commer. NOAA Tech. Memo. NMFS F/NWC-177,63p. Pitcher, W., and D. G. Calkins. 1981. Reproductive biology of Steller sea lions in the Gulf of Alaska. J. Mammal. 62:599-605. Porter, B. 1997. Winter ecology of Steller sea lions iEumetopias juba- tus) in Alaska. M.S. thesis, 84 p. Univ. British Columbia, Vancouver, B.C., Canada. Raum-Suryan, K. L., K. W. Pitcher, D. G. Calkins, J. L. Sease, and T. R. Loughlin. 2002. Dispersal, rookery fidelity, and metapopulation struc- ture of Steller sea lions (Eumetopias jubatus) in an increas- ing and declining population in Alaska. Mar. Mamm. Sci. 18:746-764. Schreer, J. F., and K. T. Kovacs. 1997. Allometry of diving capacity in air-breathing verte- brates. Can. J. Zool. 75:339-358. Service-Argos. 1984. Location and data collection system user's guide, 36 p. Service-Argos, Toulouse, France. Sinclair, E. H., T. R. Loughlin, and W. Pearcy. 1994. Prey selection by northern fur seals (Callorhinus ursi- nus) in the eastern Bering Sea. Fish. Bull. 92:144-156. Sinclair, E. H., and T. K. Zeppelin. 2002. Seasonal and spatial differences in diet in the western stock of Steller sea lions {Eumetopias jubatus). J. Mamm. 83:973-990. Stewart, B. S., S. L. Leatherwood, P. K. Yochem, and M. P. Heide-Jorgensen. 1989. Harborseal tracking and telemetry by satellite. Mar Mamm. Sci. 5:361-375. Trillmich, F, and K. A. Ono eds. 1991. Pinnipeds and El Nifio, responses to environmental stress. Ecological studies 88, 293 p. Springer-Verlag, Berlin. Vincent, C, B. J. McConnell, V. Ridoux, and M. A. Fedak. 2002. Assessment of Argos location accuracy from satellite tags deployed on captive gray seals. Mar. Mamm. Sci. 18: 156-166. Werner, R., and C. Campagna. 1995. Diving behaviour of lactating southern sea lions (O/aria flavescens) in Patagonia. Can. J. Zool. 73:1975-1982. 583 Abstract— Two halfbeak species, bal- lyhoo {Hemiramphus brasiliensis) and balao (//. balao), are harvested as bait in south Florida waters, and recent changes in fishing effort and regula- tions prompted this investigation of the overlap of halfbeak fishing grounds and spawning grounds. Halfbeaks were sampled aboard commercial fishing ves- sels, and during fishery-independent trips, to determine spatial and tem- poral spawning patterns of both spe- cies. Cyclic patterns of gonadosomatic indices (GSIs) indicated that both species spawned during spring and summer months. Histological analysis demonstrated that specific stages of oocyte development can be predicted from GSI values; for example, female ballyhoo with GSIs >6.0 had hydrated oocytes that were 2.0-3.5 mm diameter Diel changes in oocyte diameters and histological criteria demonstrated that final oocyte maturation occurred over a 30- to 36-hour period and that ballyhoo spawned at dusk. Hydration of oocytes began in the morning, and ovulation occurred at sunset of that same day; therefore females with hydrated oocytes were ready to spawn within hours. We compared maps of all locations where fish were collected to maps of locations where spawning females (i.e. females with GSIs >6.0) were collected to deter- mine the degree of overlap of halfbeak fishing and spawning grounds. We also used geographic information system (GIS) data to describe the depth and bottom type of halfbeak spawning grounds. Ballyhoo spawned all along the coral reef tract of the Atlantic Ocean, inshore of the reef tract, and in association with bank habitats within Florida Bay. In the Atlantic Ocean, balao spawned along the reef tract and in deeper, more offshore waters than did ballyhoo; balao were not found inshore of the coral reef tract or in Florida Bay. Both halfbeak species, considered together, spawned throughout the fish- ing grounds of south Florida. Spawning cycles and habitats for ballyhoo {Hemiramphus brasiliensis) and balao (H. balao) in south Florida Richard S. McBride Justin R. Styer Rob Hudson Florida Marine Research Institute Florida Fish and Wildlife Conservation Commission 100 8th Avenue SE St. Petersburg, Flonda 33701-5095 E-mail address (for R. S. McBride): nchardmcbnde^fwc-Statefl-US Manuscript approved for publication 30 January 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish. Bull 101:583-589 (2003). Combined landings of two halfbeaks species, ballyhoo (Hemiramphus brasil- iensis) and balao (H. balao), constitute a small but valuable bait fishery in south Florida (Berkeley et al., 1975; McBride et al., 1996). Both species occupy coastal pelagic habitat in association with coral reefs (Starck, 1968; Nybakken, 1997). During the 1990s two changes in the halfbeak fishery occurred that caused concerns regarding the exploi- tation levels in this fishery (McBride, 2001). First, geographic shifts occurred when halfbeak fishing expanded from the Atlantic Ocean into the nearshore waters north of the Florida Keys, an area known as Florida Bay. Second, changes in statewide net fishing regu- lations^ created concerns that the net fishermen displaced from other fisheries might preferentially enter the halfbeak fishery, thereby increasing halfbeak fishing effort. These two changes could have specific consequences on halfbeak reproductive output. For example, because some fishermen viewed Florida Bay as a spawning or nursery ground for halfbeaks, it was of interest to learn exactly how concentrated spawning might be in Florida Bay and whether spawning occurred outside Florida Bay. In addition, because halfbeak landings are dominated by a single species, bal- lyhoo (Berkeley et al., 1975; McBride et al., 1996), it could be argued that these changes in the fishery could dis- proportionally affect spawning by the less abundant, and potentially more vulnerable target species, balao. Both ballyhoo and balao are distrib- uted widely in the western and eastern Atlantic Ocean (Collette, 1965), but no study has defined their spawning grounds. Berkeley and Houde (1978) described both species to be small (<32 cm fork length) summer-spawners that rarely live past two years, but in terms of spatial coverage, they collected fish principally from the Miami, Florida, area. We reviewed reports on regional ichthyoplankton collections (e.g. Powell et al., 1989; Limouzy-Paris et al, 1994) and found that the numbers of halfbeak eggs and larvae were too few for char- acterizing the spawning grounds. More- over, Berkeley and Houde (1978) sug- gested that standard ichthyoplankton survey data would underestimate the abundance of halfbeak eggs or larvae for three reasons. First, halfbeak eggs appear to attach to vegetation; therefore oblique tows may not target the appro- priate habitats (i.e. benthic or floating vegetation) and halfbeak eggs would be completely lost if pleuston was dis- carded from ichthyoplankton samples. Second, halfbeak eggs hatch 8-9 days after fertilization and may disperse far away from spawning locations. Third, halfbeak larvae hatch at 5-7 mm and have pigmented eyes; therefore they appear capable of avoiding plankton nets. Various plankton sampling strate- ■ This referendum (s. 16, Art. X of the Florida Constitution, enacted July 1, 1995) prohib- its entangling nets in waters inshore of 3 miles on the Atlantic coast and 9 miles on the Gulf coast of Florida (including Florida Bay). It also prohibits non-entangling nets larger than 500 fl^ (such as those nets used by commercial halfbeak fishermen), in waters less than 1 mile of Florida's Atlan- tic coast and 3 miles of the Gulf coast. 584 Fishery Bulletin 101(3) gies could be developed to overcome these problems, but we chose an alternative to plankton sampling as a way to define halfbeak spawning grounds. B2 00' 81 30' &^^~ Atlantic Ocean 82°00*W 8r30W 8100W 80'30'W SO'OO' W 82°0Q' 8130' 8100' 80''3Q' 80°00' >00 m Isobath Atlantic Ocean . i"> ■ ^ ^^ 8200' W BV'30' W B2 00' 81 '30' 00' W 80"30' W 80°00' W iO'3Q' 60 00' Allanllc Ocean B2 00 w erao'W eroo'W aoao'W 80"00'W In this study, we used collections of adult ballyhoo and balao to define each species" spawning grounds in south Florida. Analyses of gonad histological preparations iden- tified a discrete range of gonadosomatic indices (GSI) for females that were ready to spawn within hours, and the locations of these fishes were plotted by using geographic information system (GIS) software (Arc View, version 3.3., Environmental Systems Research Institute, Inc., Redlands, CA). This synthesis of GIS and GSI data was used to map the spawning grounds of ballyhoo and balao. Materials and methods Sampling occurred throughout the south Florida com- mercial halfbeak fishing grounds, from Palm Beach to the Marquesas Keys (Fig. lA). The area immediately sur- rounding Vaca Key was well sampled, but other sections of the middle Florida Keys were not because commercial fishermen used only six fishing ports and their day trips were of limited range. Few samples were obtained from the Palm Beach area because net fishing is no longer allowed in much of this area (McBride, 2001). Halfbeak fishing trips by commercial fishermen were monitored from November 1995 to April 1999 by an onboard biologist during as many as four trips per month. A subsample of fish from the first successful net (a lampara net) set, and occasionally from later sets within a day, was obtained by filling a 5-gallon bucket from the catch as it was transferred from the net to holding boxes. This bucket held 100 to 200 halfbeaks, and these fishes were kept on ice and brought back to the laboratory for processing. Fishery-independent collections were made by using cast nets and small hooks (sabiki rigs) in the middle Florida Keys. This sampling was specifically designed to include inshore areas where lampara net fishermen could not fish because of regulations associated with Florida's net limita- tion referendum.' The target number of these fishery-inde- pendent trips from July 1997 to October 1998 was four per month, and the target sample size for each trip was twelve fish. Additional fishery-independent sampling occurred in the springs of 1997, 1998. and 1999. In the laboratory, whole body weight was recorded to the nearest 0.1 gram, and the gonads were removed and weighed to the nearest 0.01 g. Sex was identified with the aid of a dissecting binocular microscope (25-50x) when nee- Figure 1 (A) Sampling area for halfbeaks {Hi'iniramphu.'! spp.) during 1995-99 in the Atlantic Ocean and in Florida Bay. Each symbol represents an individual sample location where halfbeaks were caught. Fishery-dependent samples (triangles) were taken from commercial lampara net vessels. Fishery-independent samples (squares) were collected in the middle Florida Keys, near Vaca Key (not labeled because of the density of square symbols). Locations of ripe female (B) ballyhoo (Hemiramphus brasiliensis) and (C) balao (H. balao) are plotted separately. Ripe females have hydrated eggs, and this condition was determined in B and C by a gonadosomatic index >6.0 (see text for supporting evidence). McBride et al.: Spawning grounds of Hemiramphus brasiliensis and H. balao 585 essary. Weights offish collected from July 1997 to October 1998 were measured for up to 30 females per species per trip. Fish body and gonad weights were only occasionally recorded for other trips during 1995-99, but these data were included in the mapping of ripe females (i.e. females with hydrated oocytes in ovigerous lamellae) to increase overall sample size. In total, weight data were collected for 2908 halfbeak females from 79 commercial fishing net sets (63 different fishing days) and 59 fishery-independent sampling events (50 different sampling days). Commercial catches contributed 1649 ballyhoo and 757 balao females, and fishery-independent collections added another 497 bal- lyhoo and 5 balao females. The gonadosomatic index (GSI) was calculated as GSI=(GWnBW-GW)) X 100, locations were recorded by using a global positioning sys- tem hand-held unit. The latitude and longitude of fishery- dependent samples were taken onboard the fishing vessel once the lampara net enclosed the fish. Location data for fishery-independent samples were taken from an anchored position. Depth information was divided into the following categories: area exposed at low tide, 0-1 m, 1-2 m, 2-4 m, 4-6 m, 6-10 m, and 10-20 m; only one of 159 locations was without depth information. Substrate information was di- vided into one of the following categories; platform margin reefs, patch reefs, other hard bottom, seagrass beds, and bare substrate; 32 of 159 locations did not have substrate information. Results where GW = gonad weight; and BW = body weight. The processes of final oocjrte maturation (FOM) were examined by comparing GSIs with changes in whole oocyte size and histological criteria. Oocyte diameters were measured for 39 ballyhoo collected in April 1998 and March-April 1999. Fixed ovary tissue was washed, teased apart, and placed in a solution of 33% glycerin to 67% water Measurements of at least 300 oocytes per fish were made to the nearest micron with the aid of a video system and image-analysis software (Optimas, vers. 100, Media Cyber- netics, Inc., Silver Spring, MD). A minimum-size cut-off of 0.15 mm was used to exclude debris within the petri dish. Initially, oocyte diameters from six ballyhoo were measured from four separate sections of ovaries (left, right, anterior, posterior), but the modal oocyte diameter within each in- dividual was the same for all four sections; therefore tissue from other fish was extracted without regard to location within the ovary. Berkeley and Houde (1978) performed a similar test and came to a similar conclusion. Ovaries from fish {n=930 females) collected during March and May 1997, July 1997-October 1998, and March-April 1999 were pre- pared for histological examination. Histological methods are presented in McBride and Thurman (2003). Here, the most advanced oocyte stage was recorded (in increasing order of oocyte maturity) as either perinucleolar, cortical al- veolar, vitellogenic. or as either of two stages for oocytes in final maturation; nucleus migration and hydration (West, 1990 ). McBride and Thurman ( 2003 ) have reported the size at 50% maturity to be >160 mm FL for female balao and >198 mm FL for female ballyhoo (approximately 31.5 gand 60.9 g, respectively, using length-weight relationships from Berkeley and Houde [1978; their Fig. 71). Mean GSIs and 95% confidence limits were determined for fish with regard to their most advanced stage of oocyte development, and a minimum cut-off value was established for GSI values indicating ripe females. The locations of ripe females were plotted to indicate spawning grounds. Water depth and bottom type of these spawning locations were determined by using the Marine Resources Geographic Information System at the Florida Marine Research Institute (w-ww.floridamarine.org). Point Spawning cycles Ballyhoo and balao had prolonged spawning seasons that peaked in late spring and early summer (Fig. 2). Monthly average GSIs of mature females increased from a low of <0.4 for both species to a high of 6.4 for balao and 6.9 for ballyhoo. The average GSI of individual females with only primary growth oocytes (i.e. their most advanced oocyte stages were perinucleolar or cortical alveolar) fell within a narrow interval of 0.1-0.3 (maximum=0.95, Fig. 3). These females were either small fishes that were immature or they were larger fishes that were regressed (i.e. mature but inactive). Vitellogenesis more than doubled the aver- age GSI values for both species, but all females whose most advanced oocyte stage was vitellogenic had GSIs less than 1.37. Dramatic increases in GSI values also occurred during FOM, and significant differences were evident in the sequential FOM steps of nucleus migration and nucleus breakdown. During nucleus migration, but before hydra- tion, ballyhoo GSIs averaged 3.4 (3.3-3.6; 95% CD and balao GSIs averaged 5.2 (4.6-5.9). Females with hydrated oocytes had GSIs averaging 7.4 (7.0-7.9; 95% CD for ballyhoo and 8.7 (6.6-10.8) for balao. Individual female GSIs reached an observed maximum of 13.3 for ballyhoo and 14.2 for balao. By applying these GSI criteria, which indicate that females with a GSI greater than about three had oocytes in FOM, it is evident that the average mature halfbeak female is actively spawning from at least March to August. Final oocyte maturation also followed a diel cycle. For ballyhoo, FOM began about 30-36 hours before ovulation, hydration of oocytes began about 8-12 hours before ovula- tion, and ovulation occurred at sunset. Ballyhoo oocytes de- veloped in a group-synchronous pattern, and during FOM, a batch of oocytes increased rapidly in diameter (Fig. 4). Mature female ballyhoo had a bimodal or trimodal distri- bution of oocyte diameters when spawning. The smallest mode (<1.0 mm oocyte diameter) represented a reservoir of primary growth oocytes and vitellogenic oocytes. Larger modes, between 1 and 3 mm, represented oocytes in FOM. The presence of two larger ooctye modes (-1.0-2.0 and >2.0 mm) in all females sampled during the afternoon period in- dicated that female ballyhoo typically spawn every day dur- 586 Fishery Bulletin 101(3) ing March-April. In total, the trimodal oocyte frequency represents a reservoir of oocytes prior to FOM, one batch of oocytes beginning FOM, and one batch completing FOM. Hydrated oo- cytes were not observed in balao prior to 1 100 h. However, in several ballyhoo collected around dawn (i.e. at approximately 0600-0700 EST), the nucleus in oocytes of the advanced batch was still visible along the chorion but the cy- toplasm was lightening in color. This suggested that initiation of hydration at daybreak briefly preceded nucleus breakdown. During the fol- lowing 12-hour period, oocytes in this maturing clutch advanced from late nucleus migration to nucleus breakdown, and increased in diameter from 1.5-2.0 mm in the morning to 2.0-3.0 mm in the afternoon. Modal egg size for each of three running-ripe ballyhoo (i.e. females with hydrat- ed, ovulated eggs in the ovarian lumen) was 2.35, 2.60, and 2.80 mm diameter The complete size range of these ovulated eggs was 2.2-3.4 mm diameter These females were collected at or just before sunset (time: 1810-1855). Most efforts to sample across the full 24-hour cycle failed, apparently because halfbeak do not bite hooks after sundown. A sample of 12 ballyhoo was collected one night, however, by randomly throwing a cast net on dense schools of fish. These fish, collected between 2200 and 2359 hours during March 1997, all appeared to have recently spawned. Histological preparations demonstrated that they had fresh postovulatory follicles, and they contained a distinct clutch of oocytes in early nucleus migration. Whole oo- cytes from these fish collected at night were not archived in formalin; therefore they were not measured for comparisons to whole oocytes col- lected at other times during the diel cycle. These patterns of diel reproductive periodicity also ap- peared to apply to balao, but the available data were not conclusive. Spawning habitat o a -Ballyhoo, n=1,791 Female GSI Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct 1997 Month 1998 Figure 2 Mean (±95'^ confidence limits) gonadosomatic indices by month for bal- lyhoo (Hemiramphus brasiUensis) and balao (H. balao) females. Values are calculated for fish larger than size at 50% maturity, reported by McBride andThurman (2003) as >198 mm fork length (PL) for ballyhoo and >160 mm FL for balao. n = number of females. Perinucleolar Cortical Vitellogenic Nucleus Hydration 154/7 Alveolar 24/8 Migration -|20/7 159/24 360/57 Most advanced ooctye stage Figure 3 Mean (±95% confidence limits) gonadosomatic indices of female bally- hoo (Hemiramphus brasiUensis) and balao (H. balao) in various stages of oocyte development. A two-way ANOVA demonstrated a significant effect of both developmental stage and species (P<0.0001 ) on gonadoso- matic index. Numbers indicate the number of fish, by stage, for each species. Hydration = nucleus breakdown. In our study it was shown that hydrated oocytes can be inferred from a threshold criterion of GSI >6.0 (Fig. 3), and ripe females (i.e. with a batch of hydrated oocytes) will spawn within hours. Ripe ballyhoo females were distributed throughout the fishing grounds in both the Atlantic Ocean and Florida Bay (Fig. IB). In the Atlan- tic, ripe ballyhoo females were caught in water depths from 1 to 20 m (mode: 6-10 m, 36.3'7( of the sets containing ripe ballyhoo in the Atlantic Ocean). In Florida Bay, ripe bal- lyhoo females were caught in areas that were exposed at low tide and out to 6-m deep (mode: 2-4 m; 57.9'7f of the positive sets in Florida Bay). Ripe ballyhoo females were mainly associated with hard bottom or vegetated habitats in both areas. In the Atlantic Ocean, ripe ballyhoo females were collected above platform reefs in 51.79i of the sets, above seagrass beds in 37.9% of the sets, near patch reefs in 5.2% of the sets, and over bare substrate in 5.2% of the sets. In Florida Bay, these fish were also associated with hard bottom substrates, specifically with vegetated bank habitat, in 44.77r of the sets and with seagrass beds in 55.3% of the sets. Ripe balao females were distributed throughout the Atlantic fishing grounds but not in Florida Bay (Fig. IC). In the Atlantic, they tended to occur in deeper water than did ripe ballyhoo females (range: 2-20 m; mode: 10-20 m, 51.37f of the sets containing ripe balao). The habitat as- sociations of ripe balao females were similar to those of ripe ballyhoo females in the Atlantic Ocean, but typically reflected areas offshore rather than inshore of the reef In the Atlantic Ocean, ripe balao females were collected above platform reefs in 58.0% of sets, above seagrass beds McBride et al.: Spawning grounds of Hemiramphus braslllensis and H. balao 587 n=3 1542- 1547 h 00 05 10 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Oocyte diameter (mm) Figure 4 Oocyte diameters for ballyhoo [Hemiramphus brasiliensis) collected at different times of the day (EST). At least 300 oocytes were measured per female. Sizes <0.25 mm were not measured representatively because of the minimum size criteria of the image analysis software. Sunset occurred at 1830-1900 EST during this collection period (March-April), n = number of females. in 25.8% of sets, over bare substrate in 9.7% of sets, and above undefined hard bottom in 6.5% of sets. Discussion These detailed findings of prolonged summer-spawning sea- sons, extreme iteroparity, and diel reproductive periodicity are consistent with other studies of halfbeak reproductive biology. Graham (1939), Ling (1958), Talwar (1962, 1967), and Berkeley and Houde ( 1978) noted a protracted spawn- ing season by hemiramphids during warm months. McBride and Thurman (2003) examined the frequency of postovula- tory follicles and reported that both species spawn daily during late spring and early summer, but also that some portion of the ballyhoo population spawns year-round. The present study is the first to follow the diel progression of FOM within the family Hemiramphidae. Lunar periodicity was not evident but it may have been confounded by the highly iteroparous nature of both species. Spawning halfbeaks were distributed so widely through- out the fishing grounds that no specific areas were identi- fied for the protection of spawning individuals. We noted interspecific differences in spawning areas, but these are not necessarily related to preferences by spawning females per se. Instead these differences appeared to be the result of interspecific distribution patterns of adult halfbeaks in general (i.e. adult ballyhoo are a more inshore species com- pared to adult balao [McBride, pers. obs.] ). Because balao were not found in Florida Bay, fishing in Florida Bay does not affect this species. Spawning by ballyhoo was evident in Florida Bay, as predicted by fishing industry partici- pants, but spawning ballyhoo were also widespread along south Florida's coral reef tract. Existing, albeit recent, 588 Fishery Bulletin 101 (3) regulations^ should provide some measure of protection for spawning ballyhoo in inshore waters. Our study design was limited to the presence and ab- sence of spawning females and did not identify concentra- tions of spawning activity associated with specific habitats. Presumably submerged vegetation is an important micro- habitat. Several authors have noted that hemiramphid eggs, including those of ballyhoo, attach by filaments (of the chorion) to vegetation such as Syringoditim fill forme and Sargassum sp. in waters less than approximately 6 m deep (Graham, 1939; Ling, 1958;Talwar, 1962, 1967; Berke- ley and Houde, 1978). However, Berkeley and Houde ( 1978) collected eggs in plankton tows. The specific importance for halfbeak reproductive success of attached versus floating vegetation, or no vegetation, has not been identified. The methods of this study define the macroscale spawn- ing habitat of halfbeaks based on the distribution of spawn- ing females. We demonstrate here that GSl values, even for highly iteroparous species, can distinguish females with hydrated oocytes from females in a less advanced stage of oocyte development. The GSI value is simple and inex- pensive to measure, and by including individual halfbeaks for which we had GSI values but no histological data, we more than tripled our sample size with little additional laboratory cost. We could have instead characterized oo- cyte development macroscopically and such a modification is well suited when conditions affect weighing devices. But macroscopic characterization of oocyte development usually follows an ordinal scale that may vary between observers. The distribution of females with hydrated eggs may be a better indication of spawning habitat than the distribu- tion of eggs because hydration occurs for only a few hours (DeMartini and Fountain, 1981; Hunter and Macewicz, 1985; Brown-Peterson et al., 1988; McBride et al., 2002), whereas egg dispersal may occur over several days. In this study we assumed that spawning females move only limited distances within the few hours of the hydration process, and although limited movement has not been documented for either ballyhoo or balao, we believe that our interpretation of the data supports this assumption. The size of the study area was approximately 200 km by 250 km, and it seems reasonable that spawning halfbeaks were not moving extensively within this spatial boundary on an hourly basis. The approach discussed in the present study may meet the needs of other investigators wanting to generate a first approximation of spawning habitats for management purposes, which was the goal of this study. Also, this approach has good potential for use in areas were species identification of halfbeak eggs or larvae is problem- atic (Noell et al., 2001). Analyses requiring a smaller area or finer spatial resolution will depend on verification of a hydration period that is short in relation to expected fish movements. The specific example presented in our study was limited because we collected the fish using commercial fishing vessels on routine fishing operations. This was cost-effec- tive, but wo were not able to identify spawning habitat preference or to define the complete geographic extent of the spawning grounds within south Florida. Gaps in the distribution of ripe females, which were particularly evi- dent in the middle Florida Keys, were typically related to gaps in sampling coverage. In addition, both species pre- sumably spawn outside the area we sampled. Still, much of the reported geographic range of ballyhoo and balao in the western Atlantic Ocean has been covered in the pres- ent study. The remaining shortcomings of this specific example could be resolved by using this approach within a statistically valid sampling design and estimating size- specific batch fecundity to map reproductive rates within a spatial and temporal context. The data resulting from such a comprehensive sampling design would be well-suited for identifying essential spawning habitat, for siting habitat- specific investigations of spawning dynamics, or for validat- ing dispersal models for early life stages of marine fish. Acknowledgments We are grateful to many individuals for assistance in this research. First, to the fishermen and processors in the halfbeak fishery, all of whom participated in this survey. T. Brown, J. Hunt, and R. Moretti provided logistical support in Marathon. R. Beaver, K. Krumm, E. Robillard, D. Snod- grass, and J. Whittington assisted in fish collection and processing. G. Gerdeman, P. Nagle, F Stengard, C. Stevens, and P. Thurman assisted with tissue processing and repro- ductive staging. C. Anderson assisted in preparing Figure 1 and GIS habitat analyses. B. Mahmoudi, R. Taylor, M. Zimmermann, and two anonymous reviewers provided constructive comments. Editorial assistance was provided by J. Leiby and J. Quinn. This research was funded in part by a grant from the National Oceanic and Atmospheric Administration (NOAA) to the Florida Fish and Wildlife Conservation Commission (Saltonstall-Kennedy Program, NOAA award no. NA77FD0069). Literature cited Berkeley, S. A. and E. D. Houde. 1978. Biology of two exploited species of halfbeaks, Hemir- amphus brasiliensis and H. balao from southeast Florida. Bull. Mar. Sci. 28:624-644. Berkeley, S. A., E. D. Houde. and F. Williams. 1975. Fishery and biology of ballyhoo on the southeast Flor- ida coast. In Sea Grant Special Report 4, 1-15 p. Univ. Miami Sea Grant Program, Coral Gables, FL. Brown-Peterson, N., P. Thomas, and C. R. Arnold. 1988. Reproductive biology of the spotted seatrout.CvMosf 10?! nebiihsiis, in south Texas. Fish. Bull. 86:.373-88. Collette, B. B. 1965. Hemiramphidae (Pisces, Synentognathi) from tropical west Afinca. Atlantidc Report 8:217-235. DeMartini, E. E., and R. K. Fountain. 1981. Ovarian cycling frequency and batch fecundity in the queenfish, Seriphus politiis: attributes representative of serial spawning fish. Fish. Bull. 79:547-60. Graham, D. H. 1939. Breeding habits of the fishes of Otago Harbour and adjacent seas. Trans. Proc. Royal Soc. New Zealand 69:361 - 372. McBride et al : Spawning grounds of Henvramphus biasiliensis and W. balao 589 Hunter, J. R. and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy {Engraulis mordax) (R. Lasker, ed.), 79-94 p. U.S. Dep. Commer.. NOAA Tech. Rep. NMFS 36. Limouzy-Paris, C. B., M. F. McGowan, W. J. Richards, J. P. Umaran, and S. S. Cha. 1994. Diversity of fish larvae in the Florida Keys: results from SEFCAR. Bull. Mar. Sci. 54:857-870. Ling, J. K. 1958. The sea garfish, Reporbamphiis melanochir (Cuvier & Valenciennes) (Hemiramphidae), in South Australia: breeding, age determination, and growth rate. Aust. J. Mar. Fresh. Res. 9:60-110. McBride, R. S. 2001. Landings, value, and fishing effort for halfbeaks, Hemiramphus spp., in the south Florida lampara net fishery. Proc. Gulf Carib. Fish. Inst. 52nd Ann. Meeting, Key West, FL. McBride, R. S., L. Foushee, and B. Mahmoudi. 1996. Florida's halfbeak, Hemiramphus spp., bait fishery. Mar Fish. Rev 58:29-38. McBride. R. S., F. J. Stengard, and B. Mahmoudi. 2002. Maturation and diel reproductive periodicity of round scad (Carangidae: Decapterus punctatus). Mar. Biol. 140: 713-722. McBride, R. S., and R E. Thurman. 2003. Reproductive biology of Hemiramphus brasilieiisis and H. balao (Hemiramphidae): maturation, spawning frequency, and fecundity. Biol. Bull. 204:57-67. Noell, C. J., S. Donnellan, R. Foster, and L. Haigh. 2001. Molecular discrimination of garfish Hyporhamphus (Beloniformes) larvae in southern Austrailian waters. Mar. Biotechnol 3: 509-514. Nybakken, J. W. 1997. Marine biology: an ecological approach, 4th ed., 481 p. Addison Wesley Longman, Inc., Menlo Park, CA. Powell, A. B., D. E. Hoss, W F Hettler, D. S. Peters, and S. Wagner. 1989. Abundance and distribution of ichthyoplankton in Florida Bay and adjacent waters. Bull. Mar. Sci. 44: 35-48. Starck,W.A.,Jr. 1968. A list of fishes of Alligator Reef, Florida, with com- ments on the nature of the Florida reef fish fauna. Un- dersea Biol. 1:4-40. Talwar, P K. 1962. A contribution to the biology of the halfbeak, Hyporh- amphus georgu (Cuv. & Val.) (Hemirhamphidae) [s;c|. In- dian J. Fish. 9:168-196. 1967. Studies on the biology of Hemirhamphus [sic] mar- ginatus ( Forsskal ) ( Hemirhamphidae-Pisces). J. Mar. Biol. Assoc. India 9:61-69. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshwater Res. 41:199-222. 590 Abstract— Tope shark iGaleorhi- nus galeus) and thornback ray {Raja clavata) are the two most captured elasmobranch species by the Azorean bottom longline fishery. In order to better understand the trophic dynam- ics of these species in the Azores, the diets of thornback ray and tope shark caught in this area during 1996 and 1997 were analyzed to describe feed- ing patterns and to investigate the effect of sex, size, and depth and area of capture on diet. Thornback rays fed mainly upon fishes and reptants, but also upon polychaetes, mysids, natant crustaceans, isopods, and cephalopods. In the Azores, this species preyed more heavily upon fish compared with the predation patterns described in other areas. Differences in the diet may be due to differences in the environments (e.g. in the Azores, seamounts and oce- anic islands are the major topographic features, whereas in all other studies, continental shelves have been the major topographic feature). No differ- ences were observed in the major prey consumed between the sexes or between size classes (49-60, 61-70, 71-80, and 81-93 cm TL). Our study indicates that rays inhabiting different depths and areas (coastal or offshore banks) prey upon different resources. This appears to be related to the relative abundance of prey with habitat. Tope sharks were found to prey almost exclusively upon teleost fish: small shoaling fish, mainly boarfish (Capros aper) and snipefish {Macroramphosus scolopax), were the most frequent prey. This study illus- trates that thornback rays and tope sharks are top predators in waters off the Azores. Diets of thornback ray {Raja clavata) and tope shark iGaleorhinus galeus) in the bottom longline fishery of the Azores, northeastern Atlantic Telmo Morato Encarnacion Sola Maria P. Gros Gui Menezes Departamento de Oceanografia e Pescas Universidade dos A(;ores PT-9901-862 Horta, Portugal E-mail address: telmo@noles horta uac pt Manuscript approved for publication 19 February 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:590-602 (2003). The thornback ray {Raja clavata L.), is a shallow water bottom-living elasmo- branch found in the Atlantic from Ice- land and Norway southwards to South Africa, including Madeira and Azores islands. This species is also found in the Mediterranean, western Black Sea, and southwestern Indian Ocean (Stehmann and Burkel, 1984). The thornback ray is commercially exploited in several countries. In the Azores it is a bycatch of the bottom longline fishery directed toward demersal and deepwater teleost species. Food and feeding habits of the thornback ray have been intensively studied since the end of the 19'^^ century (e.g. Day, 1880-84) and more recently (e.g. Smale and Cowley, 1992; Ellis et al., 1996; Daan et al.'l. However, only two studies have been conducted on the thornback ray off Portuguese conti- nental waters (Marques and Re, 1978; Cunha et al., 1986), and none exists for populations inhabiting waters around the oceanic islands or seamounts in the northeastern Atlantic. The tope shark {Galeorhinus galeus (L.)), is a cosmopolitan species that can be found from about 70°N to about 55"S. Distribution of this species in- cludes the Atlantic, Pacific and Indian Oceans (Compagno, 1984). Tope shark is also commercially exploited by sev- eral countries around the world, includ- ing the Azores, where it is a bycatch of the bottom longline fishery. Clompagno (1984) and Olsen (1984) reviewed the biology of this shark; however, there have been relatively few studies on their feeding habits. The diet of tope shark was described by Ford (1921) for individuals landed at Plymouth U.K., by Olsen ( 1954 ) in southeastern Australia, and by Ellis et al. (1996) in the north- eastern Atlantic Ocean. Elasmobranchs are among the top predators in marine environments ( Ellis et al., 1996); thus they affect the popu- lations of both fish and invertebrates at lower trophic levels. However, feed- ing studies of elasmobranches in the Azores have been limited to the blue shark (Prionace glaiica) (Clarke et al., 1996). Tope shark and thornback ray are the two most abundant elasmo- branch species landed by the Azorean bottom longline fishery. Information on the feeding habits of these two species contributes to a better understanding of trophic dynamics and food webs — in- formation which is needed as fisheries scientists advance ecosystem principles to fisheries management (Pauly et al., 2000; Pitcher, 2000; Whipple et al., 2000). The purpose of this study was to examine the diet of thornback ray and tope shark, to describe their feed- ing patterns and the effect of sex, size, depth, and location on their diet. Materials and methods Thornback rays and tope sharks were collected between March and May (spring) of 1996 and 1997 during a ' Daan, N., B. Johnson, J. R. Larsen and H. Sparholt. 1993. Analysis of the ray [Raja spec.) samples collected during the 1991 international stomach sampling project. ICESC.M. 1993/0:15,17 p. Morato et al.: Diets of Rqa davata and Galeorhinus galeus 591 39° N 3rw CORVO P FLORES c 39° N 38° N N t 28° W 27° W GRACIOSA r~ -t& • ^r^ TERCEIRA S. JORGE ^"^"^^^^A * , FAIAL ^^^V^ ^^^^^^^^ ^^ ^^k^ ^^^PICO Banco D Jo^o de Ci 9H^ 4^^^C ^ Baixo de S Mateus ■ Banco Pri^sa Alice ,6° W 25"W • ^^^^^ 38° N—^ •— than 60n m ddoth ^ ► Less than 600 m depth 30° W 28" W 37° N 39° N 38° N Mar da Pra^B f '4 24° W Figure 1 Locations of the longline sets made in the Azores during the spring of 1996 (•) and 1997 (D). study on demersal fisheries in Azorean waters (Fig. 1). Fishes were caught by longhne onboard the RV Arquipe- lago. Line setting began before sunrise (approx. 05:00 hi and hauhng started about two hours after setting. From the fish sampled, total length (TL, to the nearest cm) was measured, and sex and maturity were determined by mac- roscopic examination of gonads and claspers with maturity scales, as proposed by Stehmann (1987). Stomachs were removed and classified as either everted, regurgitated, with bait, empty, or with contents. Individuals falling in any of the first three categories, as well as those that had obviously eaten fish hooked on the longline, were excluded from further analysis. Stomachs with contents were placed in plastic bags and frozen (within about 2 h of capture) for subsequent analysis. Stomach contents, which partly consisted of a turbid suspension, were washed with water in a nylon net of approximately 0.5-mm mesh size to allow easier examination. The items were carefully separated, weighed (after removing the surface water by blotting them in tissue paper), and identified to the lowest possible taxonomic level. Individuals of each identified taxon were counted. Whenever fragments were found, the number of individuals was taken as the smallest possible number of individuals from which fragments could have originated. Precision estimates in diet studies have been advocated and used by several authors (Ferry and Cailliet, 1996; Morato et al., 1999). We used the cumulative trophic diver- sity, measured with the Shannon-Wiener index [as H'=— Z P,(log^P|), where P, is the proportion of individuals in the ;th species] to measure sample size sufficiency (Hurtubia, 1973). Cumulative numbers of randomly pooled stomachs were plotted against the cumulative trophic diversity. The asymptote of the curve indicates the minimum number of stomachs required. Frequency of occurrence (%0), percent- age number (%N), and weight (%W) for each prey type were used to describe the diet of both species (for a review see Hyslop, 1980; Cortes, 1997). Wet weight was used to de- termine the latter value. The index of relative importance 592 Fishery Bulletin 101(3) E t/5 o 3.2 2.8 2.4 2.0 1.6 1.2 0.8 0.4 0.0 [IRI={%N + %W) X %0] (Pinkas et al., 1971) and the 9(IRI (as %77?7, = 100 X IRIJ1IRI,) were calculated for each prey category and used in diet comparisons. Prey taxa occurring in less than five stomachs were grouped into higher taxonomic categories. Ontogenetic differences in the diet of thornback rays were examined by grouping fish into four size classes (49-60, 61-70, 71-80, and 81-93 cm TL). The diet of thornback rays was also analyzed by sex, depth (0-100, 101-200, 201-350 m), and area of capture (coastal areas and off- shore banks). No further analy- ses were performed for tope shark because their diet was dominated by only one prey cat- egory (see "Results" section). To determine if the most important preys were similar for different groups of rays, weighted corre- lation and concordance analyses were used (Zar, 1999). These methods were preferred to conventional rank correlation methods (e.g. Spearman) because they emphasize the high ranking given to the most important prey categories. Differences in the rank- ings of IRI values for prey categories between three or more groups (e.g. three size classes) were tested for significance with the top-down concordance method iC\= top-down concordance coefficient) (Zar, 1999). For paired groups (e.g. males and females) the top-down correlation method (r,p= top-down correlation coefficient) was used (Quade and Salama, 1992; Zar, 1999). Schoener's dietary overlap index (Schoener, 1970) (as C„= 1-0.5 1 1 P^, - P^, | , where P^- was the proportion (based on %IRI ) of food category / in the diet ofx; and P^,, was the proportion of food category i in the diet ofy) was used to measure the diet overlap between sex, size classes, depth strata, and area of capture. Cluster analysis was used to describe geographic simi- larities in the feeding habits of thornback rays. A preda- tor-prey matrix was built from published data. When more than one index was available, the following criteria were used to choose between indexes: IRI or '/< IRI, ^O, %N, %W, %Volume. The number of prey categories included was based on the quality of the description found in the pub- lished sources. Eleven different categories were obtained. A distance matrix was then calculated by using Euclidean distance, and the hierarchical form of analysis was applied (Clarke and Warwick, 1994 ). The grouping of predators was based on the "average linkage method," and a dendrogram was used as a graphic form of representation. Finally, tro- phic levels (TLd^) were estimated for each of the samples {k) by using the method proposed by Cortes ( 1999) [as TLVf^=l+ 'I^,* X TLu^t, where TLi', is the trophic level of each prey category as estimated by the author, P,^, is the proportion of prey category i in sample k]. Mean trophic levels were also Raja clavala Galeorhimts gateus 0 10 20 30 40 30 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 Number of stomachs witti contents Figure 2 Randomized cumulative trophic diversity curves for thornback ray (Raja clavata) and tope shark (Galeorhinus galeus) . estimated for groups resulting from the cluster analysis, and differences between them were tested by using one- way ANOVA (Zar, 1999). Results Thornback rays were caught at depths ranging from 10 to 350 m, but primarily (95%) shallower than 250 m. Out of 237 stomachs examined, the contents of four appeared to have been regurgitated (1.7%), seven contained bait only (2.9%), 88 were empty (37.1%), and 138 contained prey (58.2%). Rays with stomachs containing food measured from 49.0 to 93.0 cm TL. All tope sharks were caught between 10 and 150 m depth, except for one individual taken at 300 m. Out of 365 stomachs examined, 174 (47.7%) were empty, seven ( 1.9% ) contained fish hooked on the long- line and 184 stomachs (50.4% ) contained prey. Sharks with stomachs containing food ranged from 58.0 to 153.0 cm TL. The cumulative trophic diversity curves of both spe- cies appeared to reach an asymptote, suggesting that a sufficient number of stomachs were analyzed for both the thornback ray and tope shark (Fig. 2). Thornback ray The main diet components of thornback rays were fish (%IRI=81 .6) and crustaceans reptants (%IRI=17.4) (Fig. 3). Fish occurred in 84.1% of stomachs that contained food, and represented 78.0% of total prey weight and 50.2% of total prey number (Table 1). Two benthopelagic species, the snipefish {Macroramphosus scolopax (%IRI=34.01) and the boarfish (Capros aper (%IRI=26.8]), were by far the predominant fish prey items. However, some pelagic fish Morato et al,: Diets of Raja clavata and Galeorhinus galeus 593 Table 1 Values for percentage by number C/rN), weight C^W), occurrence CXO , and index of relative importance (IRI and fr [RD for prey items observed in stomachs (n = 138) of thomback rays (Raja clavata) caught off the Azores during the spring of 1996 and 1997. | Total values are given in bold font. Prey items '7.N %.w %0' IRI '/rlRI Algae 0.3 0.0 1.5 0.5 0.0 Bivalvia — Chlamys sp. 0.1 0.0 0.7 0.1 0.0 Total Cephalopoda 1.1 1.1 5.1 11.2 0.1 Octopodoidea unidentified 0.1 0.1 0.7 0.1 0.0 Scaeurgus unicirrhus 0.7 0.8 2.9 4.4 0.1 Cephalopoda unidentified 0.3 0.2 1.5 0.8 0.0 Total Polychaeta 3.4 0.8 9.4 39.5 0.8 Hirudinea 0.1 0.0 0.7 0.1 0.0 Crustacea Stomatopoda 0.1 0.0 0.7 0.1 0.0 Total Natantia 3.1 1.0 10.1 41.4 0.3 Penaeidea unidentified 1.4 0.3 2.9 4.9 0.1 Solenocera membranacea 0.1 0.1 0.7 0.1 0.0 Solenocera sp. 0.1 0.1 0.7 0.1 0.0 Pandalidae 0.3 0.1 0.7 0.3 0.0 Processa intermedia 0.1 0.0 0.7 0.1 0.0 Processa sp. 0.1 0.0 0.7 0.1 0.0 Caridea unidentified 0.1 0.0 0.7 0.1 0.0 Natantia unidentified 0.9 0.4 2.9 3.8 0.1 Total Reptantia 31.9 17.0 47.1 2303.2 17.4 Anomura unidentified 0.1 0.1 0.7 0.1 0.0 Scyllaridae Scyllarus arctus 4.0 0.8 9.4 45.1 0.9 Diogenidae 1.1 1.7 5.8 16.2 0.3 Paguridea 0.3 0.3 1.5 0.9 0.0 Galatheidae Galathea sp. 0.3 0.1 1.5 0.6 0.0 Homolidae Paromola cuvieri 0.6 0.1 2.9 2.0 0.0 Calappidae Calappa granulata 1.8 1.3 7.3 22.6 0.5 Parthenopidae Parthenope sp. 2.8 0.7 0.7 2.5 0.0 Portunidae 0.1 0.0 0.7 0.1 0.0 Total Liocarcinus spp. 14.9 8.3 16.6 385.1 5.5 Liocarcinus marmoreus 9.8 5.1 9.4 140.1 2.8 Liocarcinus corrugatus 3.8 2.7 6.5 42.3 0.8 Liocarcinus spp. 1.3 0.5 2.2 4.0 0.1 Brachyura 4.1 2.4 11.6 75.4 1.5 Reptantia unidentified 0.6 0.4 2.9 2.9 0.1 Decapoda unidentified 0.1 0.0 0.7 0.1 0.0 Total Mysidacea 6.6 0.7 3.6 26.3 0.5 Isopoda 1.6 0.3 5.1 9.7 0.2 Amphipoda-y/fef/m sp. 0.1 0.0 0.7 0.1 0.0 Crustacea unidentified 1.1 0.8 4.4 8.4 0.2 Total Pisces 50.2 78.0 84.1 10811.2 81.6 Myctophidae 0.6 0.3 2.2 2.0 0.0 Moridae Gadella maraldi 0.1 0.2 0.7 0.2 0.0 Caproidae Capros aper 13.7 24.7 34.8 1336.3 26.8 Macroramphosidae Macroramphosus scolopax 16.7 19.3 47.1 1695.6 34.0 Sparidae Pagellus spp. 1.0 5.4 4.4 28.2 0.6 Mullidae Mullus surmuletus 0.1 3.0 0.7 2.2 0.0 Pomacentridae Chromis limbata 0.1 0.2 0.7 0.2 0.0 Carangidae Trachurus picturatus 0.9 2.6 3.6 12.6 0.3 Scombridae Scorn ber Japon icus 0.4 6.0 2.2 14.1 0.3 Pisces unidentified 16.6 16.3 43.5 1431.2 28.7 Rocks 1.0 0.3 5.1 6.6 0.1 Tissue unidentified 0.4 0.8 2.2 2.6 0.1 ' Because the %0 is a nonadditive index (Cortes, 1997) for grouping fish items into higher taxonomic categories (i.e Pisces, etc), the %0 value was recalculated by considering the number of stomachs with the respective higher taxonomic category. This recalculation affects both the IRI and %IRI | values. 594 Fishery Bulletin 101(3) A 100 Thornback ray, Raja clavata (n= 138) Tope shark, Galeorhlnus galeus (n=184) Pisces CJ o B 100 80 60 40 20 0 20 40 60 80 100 Figure 3 Relative importance of prey categories in the diet of (A) tliornback ray (/?"./o clavala) and (B) tope shark (Galeorhirtiis galeus) ranked from highest IRI values. Where the areas of the boxes are equal to the IRI value [(%N+'/r W) x %0\. ':', N is the percent number, 7,W the percent weight, and %0 the frequency of occurrence of the prey category. Each tick mark of %0 represents 10%. %0 Morato et al.: Diets of Rqa davata and Galeorhinus galeus 595 Table 2 Percentage of relative importance C/t IRI ) of food categories of Raja davata by sex, total length, depth strata, and areas (coastal and offshore banks). Prey items occurring in less than five stomachs were grouped into higher taxonomic levels. The null hypothesis of not feeding upon the same most important prey categories was tested by using the top-down correlation method (being r.^ the top-down correlation coefficient) and the top-down concordance method (being Cj the top-down concordance coefficient). NS = non significant, *P<0.01. Sex Total length (cm) Depth (m) Areas M 49-60 61-70 71-80 81-93 0-100 101-200 201-350 Banks Coastal Cephalopoda 0.52 0.03 1.44 0.00 0.38 0.63 0.03 Polychaeta 0.62 1.70 0.21 0.43 0.73 6.44 0.54 Penacidea 0.34 0.62 0.72 1.32 0.18 0.00 0.19 Other Natamia 0.10 015 0.52 O.I I 0.06 0.00 0.48 Diogenidae 0.07 1.58 1.45 0.00 0.88 0.45 0.69 Scyllarus anms 1.54 0.57 1.12 0.25 2.48 0.64 0.76 Cahippa xraimlata 0.73 0.31 0.45 0.30 0.90 0.36 0.00 Liocanimis -.pp. 8.12 0.60 1.64 3.30 9.49 0.19 10.44 Other Reptantia 9.20 8.32 1.43 11.88 22.48 0.00 47.44 Mysidacea 0.68 0.50 0.18 0.62 1.02 0.00 0.00 Isopoda 0.53 0.00 0.00 0.24 0.30 0.35 0.02 Cciproscipci- 41.20 24.16 36.26 38.11 23.39 53.34 20.06 Mmnmimphosus scolopax 35.15 58.88 53.60 41.65 34.91 37.35 15.81 Pagelhis^p. 0.46 1.07 0.00 1.08 1.24 0.00 0.56 Myctophidae 0.04 0.12 0.00 0.14 0.09 0.00 0.03 Tnulumis picnmiliis 0.14 0.57 0.34 0.00 0.87 0.26 1.25 Other Pisces 0.58 0.82 0.64 0.56 0.61 0.00 1.71 0.21 0.40 0.12 0.01 0.21 0.84 1.52 1.43 0.31 1 .00 0.47 38.26 54.84 0.19 0.04 0.01 0.15 5.60 15.13 14.72 0.00 0.00 4.35 0.00 0.00 0.00 2.21 0.00 10.63 36.72 9.99 0.65 0.00 0.00 Stomachs with contents (/I) 89 49 19 47 60 11 47 Ct=0.51NS 78 13 3.48 4.23 1.41 0.08 0.00 0.91 0.10 0.00 0.52 16.79 0.00 35.56 33.50 2.27 1.16 0.00 0.00 0.06 0.57 0.29 0.12 0.53 1.21 0.64 12.30 6.33 0.00 0.31 32.53 42.91 0.43 0.01 0.40 1.36 24 =0.44NS 110 prey were also recorded in the stomachs of thornback rays: the chub mackerel, {Scomber japonicus [7fIRI=0.3]) and the blue jack mackerel {Tyachuriis picturatus [%IRI=0.3]). Some individuals also fed upon mesopelagic myctophids (%IRI<0. 1 ) and upon shallow water benthic fish such as the red striped mullet (Mullus surmuletus [%IRI<0.1] ) and the Azorean chromis (Chrornis linibata [%IRI<0.1|). Reptants occurred in 47.1% of the stomachs examined and represented 17.0% by weight and 31.9% by number of the total prey found (Fig. 3A). Swimming crabs (Liocar- cinus spp. 1%IRI=5.5]), which include both L. marmoreiis (%IRI=2.8) and L. corrugatus (%IRI=0.8), were the most important reptant prey item in the diet of thornback ray (Table 1). Other important reptants included the lesser locust lobster (Scyllarus arctus [%IRI=0.9]), the shame- faced crab (Calappa granulata |%IRI=0.5]), as well as some unidentified Diogenidae (%IRI=0.3) and brachyura (%IRI=1.5). Polychaetes (%IRI=0.8) were the third most important prey category and occurred in 9.4% of the stomachs with food (Fig. 3A). Mysids (%IRI=0.5), natants (%IRI=0.3), isopods (%IRI=0.2), and cephalopods (%IRI=0,1) also occurred in stomachs of thornback rays sampled in the Azores (Table 1). A comparison of thornback ray's diet in relation to sex, length, depth and area of capture (Table 2) suggests that C. aper and M. scolopax were by far the most important prey for all subgroups examined. The diets of both sexes were significantly correlated (r.p=0.70, P<0.01), indicating a high degree of similarity in the diets of males and females. Both sexes fed primarily upon two benthopelagic fish species (M. scolopax and C. aper) and reptants (Table 2). Schoener's diet overlap index between males and females was 0.72, also indicating a high level of similarity between diets. Significant concordance (C^=0.74,P<0.01) was displayed among thornback rays of different size classes (49-60, 61-70, 71-80 and 81-93 cm TL). Prey categories had simi- lar %IRI values for the different size classes (Table 2 ), with the exception of reptants (both Liocarcinus spp. and "other reptants"), which were more important in the diet of the two middle size classes. Schoener's index also suggested a high degree of overlap (>0.60) among all size classes (Table 3). Examination of depth-related differences was lim- ited by the small sample size of rays from deeper waters (n2oi_350n,= 13). However, the top-down concordance coef- ficient suggested that individuals captured at different depths (0-100, 101-200, and 201-350 m) do not feed upon the same most important prey categories (Ct.=0. 52, P>0.05). Reptants (both Liocarcinus spp. and "other reptants") and the fish species T. picturatus were more important in the diet of rays captured in shallow waters (0-100 m); whereas 596 Fishery Bulletin 101 (3) 100 80 60 40 20 Group I Group I ^fr^ to in oi OD CD 00 in CD CD y- f- y- Oi 2. -— X C >-- LO CD 3- ^ — - 9? 5 5 5 m m S ■c <■ S C/3 s. --~ — < LU Z !ii ^ r r 5 < 05 Group I V ^ ^;^ I Figure 4 Dendrogram of the cluster analysis (Euclidean distance, average linkage method) for geographic patterns of feeding habits of Raja clavata. In parentheses is given the authorship of the studies: 1 = Ellis et al .( 1996 ); 2 = Gibson and Ezzi (1987 ); 3 = Ajayi ( 1982 ); 4 = Quiniou and Andriamirado ( 1979 ); 5 = Olaso and Rodriguez-Mann ( 1995); 6 = Cunha et al. (1986); 7 = the present study; 8 = Ebert et al. (1991); 9 = Smale and Cowley (1992). polychaetes, cephalopods, penaeids, mysids, seabreams (Pa- gellus sp.), and myctophids were consumed more by rays caught in deeper waters (Table 2 ). Schoener's overlap index for individuals captured at different depth intervals (Table 3) indicated low overlap (=0.50), supporting the results of the top-down concordance coefficient analysis. Finally, the diet of rays caught in coastal areas and offshore banks were not significantly correlated (C.p=0.44, P>0.05), indicating that thomback rays feed upon differ- ent prey depending on the environment. The Diogenidae, Liocarcinufs spp., "other reptants," and "other Pisces" were more important prey for rays in coastal areas, whereas polychaetes, penaeids, cephalopods, mysids, seabreams iPagellus sp.), and myctophids were more important for rays caught at offshore banks (Table 2). However, Schoen- er's index showed a high level of overlap (0.69) between the diets of rays caught in the different locations — most likely due to the high dominance of two benthopelagic fishes in their diets (75.4% and 69.1% for coastal areas and offshore banks, respectively I. Published information on the diet of Ihornback rays is summarized in Table 4. Estimations of mean trophic levels vary from 3.1, for the smallest size class ( South Wales: <25 cm TL), to 4.2 for the Azorean thomback ray (this study; size Table 3 Schoener's diet overlap index for thornback rays iRaja cla- vata) size classes and for different depth strata. Depth (m) 101-200 201-350 Total length (cm) 61-70 71-80 81-93 0-100 0.40 0.29 49-60 0.83 0.66 0.76 201-350 0.50 61-70 0.77 0.77 71-80 0.62 classes 49-60 and 81-93 cm TL). The arbitrarily chosen cutoff in the cluster analysis was set at 60% dissimilar- ity, which divided the dendrogram into three groups with similar feeding patterns (Fig. 4). Cluster group I grouped the Azorean poi)ulations (all size classes) and had an esti- mated trophic level of'4. 14 ( ±0.09 SD). Cluster group II con- tained all other medium and large size classes (i.e. >40 cm TL), with the exception of small rays from the Canta- brian Sea, North Spain (17-49 cm TL), and one small- to Morato et al : Diets of Rqa davata and Galeorhinus galeus 597 ey cat- NAT= (1996); study; > IM lO CO (M .—1 ■* ■* ■* CO in CO CO T)i ■* ■* CD CD in CM T-H o CM r- in m -J CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO ■* Tf ^' Tj co CO CO ^- ;t^ ■ -^ t*-; o I— 1 1-H t— 1 IM CO CO CO CO CO CO ^ Tf - t- c- t~ 00 03 03 CO o ~: t. (ii §1^ :^ ■■3 S H " CO o o o O 00 00 00 CO CO 00 o o o o o CM CO CM ^ Tf ,-H in o CO 1—1 CD t^ CO o 00 CD 03 c- CO o o o o o o m CO in 00 in f— t 03 »-H CM t:~ > ;^ II .. ^ II CO E c4 in t— 1 d CM 00 CO CO 1— i in d d d d d CD I> jA d ^ ^ d Tl- 00 in '— ' 1— I CM C!3 00 CD 03 CO ■S M 00 -2 O 2 2 < H ^ c a. o o o o in o ^_, 00 f— 1 CM CM ^ CO CO 03 ^ CO CO 03 CO CO ^ T-t ■* o -T^ M 3 . CO 03 CO 1—1 •—1 CO CM 00 ^ r- 1 ■* 03 in CO CO CO CM O t> CM ^. o 00 00 resentei idacea; wing St ha et al CD d CO CO CO 00 Tf CM in l> CM CM d in 00 Tl< CO CO CO in CD 1-H CO 1> cm' '^ CO c~ ^ ^ CO CD c~ 00 CO tT r^ i> in 00 c~ t-H CO ■* t^ & ^ ^ c 1 o o o o in CM CO 00 CM o I> CM i> CM o 00 CD ■* CO ■-t o iH 03 CM g S'Scj ■* o 1— 1 Oi ^ t— ( CM CD CM o 03 P CM CO 00 Tt CO CO CM ^ CM o o in c- c CO CM CM .-H CM d in I> 03 d CO ■* cd' 'S' i-H .-H d d ^ d d "a II i II in ?— 1 CM CD "* CM in CM CO 00 CM ^ I— 1' M M *J to JI! r*^ ■'-' • ' o o o o o O O O o O o o ^ (-J 5-) o o o O o o o o in CJ3 1-H H o o o o O o o o o o o O o o o o o o o o o o t^ CO t^ M o CD d d d d d d d d d d d d d d d d d d d d d CM iri ra — O ^ - CM T-H ei: t; j= *^ c w o O j-5 o o o o o o o CO ^ o o CO o CO o 00 CM CM o CO CO CO 3 3 P"^ a> 00 o in o o o o o o Tf 00 o o CM o °9 o CO O o oc> CO a ,) following a; AMP = / ?rence (Ref Rodriguez ^ ci C3 d d d d d d d d CM d d d CM d i-H d d d ,_; d ^ d 1-H c 1 S a Oh o O o o o o o o o o CM O in CM o 03 o o o o o o CM O 1-H -J ^ c +j to -2 II c — o o o o o o o o o o o O o O t:^ o o Tf o in o o o d IZ) C>1 r^ o o o o o o o o o o o o o o o o CM CO CO o o o II o o - O d d d d d d d d d d d d d d d d d d d d d d d d d ^ Table 4 of trophi opoda: IS' Numbers 979): 5 = a, o o o o o o o o o o o o o .— 1 CM Tl- o 00 CO o 00 1-^ Q ti^ lO CO CO o o o o o o o o o o o o i> o in 'ti- o CO CO o o CO \ o Tt ^ d d d d d d d d d d d d d Tf d d ^ d d d d d d i. Estimation EP = Cephal yslop, 1980). riamirado (1 evel. J= K o o o o o o o o o o o o o o o o o o o o o o o o o 1-. o. o o o d CO d CD CO o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o d o tions ;a;C id(H And hicl ^ ^ o o o o o o o o o o ■* in CO in Q ^ o o o o o o o o o T* in in CT> o o o o o o o CO t~ in "* o o o o o o o o o o o t- 2 gx-^ §• pa d CO (N d d d d d d d CM CO CO O-l d d d d d d d d d d d c OJ O OJ g g II 2 o •" -2 > J o o o o o ^ o o o CO o o o o CD o rH CO ,—1 CO CO '^ o o 1-H Q. c .S S J o <>3 c^ o ^ o o 00 03 "* t> o o o o CM o O o CM -* I- Tf o o CM ogra Echi e po: Qui; 2).T a, iri ■^ CO CM d CO CO CO i-H »-i d d d d CM d d d d d d CD d d d a* .. X II o> erent g ECH = PMist 82); 4 = ley ( 19 X T3 C S O- 5= ^ ^ ^ ^ ^ E^ ^ # ^ ^ ^ z Z Z Z > > 5^ £^ g ^ >, in diff alvia; isces. ' yi ( 19 dCow s :2 o O) o> o in 05 03 03 £73 in o o o in C33 03 o o o CO CO t^ r- C 2 " II "^ S .2 ^ V CO 'f V CM V in CO 00 00 A CO in 1 1 2 o o in V o in A CO 1 03 1 .— 1 00 1 1 .—1 00 1 o in V in A Q ■- II M II -^ T}< m in in CO in i> in ■^ CO t-- CO CO ■^ > " r^ '^ ^ " 0- " e w ~ (N CO -* CO 1 " « p ^ ■^ .M .M 4^ -4J) ^ S F CO '^s 00 II CO CO CO m m CO *j *J J3 ^ ^ CO a a a M a a .4-3 *j X CO +-» CR CO o o o O o o CO CO to to a a a o o to a t3 3 3 ;d diets of th 0L= Polychai REP = Repta n and Ezzi (] etal. (1991); o u _w o to 1 3 o m O 3 o 3 o CO o 1 CO O 3 o CO O A 3 o CO a o a a o o Z Z a o to a o Z CO a o O Z o O CO 1 o O to 1 CO 01 1-1 O N < CO OJ o to 0) ;-! s < to 1 o CO at a a a CJ o CO (U o. a O a s o 1 ca 6 Categorizi egories: P( Natantia; 2 = Gibsoi 8 = Ebert c o CS o 2 c < H 2 '-3 c < z c < Z -a c CO o o CO Wales Wales Wales Wales CO 03 1 en 1 a) C3 1= a %■. u c a ;.< m CJ C 2 fa g c a fa c 'a a CO c 'a a CO "a be 3 "a M 3 X. "a cm 3 a "a be "a be 3 t South Afri South Afri South Afri en 598 Fishery Bulletin 101(3) Table 5 Values for percentage by number CSN) , weight C^fW), occurrence C/rO), and index of relative importance (IRI and '2 IRI) for prey | items observed in stomachs of tope shark (;i= 184), Galeor hinus galeus, caught off the Azores during the spring of 1996 and 1997. Number (No.) and percent occurrence 9rO) offish lenses , fish remains , and otoliths found in stomach. are also presented. Total values are given in bold font. Prey items %N %W %0' IRI %IRI Total Crustacea 1.0 1.0 3.3 6.5 0.03 Isopoda 3.6 1.1 2.7 12.8 0.3 Crustacea unidentified 1.2 0.0 1.1 1.3 0.0 Total Cephalopoda 0.8 0.2 3.3 3.2 0.02 Octopodidae 0.6 0.3 0.5 0.5 0.0 Cephalopoda unidentified 3.0 0.0 2.7 8.1 0.2 Total Pisces- 98.2 98.8 100.0 19,700.4 99.95 Stemoptychidae unidentified 0.6 0.2 0.5 0.4 0.0 Synodontidae Synodus sp. 0.6 11.5 0.5 6.5 0.2 Trichiuridae Lepidopus caudatus 0.6 0.0 0.5 0.4 0.0 Macrouridae unidentified 0.6 0.0 0.5 0.3 0.0 Phycidae Phycis phycis 1.2 0.0 1,1 1.4 0.0 Caproidae Capros aper 65.0 25.6 38.6 3494.6 93.2 Macroramphosidae Macroramphosus scolopax 11.2 2.7 8.2 113.5 3.0 Carangidae Trachurus picturatus 2.4 7.6 2.2 21.6 0.6 Total Sparidae 6.5 32.0 4.4 169.7 4.5 Pagellus acarne 2.4 5.8 1.6 13.3 0.4 Pagellus bogaraveo 2.4 14.4 1.6 27.3 0.7 Pagellus spp. 1.2 11.7 1.1 14.1 0.4 Pagrus pagriis 0.6 0.1 0.5 0.4 0.0 Sparidae unidentified 0.6 0.5 0.5 0.6 0.0 Scombridae Scomber Japonicus 2.4 18.4 1.6 33.8 0.9 No. of %o Pairs of fish lenses 493 103 Otoliths unidentified 118 75 Fish remains 3 2 ' Because the ^0 is a nonadditive index (Cortes, 1997), when grouping fish items into higher taxonomic categories (i.e. Pisces, etc) the %0 val ie was recal- culated considering the number of stomachs with the respective higher taxonomic category. This recalculation will affect both the IRI and '7fIRI values. ■-' Including unidentified fish, pairs of lenses. otoliths, and fish remains. medium-size class of South Wales (35-45 cm TL ). Cluster group III grouped small rays from several geographic re- gions, from South Africa (which also includes some large individuals) to NE Atlantic. Estimates of trophic levels were 3.46 (±0.84 SD) for the rays of the cluster group II (i,e, medium and large), and 3.35 (±0.21 SD) for the rays composing cluster group III (i.e. small ). The estimated tro- phic levels for the three cluster groups were significantly different (P<0.001). Tope shark The diet of tope shark consisted almost exclusively offish (%IRI=99.95), along with a few crustaceans (%IRI=0.03) and ccphalopods 17r IR1=0.02) (Fig. 3B). Recognizable prey from 14 different taxa were identified (Table 5). The boarfish (C. aper) was the most important prey item (%IRI=93.2), accounting for 65.0% of food by number (%N), 25.6% by weight (%W), and occurred in 38.6% of stomachs that contained food (%0). The second most important prey item was the snipefish (M. scolopax (%IRI=3.01), which represented 11.2% of food by number and 2.7% by weight. Some commercially important fish species were also found in the stomachs of tope shark; sparids ('7,IRI=4.5, which included Pagellus acarne, P. bogaraveo. and Pagrus pagrus), the chub mackerel iS. japonicus |%IRI=0.9| ), and the blue jack mackerel (T picturatus [%IRI=0.61l. These species were more important by weight than by number or occur- rence. The stomachs of tope sharks also contained 493 pairs of eye lens and fish that were heavily digested, as well as unidentifiable otoliths. Discussion In general, the percentage of empty stomachs for thornback rays and tope sharks was relatively high compared to the percentage from literature reports. The percentage of empty Morato et al.: Diets of Raia davata and Galeorhinus galeus 599 stomachs for tope shark was 47.7% — much higher than the 4.3% obsen'ed by ElHs et al. ( 1996). The percentage of empty thornback ray stomachs was high (37.1%) when compared to values reported for the North Sea (9%, Daan et al.'; and 3.7%, Ellis et al., 1996), Carmarthen Bay, South Wales (4.5%, Ajayi, 1982), west coast of Southern Africa (4.5%, Ebert et al., 1991; and 2.6%, Smale and Cowley, 1992) and the Portuguese mainland coast (2.5%, Cunha et al., 1986). We attribute the high percentage of empty stomachs found in our study to the use of longlines to catch the fish in the Azores (trawls were used in the other studies ). Longlining is a passive fishing method, which suggests that fish that feed to satiation have a reduced response to bait odor ( Lokkeborg et al., 1995), meaning that fish with full stomachs tend not to eat the bait and be caught. Thus, only those fish with empty stomachs or partial stomach fullness were caught. Thornback rays captured by longline in the Azores dur- ing the spring of 1996 and 1997 fed upon a wide variety of organisms. Fishes (81.6 %IRI) and reptants (17.4 %IRI) dominated the diet, which also consisted of polychaetes, mysids. natants, isopods, and cephalopods. In general, thornback rays in the Azores preyed more heavily upon fish in comparison with the predation patterns described in other studies. Ajayi et al. (1982) reported a predomi- nance of crustaceans (83%W) for all size classes and a low importance offish (11.6%W) in the diet of thornback rays in Carmarthen Bay, Bristol Channel. They also reported amphipods, polychaetes, and some natants as food items. Using the points method of Hyslop (1980), Ellis et al. (1996) reported that thornback rays from the North Sea fed primarily on crustaceans (78.9% ) compared to mollusks (10.2%) and fish (7.3%). Several others have also reported a dominance of crustaceans and low importance of fish in the diet of thornback ray (Fitzmaurice, 1974; Marques and Re, 1978; Quiniou and Andriamirado, 1979; Cunha et al., 1986; Gibson and Ezzi, 1987; Smale and Cowley, 1992; Olaso and Rodriguez-Marin, 1995; Daan et al.'; Ebeling^). Polychaetes (Holden and Tucker, 1974; Marques and Re, 1978), bivalves (Quiniou and Andriamirado, 1979), holo- thurians (Ebeling-), and cephalopods (Holden and Tucker, 1974; Marques and Re, 1978; Smale and Cowley, 1992; Olaso and Rodriguez-Marin, 1995) that were considered important prey items in the other studies mentioned were not recorded or were insignificant in our samples. Differences in diet composition of several predators may reflect the geographic peculiarities in fauna composition (e.g. Smale and Cowley 1992), but when comparing diets based on higher taxonomic levels (such as fish, reptants, and natants categories), such geographic differences should not be so obvious. Our geographic analysis (see Fig. 4) distinguished three major groups: I) the Azorean individuals; II ) other large individuals; and III ) other small individuals. Further, the estimated mean trophic levels for these three major groups were significantly different: 4.14 (±0.09 SD) for the Azores; 3.46 (±0.84 SD) for other large rays; and 3.35 (±0.21 SD) for smaller rays. The higher 2 Ebeling, E. 1988. A brief survey of the feeding preferences of Raja davata in Red Wharf Bay in the Irish Sea. ICES CM. 1988/G:58, 5 p. trophic level for the Azores is a result of a higher degree of piscivory in this region and an increased consumption of decapods and fish by larger rays, compared with small rays. Notwithstanding the difference in sampling methods (longline vs. trawl caught), it appears that the Azores can be considered a separate group. In other studies, predator size played the major role in controlling feeding patterns. The diet of the thornback ray in the Azores consists of a greater proportion of fish than in any other area and may reveal differences in the function of different environ- ments, because seamounts and oceanic islands are the ma- jor topographic feature of the Azores region and the other studies were conducted on continental shelves. The general function of oceanic seamount environments is still not completely understood but they are characterized by sub- stantial enhancement of primary production due to topo- graphic effects on local hydrographic conditions (Genin and Boehlert, 1985). However, evidence for enhanced primary production leading to concentrations offish over seamounts is sparse (Rogers, 1994). Additionally, the availability and relative abundance of the two most important fish prey items found in our work (the benthopelagic species C. aper and M. scolopax ) vary considerably both seasonally (Grana- deiro et al., 1998) and annually. Therefore, the high degree of piscivory in the Azores may result from environmental features and exceptional fish prey availability during the sampled years or seasons. Thornback rays also fed on pelagic fish, as indicated by the presence of chub mackerel and jack mackerel in stom- achs— a finding that confirms previous suggestions (see Daan et al. ' ; Ebeling- ) that thornback rays are active preda- tors and able to feed semipelagically. The most important reptants in the diet, Liocarcinus spp., were also reported as the main prey item for thornback rays by Ellis et al. ( 1996). The level of importance of isopods and amphipods, mysids, cephalopods, and polychaetes in the diet of thornback rays in the Azores was similar to values reported by other au- thors (Ellis et al., 1996; Daan et al.'; Ebeling^). Differences in the dentition of females and males were reported by Quiniou and Andriamirado ( 1979) but we and Smale and Cowley (1992) observed no differences in the major prey consumed between sexes. Therefore, sexual di- morphism in dentition does not appear to be manifested in dietary preferences between sexes, as was initially expected. Several studies have demonstrated differences in preda- tion patterns for rays of different size classes — primarily a decrease in importance of crustaceans and an increase offish with size (e.g. Smale and Cowley, 1992; Elhs et al., 1996; Daan et al.'; Ebeling^). Some authors attribute these differences to the ability of large predators to prey upon larger prey (Smale and Cowley, 1992); others suggest the difference is due to a pronounced shift from a benthic to a benthopelagic feeding behavior (Skjaeraasen and Bergstad, 2000; Ebeling^) or the reverse (Quiniou and Andriamirado, 1979). We found no significant size-related differences in diet. Quiniou and Andriamirado (1979) reported shifts in diet at a size of 30 to 40 cm TL but we could not verify these conclusions because our sample included only rays larger than 49 cm. 600 Fishery Bulletin 101(3) There have been few data indicating dietary differences between thornback rays collected at different depths. Smale and Cowley ( 1992) reported that bottom type used by rays varies with depth and predicted that the prey spec- trum would thus also vary, but no depth-related analyses of diet composition were preformed in their study. Despite similarities in size (i.e. no differences in the mean size by depth strata; Menezes-*), we found that rays inhabiting dif- ferent depths prey upon different resources. The decreasing consumption of Liocarcinus spp., "other reptants," and T. picturatus, and the increasing consumption of penaeids, seabreams, and myctophids with depth of capture of rays, appears to be in general agreement with the relative abun- dance of prey with depth. Therefore, such depth-related variations in diet may simply reflect differences in prey availability. It is not clear, however, why Scyllarus arctus, a species with a known depth distribution of 4 to 50 meters (e.g. Alvarez, 1968; Castellon and Abello, 1983), appears in stomachs of thornback rays caught between 201 and 350 meters (see Table 2). There is no evidence of vertical mi- grations of thornback ray associated with feeding activity; therefore this prey was likely eaten at deep water. Thus, the depth distribution range of S. arctus in the Azores may be significantly greater than what was previously known. The only study that could corroborate this hypothesis (Fransen, 1991) reported one S. arctus caught between 420 and 700 meters depth in the Canary Islands. Our comparisons between areas (coastal and offshore banks) were unable to clearly separate the influence of depth because nearly all coastal samples were obtained from shallow waters, and offshore bank samples were collected from much deeper waters. Hence, we were incapable of determining whether the high level of poly- chaetes, penaeids, cephalopods, mysids, seabreams, and myctophids in the diet of rays caught at offshore banks reflects the availability of these prey in these areas, or in deeper waters, or both. Nevertheless, our findings indicate that coastal rays have different diets from rays taken in offshore banks. Tope sharks preyed almost exclusively upon teleosts, along with very few crustaceans and cephalopods. Previ- ous observations on the feeding behavior of this species suggested that fish and cephalopods are the main prey categories ( Elhs et al., 1996; Olsen, 1954). The diet of tope shark in the Azores consists of fewer species (mainly small shoaling fish, mainly boarfish and snipefish) compared to the diet of tope shark documented in previous studies. These two fish were also important diet components of other piscivorous species around the Azores between 1993 and 1997, namely cephalopods (Pierce et al., 1994), elas- mobranchs (Clarke et al., 1996), fishes (Clarke et al., 1995; Morato et al., 1999, 2000, 2001) and seabirds (Granadeiro et al, 1998; Ramos et al., 1998a, 1998b). The role of these two small shoaling fish in the marine food web of the Azores is not yet fully understood. The fact that these prey may exhibit strong variation in abundance, raises the question ^Menezes, G. 1995-97. Unpubl. data. Department of Ocean- ography and Fishcrie.s, University of the Azores. Cais de Santa Cruz, PT9901-862 Horta, Portugal. of how well predators can adapt to extensive changes in their availability. Stomach-content data offer a good snapshot of the feed- ing habits of fish species, but diets may vary substantially with food availability, depth, location, and season. Caution is, therefore, required when drawing conclusions about the trophic ecology of marine predators. The trophic role of thornback rays and tope sharks in the Azores could be further clarified by year round sampling and by an analysis of stable isotopes (Gu et al., 1996; Jennings et al., 1997; Pin- negar and Polunin, 2000), which could provide a less biased average estimate of predator trophic level. Acknowledgments This work is part of a more comprehensive study supported by the European Union (Design optimization and imple- mentation of demersal cruise survey in the Macaronesian Archipelagos (study contract DG XIV/94/034 and DG XIV/ 95/095). We thank Joao Gongalves, Ricardo Serrao Santos, Filipe Porteiro for help with identification of stomach con- tents, and Helena Krug for help with the identification of otoliths. Special thanks are due to the scientific staff and to the crew of the KV Arquipelago for working overtime at sea. We are also grateful to Malcolm J. Smale, Pedro Afonso, Joel Carlin, and Natacha Carvalho for reviewing the manu- script. Comments and suggestions of anonymous reviewers greatly improved the quality of the manuscript. Literature cited Ajayi, T. O. 1982. Food and feeding habits of Raja species (Batoidei) in Carmarthen Bay, Bristol Channel. J. Mar Biol. Assoc. U.K. 62:215-223. Alvarez, R. Z. 1968. Crustaceos decapodos Ibericos. Investigacion Pes- quera, tomo 32, 510 p. Imprenta Juvenil, Barcelona. Castellon, A., and P. Abello. 1983. Bathymetric distribution of some Reptantia Decapoda in the Catalan area (Spain). Rapp. Comm. Int. Mer Medit. 28(3):291-294. Clarke, M. R., D. C. Clarke, H. R. Martins, and H. M. Silva. 1995. The diet of swordfish {Xiphias gladius) in Azorian waters. Arquipelago (Life Mar. Sci.) 13A:53-69. 1996. The diet of blue shark (Pnonace glauca. L.) in Azorean waters. Arquipelago (Life Mar Sci.) 14A:41-56. Clarke, K. R., and R. M. Warwick. 1994. Change in marine communities: an approach to statis- tical analysis and interpretation, 144 p. Natural Environ- ment Research Council, UK. Compagno, L. J. V. 1984. FAO Species catalogue. Vol. 4, Sharks of the world: an annotated and illustrated catalogue of sharks species known to date. Part 2 Carcharhiniformes. FAO (Food and Agriculture Organization) Fish. Synop. 125(4) Part 2: 251-655. Cortds, 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. Morato et al.: Diets of Rqa c/avata and Galeorhinus galeus 601 1999. Standardized diet compositions and trophic levels of sharks. ICES (International Council for the Exploration of the Sea.) J. Mar. Sci. 56:707-717. Cunha, P., J. Calvario. J. C. Marques, and P. Re. 1986. Estudo comparativo dos regimes alimentares de Raja brachyura Lafont, 1873, Raja ctavata Linne, 1758, Raja montagui Fowler, 1910 e Raja naevus Miiller and Henlen, 1841 (Pisces: Rajidae) da costa Portuguesa. Arquivos do Museu Bocage Serie A III(8):137-154. Day, F. 1880-84. The fishes of Great Britain and Ireland, vol. 1, 336 p., and vol. 2, 388 p. William and Norgate, London. Ebert, D. A., P. D. Cowley, and L. J. V. Compagno. 1991. A preliminary investigation of the feeding ecology of skates (Batoidea: Rajidae) off the west coast of Southern AfHca. S. Afr J. Mar. Sci. 10:71-81. ElUs, J. R., M. G. Pawson, and S. E. Shackley 1996. The comparative feeding ecology of six species of shark and four species of ray (Elasmobranchii) in the North-East Atlantic. J. Mar. Biol. Assoc. U.K. 76:89-106. Ferry L. A., and G. M. Cailliet. 1996. Sample size and data analysis: are we characterizing and comparing diet properly? In GUTSHOP '96. Feeding ecology and nutrition in fish: symposium proceedings (D. MacKinley and K. Shearer, eds), p. 71-80. Am. Fish. Soc, San Francisco, CA. Fitzmaurice, P. 1974. Size distribution and the food of thornback rays (Raja clavata L. ) caught on rod and line on the Mayo Coast. Irish Fish. Invest, (ser. B) 11. Ford, E. 1921. A contribution to our knowledge of the life-histories of the dogfish landed at Plymouth. J. Mar Biol. Assoc. U.K. 12:468-505. Fransen, C. H. J. M. 1991. Preliminary report on Crustacea collected in the eastern part of the North Atlantic during the CANCAP and MAURITANIA expeditions of the former Rijksmuseum van Natuurlijke Historie, Leiden, 200 p. Nationaal Natuurhis- torisch Museum, Leiden. Genin, A., and G. W. Boehlert. 1985. Dynamics of temperature and chlorophyll structures above a seamount: an oceanic experiment. J. Mar. Res. 43: 907-924. Gibson, R. N., and I. A. Ezzi. 1987. Feeding relationships of a demersal fish assemblage on the west coast of Scotland. J. Fish Biol. 31:55-69. Granadeiro, J. P., L. R. Monteiro, and R. W. Furness. 1998. Diet and feeding ecology of Cory's shearwater Calonec- tris diomedea in the Azores, north-east Atlantic. Mar. Ecftl . Prog. Ser. 166:267-276. Gu, B., S. L. Schelske, and M. V. Hoyer. 1996. Stable isotopes of carbon and nitrogen as indicators of diet and trophic structure of the fish community in a shal- low hypereutrophic lake. J. Fish Biol. 49:1233-1243. Holden, M. J., and R. N. Tucker. 1974. The food of Raja clavata Linnaeus 1758, Raja mon- tagui Fowler 1910. Raja naevus Muller and Henel 1841, and Raja brachyura Lafont 1873 in British waters. J. Cons. Int. Explon Mer 35(2):189-193. Hurtubia, J. 1973. Trophic diversity measurement in sympatric preda- tory species. Ecology 19:36-58. Hyslop, E. J. 1980. Stomach contents analysis — a review of methods and their applications. J. Fish Biol. 17:411-429. Jennings, S., O. Refiones, B. Morales-Nin, N. V. C. Polunin, J. Moranta, and J. Coll. 1997. Spatial variation in the '''N and "C stable isotope composition of plants, invertebrates and fishes on Mediter- ranean reefs: Implications for the study of trophic pathways. Mar. Ecol. Prog. Ser 146:109-116. Lokkeborg, S., B. L. 011a, W. H. Pearson, and M. W. Davis. 1995. Behavioural response in sablefish Anoplopoma fim- bria, to bait odour J. Fish Biol. 46:142-155. Marques, V. M., and P. Re. 1978. Regime alimentaire de quelques Rajidae des cotes Portugaises. Arquivos do Museu Bocage. 20 serie VI(34), 8 p. Morato, T, E. Sola, M. P. Gros, and G. Menezes. 1999. Diets of forkbeard iPhycis phycis) and conger eel (Conger conger) off the Azores during spring of 1996 and 1997. Arquipelago {Life Mar Sci.) 17A:51-64. 2001. Feeding habits of two congener species of seabreams, Pagellus bogaraveo and Pagellus acarne, off the Azores (nor- theastern Atlantic) during spring of 1996 and 1997. Bull. Mar. Sci. 69(3):1073-1087. Morato, T, R. S. Santos, and P. Andrade. 2000. Feeding habits, seasonal and ontogenetic diet shift of blacktail comber, Serranus atricauda (Pisces: Serranidae), from the Azores, Northeastern Atlantic. Fish. Res. 49(l):51-60. Olaso, I., and E. Rodriguez-Marin. 1995. Alimentacion de veinte especies de peces demersales pertenecientes a la division VIIIc del ICES. Otoiio 1991. Inf Tec. Inst. Esp. Oceanogr. 157, 56 p. 01sen,A. M. 1954. The biology, migration and growth rate of the school shark, Galeorhinus australis (Macleay) (Carcharhinidae) in south-eastern Australian waters. Aust. J. Mar. Freshwat. Res. 5:353-410. 1984. Synopsis of biological data on the school shark, Galeorhinus australis (Mcleay 1881). FAG Fish. Synop. 139, 42 p. Pauly, D., V. Christensen, and C. Walters. 2000. Ecopath, Ecosim, and Ecospace as tools for evaluat- ing ecosystem impact of fisheries. ICES J. Mar. Sci. 57: 697-706. Pierce, G. J., P. R. Boyle, L. C. Hastie, and M. B. Santos. 1994. Diets of squid Loligo forbesi and Loligo vulgaris in the northeast Atlantic. Fish. Res. 21:149-163. Pinkas, L., M. S. Oliphant, and I. L. K. Iverson. 1971. Food habits of albacore, bluefin tuna and bonito in California waters. Calif Fish Game 152:1-105. Pinnegar, J. K., and N. V. C. Polunin. 2000. Contribution of stable-isotope data to elucidating food webs of Mediterranean rocky littoral fishes. Oecologia 122: 399-409. Pitcher, T J. 2000. Ecosystems goals can reinvigorate fisheries manage- ment, help dispute resolution and encourage public support. Fish Fish. 1:99-103. Quade, D., and I. Salama. 1992. A survey of weighted rank correlation. In Order sta- tistics and nonparametrics: theory and applications (P. K. Sen, and I. Salama, eds.), p. 213-224. Elsevier, New York. Quiniou, L., and G. R. Andriamirado. 1979. Variations du regime alimentaire de trois especes de raies de la bale de Douarnenez (Raja montagui Fowler, 1919; Raja brachyura Lafont, 1873; Raja clavata L., 1758). Cybium 7:27-39. Ramos, J. R., E. Sola , L. R. Monteiro, and N. Ratcliflfe. 1998a. Prey delivered to roseate tern chicks in the Azores. J. Field Omithol. 69:419-429. 602 Fishery Bulletin 101(3) Ramos, J. R., E. Sola, F. M. Porteiro, and L. R. Monteiro. 1998b. Prey of yellow-legged gull, roseate tern and common tern in the Azores. Seabird 20:31-40. Rogers, A. D. 1994. The biology of seamounts. Adv. Mar. Biol. 30: 306-350. Schoener, T. W. 1970. Non-synchronous spatial overlap of lizards in patchy habitats. Ecology 51:408-418. Skjaeraasen, J. E., and O. A Bergstad. 2000. Distribution and feeding ecology of Raja radiata in the northeastern North Sea and Skagerrak (Norwegian Deep). ICES J. Mar Sci. 57:1249-1260. Smale, M. J., and P. D. Cowley 1992. The feeding ecology of skates (Batoidea: Rajidae) off the Cape south coast. South Africa. S. Afr. J. Mar. Sci. 12: 823-834. Stehmann, M. 1987. Quick and dirty tabulation of stomach contents and maturity stages for skates (Rajidae), squaloid and other ovoviviparous and viviparous species of sharks. Am. Elasmobranch Soc. Newsletter 1987(3):5-9. Stehmann, M., and D. L. Burkel. 1984. Rajidae. /n Fishes of the North-eastern Atlantic and the Mediterranean (P. J. P. Whitehead, M. BL. Bauchot, J. BC. Hureau, J. Nielse, and E. Tortonese, eds.), vol. 1, p. 163-196. UNESCO (United Nations Educational, Scien- tific, and Cultural Organization), Paris. Whipple, S. J., J. S. Link, L. P. Garrison, and M. J. FogEirty. 2000. Models of predation and fishing mortality in aquatic ecosystems. Fish Fish. 1:22-40. Zar, J. H. 1999. Biostatistical analysis, 4"' ed., 663 p. Prentice Hall International Editions, Upper Saddle River, NJ. 603 Abstract— The U.S. Marine Mammal Protection Act requires that the abun- dance of marine mammals in U.S. waters be assessed. Because this requirement had not been met for a large portion of the North Atlantic Ocean (U.S. waters south of Maryland), a ship-based, line- transect survey was conducted with a 68 m research ship between Maryland (38.00°N) and central Florida (28.00°N) from the 10-m isobath to the boundary of the U.S. Exclusive Economic Zone. The study area (573,000 km-) was sur- veyed between 8 July and 17 August 1998. Minimum abundance estimates were based on 4163 km of effort and 217 sightings of at least 13 cetacean species and other taxonomic categories. The most commonly sighted species ( number of groups) were bottlenose dolphins, Tiirsiops truncatus (38); sperm whales, Physeter macrocephalus (29); Atlantic spotted dolphins, Stenella frontalis (28); and Risso's dolphins. Grampus griseus (22). The most abundant spe- cies ( abundance; coefficient of variation ) were Atlantic spotted dolphins ( 14,438; 0.63 ); bottlenose dolphins ( 13,085; 0.40 ); pantropical spotted dolphins, S. attenu- ata (12,747; 0.56); striped dolphms, S. coeriileoalba (10,225; 0.91); and Risso's dolphins (9533; 0.50). The abundance estimate for the Clymene dolphin, S. clymene (6086; 0.93). is the first for the U.S. Atlantic Ocean. Sperm whales were the most abundant large whale (1181; 0.51). Abundances for other species or taxonomic categories ranged from 20 to 5109. There were an estimated 77,139 (0.23) cetaceans in the study area. Bottlenose dolphins and Atlantic spot- ted dolphins were encountered primar- ily in continental shelf (<200 m) and continental slope waters ( 200-2000 m ). All other species were generally sighted in oceanic waters (>200 m). The distri- bution of some species varied north to south. Striped dolphins, Clymene dol- phins, and sperm whales were sighted primarily in the northern part of the study area; whereas pantropical spot- ted dolphins were sighted primarily in the southern portion. Abundance of cetaceans in the southern U.S. North Atlantic Ocean during summer 1998 Keith D. Mullin Gregory L. Fulling Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 3209 Frederic Street Pascagoula, Mississippi 39567 E-mail (for K. D Mullin); Keith DMullin(a'noaa-gov Manuscript approved for publication 11 February 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:603-613 (2003). The U.S. Marine Mammal Protection Act (MMPA) requires that stocks of marine mammal species in U.S. waters be maintained at or above their opti- mum sustainable population (OSP) level, defined as the number of animals that will result in maximum productiv- ity. The MMPA, as amended in 1994, requires that the U.S. National Marine Fisheries Service (NMFS) determine the potential biological removal (PBR) of each stock for management pur- poses. PBR is an estimate of the maxi- mum number of animals that may be removed from a stock due to human activities (e.g. fisheries bycatch) while allowing the stock to reach or maintain its OSP. The PBR is calculated by using the estimated minimum abundance of a stock, half its maximum net productiv- ity rate (theoretical; or estimated), and a recovery factor (Barlow et al., 1995). For the U.S. Exclusive Economic Zone (EEZ) adjacent to the Atlantic coast of the continental U.S., the NMFS cur- rently defines 27 taxa of cetaceans as stocks (Waring et al., 2001). These stocks include 24 one-stock species, bottlenose dolphins (Tursiops trunca- tus) that are divided into two stocks, and one mesoplodont beaked whale stock. Abundance estimates are avail- able for most of these stocks from U.S. waters north of the Virginia-Maryland border (38.00°N). In 1998, except for three stocks, abundance estimates were not available for Atlantic ceta- cean stocks from U.S. waters south of Maryland (Waring et al., 1997). Abun- dance estimates for these three stocks were based on a small amount of effort from a 1992 winter ship survey south of Cape Hatteras (Mullin and Ford'). Other cetacean abundance estimates from U.S. waters south of Maryland are for portions of the continental shelf or continental slope (Blaylock and Hog- gard, 1994; Blaylock, 1995; CeTAP2; Fritts et al.-'). To estimate the abundance of ceta- ceans in U.S. Atlantic waters south of Maryland, a ship survey was conducted during summer 1998 and the results are reported in this study. Abundance esti- mates from this area are combined with abundance estimates from surveys of U.S. waters north of the Virginia-Mary- land border conducted by the NMFS Northeast Fisheries Science Center to obtain overall abundance estimates for western North Atlantic cetacean stocks (e.g. Waring et al., 2001). 'Mullin, K.D., and R.Ford. 1992. Report of NOAA ship Oregon II cruise 92-01 (198) (a cetacean survey of U.S. Atlantic waters south of Cape Hatteras, winter 1992). Southeast Fisheries Science Center, P.O. Drawer 1207, Pascagoula, Mississippi 39568. - CeTAP (Cetacean and Turtle Assessment Program). 1982. A characterization of marine mammals and turtles in the mid- and north-Atlantic areas of the U.S. outer continental shelf Final Report of the Cetacean and Turtle Assessment Program Bureau of Land Management, contract no. AA551-CT8-48, 450 p. U.S. Dep. Interior, Washington DC. i Fntts, T. H., A. B. Irvine, R. D. Jennings, L. A. Collum, W. Hoffman, and M. A. McGehee. 1983. Turtles, birds, and mammals in the northern Gulf of Mexico and nearby Atlantic waters. Rep. FWS/ OBS-82/65, 455 p. U.S. Fish and Wildlife Service, Office of Biological Services, Wash- ington. D.C. 604 Fishery Bulletin 101(3) 40- 35- 30- North Atlantic Ocean -70 Figure 1 Survey effort (4163 km; thin lines) in Beaufort sea state s4 in the southern U.S. Atlantic study area (outlined by thick line) during summer 1998. Blank areas indicate Beaufort sea states >4 that were not included in the survey effort. The 200-, 500-, 1000-, 2000-, and 3000-m isobaths are shown. Methods Study area and survey design The study area (573,000 km^) was North Atlantic Ocean waters between central Florida (28.00°N) and Maryland (38.00°N) from the 10-m isobath to the boundary of the U.S. EEZ, generally 371 km (200 nmi) from the nearest U.S. point of land (Fig. 1). The study area has a diverse bottom topography and includes a very narrow continental shelf (<200 m) at Cape Hatteras which broadens to form the mid-Atlantic Bight to the north and the Florida-Hat- teras Shelf to the south. Beyond the shelf, south of Cape Hatteras are found the following features: the Florida-Hat- teras Slope, the Blake Plateau (700-1000 m deep), and the Blake Escarpment. North of the Blake Plateau, the conti- nental slope from 200-2000 m deep is steep and most of the study area has water depths >2000 m. The Gulf Stream is the dominant oceanographic feature in the study area. From the south, the Gulf Stream Front generally follows the upper continental slope northward to Cape Hatteras, where it flows to the northeast. Seaward of the Gulf Stream are Sargasso Sea waters. North ol' Cape Hatteras and the Gulf Stream Front, cooler waters, which largely originate in the Labrador Sea, drift into the study area from the north and northeast. Transects covered the study area uniformly in a saw- tooth pattern from a random start at the southernmost inshore point and were surveyed from the 68-m NOAA ship Relentless (renamed Gordon Gunter in 1999) between 8 Ju- ly and 17 August 1998 from south to north, and from north to south. Transects were placed to cross the bathymetry gradient. The narrow band of U.S. waters between central Florida and Key West, Florida, were partially surveyed but were not included in the present report. Data collection Data were collected by two teams of three observers from the ship's flying bridge, located 14.5 m above the surface of the water, during daylight hours, weather permitting (i.e. no rain, Beaufort sea state <6). Observers used standard line-transect survey methods for cetaceans that were simi- lar to those used from ships in the Pacific Ocean and Gulf of Mexico (e.g.. Barlow, 1995; Hansen et al.'*). Each team had at least two members experienced in shipboard line- transect methods and in the identification of tropical and temperate cetaceans. Two observers searched for cetaceans using 25x binoculars and another observer searched using unaided eye or 7x hand-held binoculars and recorded data. These three observers constituted the "primary team." From 18 July to 17 August, a fourth observer was added to one team to act as a conditionally independent observer (CIO, see below). The area from 90° left and right of the ship's bow to the horizon was searched by the primary team. Observers changed position (including the CIO position) every 30^0 minutes, and each team alternated two-hour watches throughout daylight hours. The survey speed was usually 18 km/li but varied with sea conditions. Data were recorded on a computer interfaced with a glob- al positioning system (GPS) by a data acquisition program. Data collected for each cetacean sighting included time, position, bearing, and reticle (a measure of radial distance) of the sighting, species, group-size, behavior, bottom depth, sea surface temperature, and associated animals (e.g. sea- birds, fish). The bearing and radial distance for sightings that were close to the ship were estimated. Survey effort data were automatically recorded every two minutes and included position, heading, effort status, observer position, and environmental conditions that could affect the observ- ers' ability to sight animals (e.g. Beaufort sea state, position of the sun). Typically, if a sighting was within a 5.5-km strip on ei- ther side of the ship, the ship was diverted from the tran- sect line and approached the group so that observers could identify species and obtain group-size estimates. For each sighting, the final group-size was estimated by a consensus ■* Hansen, L. J., K. D. Mullin, T. A. Jefferson, and G. P. Scott. 1996. Visual surveys aboard ships and aircraft. In Distribution and abundance of marine mammals in the north- central and western (lulf of Mexico: final report; vol, II: technical report (R.W. Davis and G. S. Fargion. eds.), p. 5.5-132. Outer Contnu-ntal Shelf (OCS) Study MMS 96-0027. U.S. Dep. Inte- rior, Minerals Mgnit. Service, Gulf of Mexico OCS Region, New Orleans, LA. Mullln and Fulling: Abundance of cetaceans in the southern Atlantic Ocean 605 Table 1 Number of on-effort cetacean group sightings of each species or other taxonomic category during 4163 km of survey effort in the | southern U.S. Atlantic study area during summer 1998. Species are listed in categories pooled to estimate /(O) (see Table 2). The number of sightings used for line-transect and strip-transect abundance estimates are indicated for each species. flO) groupings and species Line-transect Strip-transect Large whales Fin whale [Balaenoptera physalus) 1 0 Minke whale (B. acutorostrata) 1 0 Sperm whale (Physeter macrocephalus) 29 0 Unidentified large whale 6 0 Cryptic whales Dwarf and pygmy sperm whale (Kogia spp.) 9 0 Mesoplodon spp. 4 0 Unidentified Ziphiidae 3 0 Unidentified small whale 4 0 Unidentified odontocete 12 0 Small whales and large dolphins Pilot whale (Globicephala spp.) 10 0 Bottlenose dolphin (.Tursiops truncatus) 35 3 Risso's dolphin (Grampus griseiis) 22 0 "Coastal" Atlantic spotted dolphin (Stenella frontalis) 24 3 Unidentified T. truncatus or S. frontalis 7 1 Rough-toothed dolphin (Steno bredanensis) 1 0 Small dolphins Pantropical spotted dolphin (Stenella attenuata) 6 0 Striped dolphin [Stenella coeruleoalba) 5 0 Clymene dolphin (Stenella clymene) 2 1 "Offshore" Atlantic spotted dolphin (Stenella frontalis) 1 0 Unidentified dolphins Unidentified dolphins 26 0 Stenella spp. 1 0 Total 209 8 of the primary team. Mixed-species groups were uncom- mon (five of 217 sightings) and group-size estimates were made separately for each species. Species identification Cetaceans were identified to the lowest taxonomic level possible from descriptions in field guides and scientific Uterature (e.g. Leatherwood and Reeves, 1983; Jefferson et ah, 1993; Carwardine, 1995) (Table 1). An observer's ability to make identifications depended on weather and animal behavior. The study area was potentially inhabited by short-finned pilot whales (Globicephala inacrorhynchus), which are thought to occur within the study area from about Virginia south, and long-finned pilot whales (G. melas), thought to occur from near Cape Hatteras north (Pa3me and Heinemann, 1993). Because the two species cannot be reli- ably distinguished at sea, they were recorded simply as pilot whales. Two forms of the Atlantic spotted dolphin (Stenella frontalis) were tentatively identified: the larger, more coastal form, and the smaller offshore form (Perrin et al.. 1994 ). Abundances were estimated for each form and for all Atlantic spotted dolphins combined because only one stock is currently designated for U.S. Atlantic waters. Coastal and offshore forms of bottlenose dolphins (Hersh and Duffield, 1990), which constitute the two stocks, were recorded, but most sightings could not be clearly categorized; therefore, all bottlenose dolphin sightings were pooled for one overall abundance estimate. Bottlenose and Atlantic spotted dol- phins could not always be distinguished at large distances and a separate estimate was made for animals that could not be approached and were identified as ''Tursiops or S. frontalis." Overall abundances for the genus Kogia and the genus Mesoplodon were estimated. Dwarf sperm whales (K. sima) and pygmy sperm whales (K. breviceps) were difficult to distinguish and stranding records of both species are numerous from U.S. Atlantic shores (Schmidly''). Based on 5 Schmidly, D. J. 1981. Marine mammals of the southeastern United States and the Gulf of Mexico. U.S. Dep. Interior, U.S. Fish and Wildlife Service Biological Services Program FWS/ OBS-80/41, 165 p. 606 Fishery Bulletin 101 (3) Table 2 Estimate of /lO) for each species group (see Table l).n = number of sightings used for the estimate of /JO ) before truncation (included in n is the number of sightings in parentheses that occurred while the ship was in transit in or near the study area). Truncation = | the perpendicular distance, >>, at which groups with a greater y were excluded from the analysis. ESW = effective strip width. /lO) CV Truncation ESW Species group n (/km) 1/tO)] (m) (m) Large whales 38 (1) 0.300 0.12 5500 6666 Cryptic whales 33 (1) 0.561 0.13 3000 3565 Small whales and large dolphins 121 (22) 0.498 0.10 4000 4016 Small dolphins 20 (6) 0.398 0.11 4500 5025 Unidentified dolphin 27 (0) 0.496 0.10 4000 4032 Total 239 (30) stranding records of mesoplodont whales from U.S. Atlantic shores, sightings of Mesoplodon were probably True's (M. mirus), Gervais's (M. europaeus) or Blainville's (M. densiro- stris) beaked whales (Mead, 1989). In some cases cetaceans could only be identified as large whales (>7 m long), small whales (nondolphin, <7 m), dolphins, or odontocetes. Analytical techniques For each species or taxonomic category, abundance esti- mates (N) were made with line-transect methods by using the software program DISTANCE (Colorado Coop. Fish and Wildlife Research Unit, Colorado State Univ., Fort Collins, CO) (Buckland et al., 1993) with the equation A' = A n 5/(0) 2Lg{0) where A = size of the study area; n = number of on-efTort group sightings; S = mean group-size estimate; /!0) = sighting probability density function at per- pendicular distance zero; L = total length of transect line; and giO) = probability of seeing a group on the transect line. The log-normal 95% confidence interval was computed for each abundance estimate because it was a product of estimates and tends to have a skewed distribution. The variance of N was estimated as \-dT(N) = N- var(») var(5) var[/(0)] var[i,'(0)] »r "^ S' f(Of f(Of and the coefficient of variation (CV) was estimated as CViN): Vvar(/V) N The sampling unit was the length of the transect completed on-efTort each day when the Beaufort sea state was <5. The formula used to estimate each component of the variance is given in Buckland et al. ( 1993). Var(«) was length-weighted and based on the variation in the number of on-effort group sightings between sampling units that ranged in length from 39 to 229 km/day. Estimation of M) The perpendicular distance, >>, was estimated by using bear- ing and reticle measurements. The reticle readings were converted to radial sighting distances (/?) by the method of Lerczakand Hobbs (1998), and the formula v = i?sin(6), where b = angle between the sighting and the transect line. Estimates of /iO) were made by using a hazard-rate, uniform, or half-normal model with exact perpendicular sighting distances. For each species group, outlying values ofy were truncated to improve the fit of the model (Table 2). Model selection was determined by using Akaike's informa- tion criterion (AIC; Buckland et al., 1993). The number of groups sighted of most species was insuf- ficient to obtain an estimate of /lO). Therefore, sightings of species with similar sighting characteristics (i.e. body size, group-size, surface behavior, blow visibility) were pooled to estimate /(O) for five categories (Table 1). The abundance for each species was estimated by using the pooled /1 0) and var|/10)l for its category. The varl/!0)l was assumed to be zero for the strip-transect estimates explained below. If the individual detection functions of all species within a category are indeed very similar, by pooling, the variance, CV, and confidence interval of each abundance estimate was probably underestimated because the variance of /lO) was based on an artificially high sample size. On the other hand, if the true detection functions of the species within a category are highly variable, the variance of /lO) for an individual species may be overestimated. During the study, effort was sometimes maintained while in transit to and from ports or along the border of the study area, but it usually occurred in a small range of water depths (e.g. parallel to shore) and was excluded because it could have biased abundance estimates. However, due to the small number of sightings for the survey, y from the "transit" sightings were pooled with the on-effort sightings for estimates of /lO) (Table 2). Mullin and Fulling: Abundance of cetaceans in the southern Atlantic Ocean 607 Estimation of mean group-size The group-sizes for most species tended to be related to y, because in many cases larger groups are easier to see than small groups with increasingy. In general, the arith- metic mean of group-size may be an overestimate of the true mean group-size and could lead to positively biased abundance estimates. Therefore, a regression of group-size by y was used to estimate an "expected mean group-size" (program DISTANCE). The expected mean group-size was used in the abundance estimate if it was smaller than the arithmetic mean group-size. For estimates based on a small number of sightings, the expected mean group- size was sometimes greater than the arithmetic mean. Because group-size estimates were usually made after the ship approached the group, this was assumed to be an artifact of the small sample size, and the arithmetic mean was used in these cases. Var(S) was the analytical vari- ance for mean group-sizes based on arithmetic means or was estimated as in Buckland et al. ( 1993:79) for expected mean group-sizes. Strip-transect estimates One requirement for unbiased line-transect estimates of abundance is that the cetacean group should not move in response to the ship before it is sighted (Buckland et al., 1993). If cetaceans are not sighted before they respond to the ship, in cases of attraction to the ship, /TO) and abun- dance will be overestimated. In the Gulf of Mexico, five species appear to be consistently attracted to ships to ride the bow waves (i.e. bottlenose, Atlantic spotted, spinner [S. longirostris], Clymene [S. clymene], and pantropical spotted dolphins [S. attenuata]) (Wiirsig et al., 1998). All sightings made with 25x binoculars had radial distances >665 m and were assumed to be made before these species were attracted to the ship. If sightings of these species were made at radial distances <665 m, because of the possibility of attraction, they were not included in the line-transect abundance estimate, and a separate strip-transect abun- dance estimate was made with these sightings. For each species, the width of the strip for strip-transect estimates was set at the line-transect strip width (l/2/(0)) for that species (Tables 1 and 2). This procedure yields the same result as the formula given above with flO) for the spe- cies-group category. However, f\0) for small dolphins and for small whales and large dolphins combined was not positively biased by including sightings of groups that were probably attracted to the transect line. For each spe- cies, the line- and strip-transect estimates were summed for one overall abundance estimate. Conditionally independent observer The central assumption for estimating abundance with line-transect methods is that cetacean groups on the tran- sect line are detected with certainty (i.e.giO) = 1; Buckland et al., 1993). However, this assumption is usually not met during cetacean surveys because of availability and per- ception bias (i.e.g(O) < 1) (Marsh and Sinclair, 1989). Some groups on the transect line are missed because they may not be at the surface during the time the ship is in the area and are not available to be seen, whereas other groups at the surface are missed by observers (i.e. not perceived) because of factors such as observer experience, sea state, and animal behavior, among others. An attempt was made to estimate g(0) due to perception bias with a conditionally independent observer (CIO) by using methods based on Barlow ( 1995). The CIO was used when the 4-observer team was on duty and was stationed at 25x binoculars located on a bridge-wing 2.7 m below the primary team. One individual switched teams each day; therefore all seven observers on the ship acted as the CIO at different times. The CIO searched for cetaceans near the transect line (from 30° left to 30° right of the bow) when the primary observers were on-effort. The CIO and the primary team could not see or hear each other. Whenever the primary team made a sighting, the data recorder relayed its bearing and reticle to the CIO. When the CIO made a sighting, the time, bearing and reticle were noted by the CIO, and the sighting was monitored until it was sighted by the primary team or, theoretically, passed abeam, at which time the CIO was to notify the primary team to divert the ship to identify the species and estimate group-size. Results Abundance estimates were based on 4163 km of effort in Beaufort sea states <,4 and 217 on-effort sightings of cetacean species or other taxonomic categories (Fig. 1 and Table 1). At least 13 cetacean species were sighted. The most commonly sighted species (number of sightings) were bottlenose dolphins (38), sperm whales iPhyseter macrocephalus) (29), Atlantic spotted dolphins (28), and Risso's dolphins (Grampus griseus) (22). Thirty sightings occurred during transit in Beaufort sea states s4 (861 km) and were used to estimate f(Q). Estimates of /iO) ranged from 0.300/km for large whales to 0.561/kin for cryptic whales (Table 2). Conditionally independent observer The CIO achieved 1775 km of effort (35% of effort, including transit, with Beaufort sea state s4) and sighted 21 cetacean groups. Of these, six groups ranging in size from 1 to 10 ani- mals were missed by the primary team and included three unidentified dolphin groups, two unidentified odontocete groups, and one Mesoplodon sp. Each of these sightings was observed briefly by the CIO but could not be tracked until they passed the beam of the ship; however, in each of the six cases no sightings were made by the primary team during the time frame it would have been possible to sight them. To estimate g^(O) following the analytical methods described by Barlow (1995), a separate estimate of /(O) is therefore required for CIO sightings missed by the primary team. Because there were only six of these, g(0) could not be esti- mated for any /lO) category, andi^(O) = 1 and varlgiO)] = 0 was used in each abundance estimate. 608 Fishery Bulletin 101(3) Figure 2 Locations of on-effort sightings of bottlenose dolphins (n=38), "coastal" (n=27) and "offshore" (n = l) Atlantic spotted dol- phins, pantropical spotted dolphins (n=6), striped dolphins (n=5), and Clymene dolphins (n=2). The 200-, 500-, 1000-, 2000-, and 3000-m isobaths are shown. Abundance The following were the most abundant species (abundance; coefficient of variation) observed in our study: Atlan- tic spotted dolphins (14,438; 0.63); bottlenose dolphins (13,085; 0.40); pantropical spotted dolphins (12,747; 0.56); and striped dolphins (S. coeruleoalba) (10,225; 0.91); and Risso's dolphins (9533; 0.50). Sperm whales were the most abundant large whale (1181; 0.51). Abundances for other species or taxonomic categories ranged from 20 to 6086. There were an estimated 77,139 (0.23) cetaceans in the study area. Group sizes Mean group sizes for balaenopterids, physeterids, and ziphiids were less than three animals per group. Bottle- nose dolphins, pilot whales, Risso's dolphins, and "coastal" Atlantic spotted dolphins were in groups that averaged 12-18 animals. The average group sizes of pantropical spotted, Clymene, and striped dolphins ranged from 75 to 110 individuals (Table 3). Distribution Cetaceans were distributed throughout the study area, but few sightings occurred on the eastern Blake Plateau Figure 3 Locations of on-effort sightings of Risso's dolphins (7i=22), pilot whales (n=10), rough-toothed dolphins (n = l). Mesoplo- don spp.(n=4), and unidentified beaked whales (n=3). The 200-, 500-, 1000-, 2000-, and 3000-m isobaths are shown. (Fig. 1). The distribution of species varied regionally and by water depth (Fig. 2-4, Table 4). Bottlenose dolphins and "coastal" Atlantic spotted dolphins were sighted throughout the study area but primarily in or near continental shelf waters. Pilot whales and Risso's dolphins were widely dis- tributed seaward of the continental shelf Sperm whales, unidentified large whales, "offshore" Atlantic spotted dol- phins, striped dolphins, and Clymene dolphins occurred almost exclusively in oceanic waters (>200 m) from Cape Hatteras northward. Most pantropical spotted dolphin sightings were in the southern part of the study area. Discussion Abundance Cetacean abundances for the entire study area have not been estimated previously. Based on stranding records and previous surveys within or near the study area, all the species encountered were expected to be sighted. Pre- vious abundance estimates for the western North Atlantic stocks of short-finned pilot whales and of dwarf and pygmy sperm whales, 749 (0.64) and 420 (0.60), respectively, were based on a winter 1992 ship survey in U.S. oceanic waters (>200 m) south of Cape Hatteras (Waring et al., 1997). The 1992 dwarf and pygmy sperm whale estimate is similar to our estimate (580; 0.57); although the 1998 study area was Mullin and Fulling: Abundance of cetaceans In the southern Atlantic Ocean 609 Table 3 Group size, density and abundance estimates of cetaceans n the southern U.S. Atlantic Ocean during summer 1998 (n = number of on-effort group sightings after truncation, S = mean group-size estimate, D = animals/100 km''^ N = abundance estimate, CV = coefficient of variation, LCI and UCI = = lower and upper limits of a log- normal 95% confidence interval). Species /! S CV(S) D AT Cy(N) LCI UCI Fin whale 1 2.0 — 0.007 41 1.15 6 270 Minke whale 1 1.0 — 0.004 20 1.29 3 156 Sperm whale 28 2.1 0.12 0.206 1181 0.51 445 3136 Dwarf/pygmy sperm whale 8 1.9 0.16 0.101 580 0.57 197 1708 Mesoplodon spp. 4 2.3 0.28 0.061 348 0.76 88 1376 Unidentified Ziphiidae 3 1.7 0.40 0.034 193 0.71 49 755 Pilot whale 9 16.6 0.19 0.892 5109 0.41 2302 11,341 Bottlenose dolphin line-transect 31 11.8 0.29 2.194 12,571 0.42 5600 28,222 strip-transect 3 5.0 0.12 0.090 514 0.82 118 2249 sum 2.284 13,085 0.40 6098 28,077 Risso's dolphin 18 15.4 0.26 1.664 9533 0.50 3684 24,671 Atlantic spotted dolphin "coastar line-transect 21 17.6 0.25 2.211 12,670 0.71 3471 46,244 strip-transect 3 7.3 0.39 0.132 754 0.64 211 2696 sum 2.343 "offshore" line-transect 1 37.0 — 0.177 1014 0.85 223 4618 strip-transect 0 sum (coastal and offshore) 2.520 14,438 0.63 4672 44,618 Unid. T. truncatus or S. frontalis line-transect 7 3.9 0.27 0.162 926 0.73 246 3480 strip-transect 1 1.0 — 0.006 34 0.99 6 189 sum 0.168 960 0.71 276 3334 Rough-toothed dolphin 1 8.0 — 0.048 274 1.03 47 1584 Pantropical spotted dolphin 6 77.5 0.25 2.225 12,747 0.56 4420 36,763 Striped dolphin 5 74.6 0.21 1.785 10,225 0.91 2072 50,449 Clymene dolphin line-transect 2 110.0 0.37 1.053 6031 0.94 1138 31,963 strip-transect 1 2.0 — 0.010 55 1.15 8 361 sum 1.063 6086 0.93 1293 28,652 Stenella spp. 1 15.0 — 0.089 512 1.15 77 3392 Unidentified large whale 6 1.2 0.14 0.025 143 0.58 48 426 Unidentified small whale 3 2.7 0.63 0.054 309 0.86 53 1796 Unidentified odontocete 11 1.4 0.14 0.101 580 0.36 284 1181 Unidentified dolphin 20 1.2 0.13 0.113 775 0.51 291 2066 Sum (all cetaceans) 13.462 77,139 0.23 49,649 119,850 much larger and many sightings occurred north of Cape Hatteras(Fig.4). Abundances have also been estimated for small portions of the study area, but direct comparisons to our estimates are difficult. Seasonal abundances were estimated for about 26 cetacean species or genera encountered in U.S. continental shelf and slope waters between Cape Hat- teras and Canada during aerial surveys conducted from 1978 to 1982 (CeTAP2), including at least 11 species or genera sighted during our survey. Fritts et al."' sighted 12 cetacean species during seasonal aerial surveys of the continental shelf and southern Blake Plateau off central 610 Fishery Bulletin 101 (3) North Atlantic Ocean X Fin whale O Minke whale + Sperm whale n Dwarf/pygmy sperm whale 70 Figure 4 Locations of on-effort sightings of sperm whales (;!=29), minke whales (n=l), fin whales (n=l), and dwarf and pygmy sperm whales (n=9). The 200-, 500-, 1000-, 2000-, and 3000-m isobaths are shown. Florida from 1980 to 1981; eight of these were sighted during our survey. The abundance estimate reported in the present study for the Clymene dolphin represents the first for this spe- cies in any portion of the U.S. Atlantic EEZ. However, the estimate was based on only three sightings and has a large 95% confidence interval (1293-28,652 dolphins). The Cly- mene dolphin was recognized as a valid species in 1981 and is sympatric with the spinner dolphin in the tropical Atlantic (Perrin et al., 1981). The two species have similar color patterns, and in previous studies both were possibly recognized as S. longirostris and were not distinguished (CeTAP2; Fritts et al.^). The identifications of Clymene dolphins were made by observers with experience from the Gulf of Mexico where both species are relatively com- mon (Hansen et al.*- Hansen et al.^). There is currently no stock designation for the Clymene dolphin in U.S. Atlantic waters (Waring et al., 2001). Our estimate of bottlenose dolphins is for waters >10 m in depth; however, this estimate does not include their entire water depth range in U.S. Atlantic waters south of Maryland. Bottlenose dolphins occur year-round in coastal waters <10 m in depth (the inshore boundary of the study area) and in some bays and estuaries from Cape Hatteras south. North of the Cape they have been found close to shore in waters <25 m in depth only during warm months '^ Hansen, L. J., K. D. Mullin, and C. L. Roden. 1995. Estimates of cetacean abundance in the northern Gulf of Mexico from vessel surveys, 9 p. Southeast Fisheries Science Center, P.O. Drawer 1207, Pascagoula, Mississippi 39568. Table 4 Mean water depth and sea surface temperature of cetacean species sighted in the southern U.S. Atlantic Ocean during summer | 1998 (ra=number of groups sighted on-effort; SE = = standard error). Species n Water depth (m) Sea surface temperature (°C) Mean SE Range Mean SE Range Fin whale 1 48 — — 25.1 — — Minke whale 1 3475 — — 29.5 — — Sperm whale 29 3252 122 2195-4389 29.0 0.28 22.8-29.9 Dwarf and pygmy sperm whale 9 2586 493 766-4079 29.6 0.37 26.9-30.9 Mesoplndon spp. 4 2699 735 774-4353 27.2 1.65 24.0-31.1 Unidentified Ziphiidae 3 1817 832 878-3475 29.8 0.15 29.6-30.1 Pilot whale 10 1527 387 251^280 28.5 0.73 23.2-31.5 Bottlenose dolphin 38 371 89 12-2561 29.3 0.27 23.2-31.3 Risso's dolphin 22 1300 285 44-4755 28.4 0.62 22.9-31.3 "coastal" Atlantic spotted dolphin 27 216 117 13-2524 29.1 0.26 25.1-31.3 "offshore" Atlantic spotted dolphin 1 4298 — — 27.9 — — Rough-toothed dolphin 1 4353 — — 27.3 — — Pantropical spotted dolphin 6 1498 708 598-5030 30.5 0.61 27.6-31.6 Striped dolphin 5 2736 237 2012-3475 23.9 0.37 22.9-25.1 Clymene dolphin 3 756 ,538 139-1829 27.9 0.59 26.8-28.8 Mullin and Fulling: Abundance of cetaceans in the southern Atlantic Ocean 611 and are assumed to have migrated along-shore from the south (Mead, 1975: Kenney, 1990). Aerial surveys of bottle- nose dolphins conducted in the past along the U.S. Atlantic included waters from the shore to 10 m in depth. For waters typically <75 m deep south of Cape Hatteras, the winter 1992 abundance from an aerial survey was 12,435 (0.18) (Blaylock and Hoggard, 1994). For waters <25 m deep from Cape Hatteras to northern New Jersey, the abundance from a summer 1994 aerial survey was 26,809 (0.40) (Blaylock, 1995). The frequency of bottlenose dolphin sightings during these surveys increased substantially inshore of the 10-m isobath boundary of the ship study area and, compared with the estimate from the ship, may account for the gen- erally larger aerial survey estimates even though they are for smaller study areas. There are currently two genetically distinguishable bot- tlenose dolphin stocks designated in the U.S. Atlantic: the coastal stock and the offshore stock (LeDuc and Curry, 1998; Waring et al., 2001). Using mitochondrial DNA from skin biopsy samples obtained during the summer 1998 study and other sampling efforts, Torres et al. (in press) reported no offshore form was sampled within 6 km of shore and no coastal from was sampled beyond 39 km from shore or in waters >34 m deep. Therefore an area of overlap of the two forms occurs within the 1998 study area but the fraction of each stock in our estimate (13,085: 0.40) is unknown because the number of biopsy samples between the two boundaries was very small in the Torres et al. (in press) study. However, 20 of the 38 bottlenose groups we used to estimate abundance were found in waters >50 m deep. Abundances were estimated for ten species and three other genera of cetaceans, but other species are known or expected to occur in the study area. Three of these species, right whales iEubalaena glacialis), humpback whales (Megaptera novaeangUae), and harbor porpoises (Phocoena phocoena), occur in the study area seasonally, primarily in months other than summer months, and abundances have been estimated from studies of their primary summer ranges north of the study area (e.g. Knowlton et al., 1994; Palka, 1995; Smith et al., 1999). Additional species expected in at least part of the study area include Bryde's whale (B. edeni), Cuvier's beaked whale (Ziphius cauirostris), pygmy killer whale (Feresa at- tenuata), false killer whale (Pseudorca crassidens), melon- headed whale iPeponocephala electro), killer whale (Orci- nus orca), common dolphin (Delphinus delphis), spinner dolphin, and Fraser's dolphin (Lagenodelphis hosei). Each of these species is thought to have a tropical to subtropical or broader distribution worldwide (Jefferson et al., 1993), and except for the common dolphin, an abundance estimate for each species is available for the adjacent northern Gulf of Mexico (Hansen et al.^). However, except for the spin- ner dolphin, each of these species is relatively uncommon in the northern Gulf of Mexico and was not encountered every year during four annual spring surveys with effort similar to that in our survey (Hansen et al.''). Therefore, many of these species may also be uncommon in the At- lantic study area and were simply not encountered during the 1998 survey. During a late summer 1999 ship survey of the inner half of the southern Atlantic study area that targeted bottlenose dolphins, a group of Fraser's dolphins and melon-headed whales was sighted in water 3000 m deep east of Cape Hatteras (Roden^). Some species may also inhabit the study area seasonally. During the 1992 winter ship survey south of Cape Hatteras (Mullin and Ford'), five groups of balaenopterid whales were recorded, three of which were classified as uniden- tified Bryde's or sei whales. Also during the winter 1992 survey, groups of false killer whales and Cuvier's beaked whales were sighted twice, and pygmy killer whales once. Common dolphins were sighted between Cape Hatteras and Maryland in all seasons, except summer, during the CeTAP^ study but were sighted once in this area during the late summer 1999 survey (Roden^). Common dolphins are expected to occur throughout the area surveyed in 1998 but they may not. Although there are stranding records south of Cape Hatteras (Schmidly''), there are no valid stranding or sighting records of common dolphins in the adjacent Gulf of Mexico despite extensive seasonal surveys of the northern Gulf (Jefferson, 1995; Hansen et al."*). Precision The precision of the abundance estimates was generally poor. For species or genera abundances, only the estimate for bottlenose dolphins (the most commonly sighted spe- cies), Risso's dolphins, and pilot whales had a CV s 0.50 (Table 3). The abundance estimate for the Atlantic spotted dolphin, the most abundant species, had a CV = 0.63. In cases where there is human-caused mortality in a cetacean stock, abundance estimates with a CV < 0.50 are gener- ally required to avoid incorrectly classifying a cetacean stock as "strategic" under the U.S. MMPA (i.e. annual human-caused mortality > annual PBR) less than 10*^ of the time (Wade and DeMaster, 1999). For most species, the variance in the encounter rate, varin ), accounted for more than 70% of the var(N). The distribution of most species was not uniform in the study area and precision might be improved by stratifying estimates by water depth (e.g. shelf and nonshelf) and by area (e.g. north and south of Cape Hatteras). Biases in abundance estimates The survey was designed to meet the assumptions of line- transect theory (Buckland et al., 1993). However, the abun- dance estimates are negatively biased to varying degrees because the central assumption, that cetacean groups on the transect line are detected with certainty (i.e. ^(0)=1), was not met, and data were not available to correct esti- mates for perception and availability bias. By using the CIO methods described by Barlow (1995), we attempted to estimate the fraction of groups missed on the transect line by the primary observers due to perception bias. How- ' Roden, C. L. 1999. Report of NOAA ship Oregon II cruise 99-05 (236) (a cetacean survey of U.S. Atlantic continental shelf and slope waters between New Jersey and central Florida. August-September 1999), 32 p. Southeast Fisheries Science Center, P.O. Drawer 1207, Pascagoula, Mississippi 39568. 612 Flshen/ Bulletin 101(3) ever, there were too few sightings to make ^^(0) estimates because the overall group encounter rate was lower than anticipated and the CIO was only used for 35% of the total survey effort. In the future, a CIO should be used whenever the primary team is on-effort and the CIO should search an area larger than 30° left and right of the bow. Although the data in proximity of the transect line are most critical for estimating ^(0), it is also necessary to have enough data to estimate fiO) for groups missed by the primary team. More work is needed to develop methods for estimating g(0) in relation to perception bias in the southern U.S. At- lantic. Completely independent observers cannot be used because the ship has to be diverted from the transect line to identify species and make group-size estimates. Because many groups can easily be lost once sighted, the ship must be diverted well before the group passes abeam. Barlow ( 1995) used a CIO that searched the same area as the primary team with unaided eye or 7x binoculars. The 25x binoculars were used in our study to increase the number of CIO sightings and avoid attraction bias in/tO). Previous experience in the Gulf of Mexico has indicated that many unaided-eye sight- ings would be of small groups of species that are attracted to the ship to ride the bow waves. Conversely, small groups are the most difficult for an independent observer to track with 25x binoculars because the ship is not diverted and the bearing to the group is constantly changing. Similar to Barlow's (1995) findings on perception bias, the majority of groups missed by the primary team were apparently small groups, although the group-sizes were not estimated at close range. Barlow (1995) estimated g'(O) ranging from 0.73 and 0.79 for small groups of delphinids (<21l and cryptic species (which usually occur in small groups), andg(O) = 1 for groups of >20 delphinids. In ad- dition to group-size, the magnitude of perception bias is dependent on behavior, weather (e.g. Beaufort sea state), and the observer: active groups are less likely to be missed than resting groups or species whose behavior does not produce pronounced cues (e.g. blows, splashes). Availability bias varies by species because of differences in individual dive cycles, group diving behavior, and group- sizes. Long-diving sperm whales and beaked whales will be at the surface for much less time than will many small delphinids, which have much shorter dive cycles. Diving syn- chrony among members of a group also affects availability bias; if dives are a-synchronous, the probability that at least one animal will be at the surface increases with group size. Barlow ( 1999) estimated both availability and perception bias for long-diving whales during ship surveys using 25x binoculars in a simulation study and estimated that for dwarf and pygmy sperm whales, Cuvier's beaked whales, and Mesoplodon spp., abundance estimates need to be in- creased 2 to 4 times (i.e.g(0)=0.50 to g( 0=0.25) to account for these biases. Barlow's (1999) estimates of^(O) for per- ception or availability bias (or both) are probably represen- tative of the bias in the southern Atlantic survey because similar ship survey methods were used. However, it may not be valid to apply them directly to our abundance esti- mates because cetacean diving behavior and group sizes may be temporally and geographically specific, and survey conditions and observers may vary among surveys. For the strip-transect estimates (Table 2), use of the line-transect strip width (2xl//(0)] from the 25x binocular sightings as the strip width was assumed to be conserva- tive and somewhat negatively biased. The distance from which animals will come to the ship to ride the bow is unknown, and variable, depending on factors such as the animals' previous behavior, number of opportunities for riding bow waves, and the type of ship. If the strip width was too narrow, the strip-transect estimates would overes- timate abundance. The geographical bathymetric range of the bottlenose dolphin was not covered during the survey. Because bottle- nose dolphins undertake seasonal movements in the study area, in order to estimate the entire population size, ship survey estimates need to be combined with same-season abundance estimates from coastal waters <10 m and in- shore waters (bays, sounds, and estuaries). Distribution Water-depth distributions of cetacean species were for the most part similar to those in the Gulf of Mexico (Mullin et al., 1994; Davis et al., 1998). Bottlenose dolphins and Atlantic spotted dolphins inhabit the continental shelf and shelf-edge region, whereas most other species have primarily oceanic distributions. The offshore form of the Atlantic spotted dol- phin has not been identified in the northern Gulf of Mexico. The sightings of some species were highly regional (e.g. sperm whales, striped dolphins, Clymene dolphins, pantropi- cal spotted dolphins) were probably heavily influenced by oceanographic features such as the Gulf Stream. Much more survey effort is needed in summer and other seasons before conclusions can be drawn about each species' distribution. Acknowledgments Many people made significant contributions to the success of this survey including the officers and crew of NOAA ship Relentless and C. Roden, the Field Party Chief The Relent- less was configured for marine mammal surveys through the dedicated efforts of W. Hoggard. The marine mammal observers were C. Brown, C. Burks, C. Gates, W. Hoggard, C. Hubard, T Martinez, K. Maze-Foley, M. Newcomer, S. Swartz, J. Tobias, and K. Touhey. Environmental and ich- thyoplankton data were collected by L. Bero, P. Brown, W. Fambrough, D. Fertl, A. Hamilton, A. Hohn, R. Holmes, E. Keith, E. LaBrecque, J. Litz, and J. Taylor. W. Irvin and T. Pusser were seabird observers. The survey was designed with the help of S. Swartz and the late R. Blaylock. Com- ments by C. Hubard, K. Maze-Foley, and three anonymous reviewers were very helpful in completing the manuscript. Literature cited Barlow. J. IBM. The abundance of cetaceans in California waters. Part I; Ship surveys in summer and fall of 1991. Fish. Bull. 93: 1-14. Mullin and Fulling: Abundance of cetaceans in the southern Atlantic Ocean 613 1999. Trackline detection probability for long-diving whales. In Marine mammal survey and assessment methods (G. W. Garner et al., eds. I, p. 209-221. A.A. Balkema, Rotterdam. Bariow, J., S. L. Swartz. T. C. Eagle, and P. R. Wade. 1995. U.S. marine mammal stock assessments: guidelines for preparation, background, and a summary of the 1995 assessments. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-OPR-6, 73 p. Blaylock, R. A. 1995. A pilot study to estimate abundance of the U.S. Atlantic coastal migratory bottlenose dolphin. U.S. Dep. Commer.. NOAA Tech. Memo. NMFS-SEFSC-362, 9 p. Blaylock, R. A., and W. Hoggard. 1994. Preliminary estimates of bottlenose dolphin abun- dance in the southern U.S. Atlantic and Gulf of Mexico continental shelf waters. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-356, 10 p. Buckland, S. T, D. R. Anderson, K. P. Bumham. and J. L. Laake. 1993. Distance sampling: estimating abundance of biological populations, 446 p. Chapman and Hall, London. Carwardine, M. 1995. Whales, dolphins and porpoises, 256 p. Dorling Kindersley, New York. NY. Davis, R. W, G. S. Fargion, N. May, T. D. Leming, M. Baunigartner, W E. Evans, L. J. Hansen, and K. D. Mullin. 1998. Physical habitat of cetaceans along the continental slope in the north-central and western Gulf of Mexico. Mar Mamm. Sci. 14:490-507. Hersh, S. L., and D. A. Duffield. 1990. Distinction between northwest Atlantic offshore and coastal bottlenose dolphins based on hemoglobin profile and morphometry. In The bottlenose dolphin (S. Leatherwood and R. R. Reeves (eds.l, p. 129-142. Academic Press, San Diego, CA. Jefferson, T.A. 1995. Distribution, abundance, and some aspects of the biol- ogy of cetaceans in the offshore Gulf of Mexico. Ph.D. diss., 232 p. Texas A&M University College Station, TX. Jefferson, T. A., S. Leatherwood, and M. A. Webber. 1993. Marine mammals of the world — FAO species identifi- cation guide, 320 p. FAO, Rome. Kenney, R. D. 1990. Bottlenose dolphins off the northeastern United States. In The bottlenose dolphin ( S. Leatherwood and R. R. Reeves, eds.) p. 369-386. Academic Press, San Diego, CA. Knowlton, A. R., S. D. Kraus, and R. D. Kenney. 1994. Reproduction in North Atlantic right whales [Euba- laena glacialis). Can. J. Zoo.72:1297-1305. Leatherwood, S., and R. R. Reeves. 1983. The Sierra Club handbook of whales and dolphins, 302 p. Sierra Club Books, San Francisco, CA. LeDuc, R. G., and B. E. Curry. 1998. Mitochondrial DNA sequence analysis indicates need for revision of the genus T\irsiops. Rep. Int. Whaling Comm. 47:393. Lerczak, J. A., and R. C. Hobbs. 1998. Calculating sighting distances from angular readings during shipboard, aerial, and shore-based marine mammal surveys. Mar. Mamm. Sci. 14:590-599. Marsh, H., and D. F. Sinclair 1989. Correcting for visibility bias in strip transect aerial sur- veys of aquatic fauna. J. Wildl. Manage. 53:1017-1024. Mead, J. G. 1975. A preliminary report on the former net fishery for TYfrsiops truncatus in the western North Atlantic. J. Fish. Res. Board Can. 32:1155-1162 1989. Beaked whale of the genus Mesoplodon. In Hand- book of marine mammals. Volume 4: River dolphins and the larger toothed whales ( S. H. Ridgway and R. Harrison ( eds. ), p. 349-430. Academic Press, San Diego, CA. Mullin, K. D., W. Hoggard, C. L. Roden, R. R. Lohoefener, C. M. Rogers, and B. Taggart. 1994. Cetaceans on the upper continental slope in the north-central Gulf of Mexico. Fish. Bull. 92:773-786. Palka, D. 1995. Abundance estimate of the Gulf of Maine harbor porpoise. In Biology of the phocoenids (A. Bjorge and G.P. Donovan, eds.), p. 27-50. Rep. Int. Whaling Comm. (special issue 16). Payne, P. M., and D. W. Heinemann. 1993. The distribution of pilot whales {Globicephala spp.) in the shelCshelf-edge and slope waters of the northeastern United States, 1978-1988. In Biology of northern hemi- sphere pilot whales (G. P. Donovan, C. H. Lockyer and A. R. Martin, eds.), p. 51-68. Rep. Int. Whaling Comm. (special issue 14). Perrin, W. F, D. K. Caldwell, and M. C. Caldwell. 1994. Atlantic spotted dolphin — Stenella frontalis. In Handbook of marine mammals. Volume 5: The first book of dolphins (S.H. Ridgway and R. Harrison, eds.) p. 173-190. Academic Press, San Diego, CA. Perrin, W F, E. D. Mitchell, J. G. Mead, D. K. Caldwell and P. J. H. van Bree. 1981. Stenella clyrnene, a rediscovered tropical dolphin of the Atlantic. J. Mamm. 62:583-598. Smith, T D., J. Allen, P. J. Clapman, P. S. Hammond, S. Katona, F. Larsen, J. Lien, D. Mattila, P. J. Palsboll, J. Sigurjonsson, R T. Stevick, and N. 0ien. 1999. An ocean-basin-wide mark-recapture study of the North Atlantic humpback whale (Megaptera novaeangliae). Mar. Mamm. Sci. 15:1-32. Torres, L. G., P. E. Rosel, C. D'Agrosa, and A. J. Read. In press. Improving management of overlapping bottlenose dolphin ecotvpes through spatial analysis and genetics. Mar Mamm. Sci. Wade, P. R., and D. P. DeMaster. 1999. Determining the optimum interval for abundance surveys. In Marine mammal survey and assessment methods (G. W Gamer, S. C. Amstrup, J. L. Laake, B. F. J. Manly, L. L. McDonald, and D. G. Robertson, eds.), p. 53-66. A.A. Balkema, Rotterdam. Waring, G. T, D. L. Palka, K. D. Mullin, J. W Hain, L. J. Hansen, and K. D. Bisack. 1997. U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 1996. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NE-114, 245 p. Waring, G. T., J. M. Quintal, and S. L. Swartz (eds). 2001. U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 2001. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-NE-168, 310 p. Wursig, B., S. K. Lynn, T A. Jefferson, and K. D. Mullin. 1998. Behavior of cetaceans in the northern Gulf of Mexico relative to survey ships and aircraft. Aquat. Mamm. 24: 41-50. 614 Abstract — The growth of red sea urchins (Strongylocentrotus francisca- nus) was modeled by using tag-recap- ture data from northern Cahfornia. Red sea urchins (n=211) ranging in test diameter from 7 to 131 mm were examined for changes in size over one year. We used the function J,^, = "^i + fit/,) to model growth, in which J, is the jaw size (mm) at tagging, and J,^[ is the jaw size one year later. The function RJ,), represents one of six deterministic models: logistic dose response, Gauss- ian, Tanaka, Ricker, Richards, and von Bertalanffy with 3, 3, 3, 2, 3, and 2 min- imization parameters, respectively. We found that three measures of goodness of fit ranked the models similarly, in the order given. The results from these six models indicate that red sea urchins are slow growing animals (mean of 7.2 ±1.3 years to enter the fishery). We show that poor model selection or data from a limited range of urchin sizes (or both) produces erroneous growth- parameter estimates and years-to- fishery estimates. Individual variation in growth dominated spatial variation at shallow and deep sites (F=0.246, n=199, P=0.62). We summarize the six models using a composite growth curve of jaw size, J, as a function of time, t: J =A(B - e-'^') + Dt, in which each model is distinguished by the constants A, B, C, and D. We suggest that this composite model has the flexibility of the other six models and could be broadly applied. Given the robustness of our results regarding the number of years to enter the fishery, this information could be incorporated into future fishery man- agement plans for red sea urchins in northern California. Modeling red sea urchin iStrongylocentrotus franciscanus) growth using six growth functions* Laura Rogers-Bennett California Department of Fistn and Game and University of California, Davis Bodega Marine Laboratory 2099 Westside Rd Bodega Bay, California 94923-0247 E-mail address; rogeRbennettis/ucdavis edu Donald W. Rogers Chemistry Department Long Island University Brooklyn, New York 11201 William A. Bennett John Muir Institute of the Environment University of California, Davis Davis, California 95616 Thomas A. Ebert Biology Department San Diego State University San Diego, California 92182 Manuscript approved for publication 5 February 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:614-626 Marine invertebrates are being fished at an increasing pace worldwide (Kees- ing and Hall, 1998). In California, invertebrates have a greater exvessel (wholesale) value than do fin-fish (Rogers-Bennett, 2001). Invertebrate fisheries are now experiencing seri- ous declines as have fin-fish fisheries (Dugan and Davis, 1993; Safina, 1998; Jackson et al., 2001). The once prosper- ous commercial abalone fishery in Cali- fornia which landed in excess of 2000 metric tons per year in the 1950s and 1960s was closed in 1997 (CDFG Code 5521) following the serial depletion of stocks over time (Karpov et al., 2000). Commercial divers now target red sea urchins and other invertebrates. Red sea urchin landings in California have also declined dramatically from a high of 24 metric tons (t) in 1988 to 6 t in 2002, despite management efforts (Kal- vass and Hendrix, 1997). The.se declines have generated interest in exploring the use of alternative fishery management policies, such as spatially explicit strat- egies that would protect large old sea urchins (Rogers-Bennett et al., 1995). Sea urchin growth models are criti- cal in the development of innovative management strategies to sustain the fishery because, among other things, models can be used to predict the time required for sea urchins to enter the fishery (referred to as "years to fish- ery") and the age of the broodstock. Despite the interest in examining sea urchin growth, modeling efforts have been hampered by several factors in- cluding model selection and a lack of data from a sufficiently wide range of urchin sizes. Perhaps as a consequence, estimates of red sea urchin growth have varied widely, ranging from 3 to 12 years for urchins to grow into the fishery (Kato and Schroeter, 1985; Tegner, 1989; Ebert and Russell, 1992; Smith et al., 1998). Because of the wide variation in growth estimates, the num- ber of models and methods being used, and the difficulties that these present * Contribution 2176 from the Bodega Marine Laboratory, University of Davis, Davis, CA 94923-0247. Rogers-Bennett et al,: Modeling growth of Strongylocentrotus fianciscanus 615 for management, there is a need to evaluate a number of growth models with a single data set that encompasses a large range of urchin sizes. In our study we report the results from six individual growth models applied to data from a one-year tag and recapture study of red sea urchins {Strongylocentrotus franciscanus) in northern California. We supplemented the number of juveniles in the field by stocking tagged ju- veniles. Estimates of the number of years required for ur- chins to gi-ow to minimum legal size in northern California are generated by the models. We examine the robustness of these results to changes in the parameters and the impact of a limited data set from a small range of urchin sizes on our results. We determine if there are spatial differences in growth between shallow and deep sites. Finally, we rank the models according to quality of fit, present a generic growth curve that combines the six models, and discuss the implications of our results for fishery management. Materials and methods Study sites Growth rates were determined for red sea urchins in the Salt Point (38°33'06"N, 123n9'45"W) and Caspar (39°21'49"N, 123°49'47"W) urchin harvest reserves in northern California. Commercial urchin harvesting is pro- hibited in these reserves. We examined spatial variation within Salt Point by tagging red sea urchins at one shal- low site (5 m) south of the southern border of the Gerstle Cove Reserve and at one deep site (17 m) on the leward side of a large wash-rock. In addition, laboratory-reared juvenile red sea urchins were stocked at the two sites in Salt Point. Both of these sites are relatively isolated, sur- rounded by sand and seasonally dense kelp (Nereocystis). At the Caspar Reserve, sea urchins were tagged outside a small cove with seasonally dense kelp (Nereocystis) at a single depth (7 m). Tagging Sea urchins at the study sites were tagged internally and recaptured after one year. At Salt Point, wild sea urchins were tagged with tetracycline injections in situ by using 0.5-1.2 mL of 1 g tetracycline/100 mL of seawater (cf Ebert, 1982; Ebert and Russell, 1992). Six hundred and nine red urchins were measured with vernier calipers (±1-2 mm) and tagged at Salt Point on 19 August 1992. Urchins were recaptured from the Salt Point sites on 18 September 1993 (;!=374 shallow; n=S52 deep). This data set was normalized to one year by using the factor 12/13. Our study was not a longitudinal study examining growth over many years, but rather for one year only. Juvenile urchins reared in the laboratory for one year (mean test diameter=17.6 mm) were tagged and stocked into the shallow and deep Salt Point sites. Juveniles were tagged by immersion for 24 hours in a calcein solution 125 mg/L seawater, pH adjusted to 8.0 (Wilson et al., 1987). After tagging, juveniles were transported to the Salt Point sites and released. Juveniles were stocked (120 at each of the two depths) on 31 August 1992 and harvested on 18 September 1993 with the adults (see Rogers-Bennett, 2001). Urchins at the Caspar Reserve were tagged internally with personal individual transponder (PIT) tags on 28 August 1996 and recovered 20 August 1997 (Kalvass'). PIT tags are glass coated mini-transponders with unique individual codes that can be read noninvasively by using a Destron(S) tag reader. Tags were implanted into the body cavity of the sea urchins through the peristomial mem- brane. PIT tags are too large for tagging small urchins (<40 mm). Estimates of urchin density were made within a circle (12 m in radius) at each of the two Salt Point sites at the time of harvest. Drift algae collections were made along a 2 X 10 m transect (20 m^) at each site. Gut contents were collected from a subsample of 20 urchins from each site. Gut contents were fixed in alcohol, sorted on a petri dish, and the most abundant items were recorded from 5 out of 25 lO-mm'^ grids (Harrold and Reed, 1985). We used a conservative definition of optimal foods, defining them as fleshy red or brown algae (Harrold and Reed, 1985). Sub- optimal foods included green algae, upright and encrust- ing coralline algae, detritus (animal, plant, and inorganic), plants (Phyllospadix), mud, and sand. Growth measurements Sea urchins can not be reliably aged by using rings on test ossicles (Pearse and Pearse, 1975; Ebert 1988; Gage, 1992), therefore growth increments after one year must be measured directly. For the urchins tagged with fluorescent dyes (tetracycline and calcein), growth was measured as the change in urchin jaw length (AJ =J,^j-J,) after one year (Ebert and Russell, 1993). Urchin jaws were dissected from Aristotle's lantern, excess tissue was removed with 10% sodium hypochlorite, and the jaws were measured to the nearest 0. 1 mm. Growth was measured by determining the width of the calcium deposit one year after tagging. Tags on jaws are more accurate than tags on test ossicles because ossicles move toward the oral surface during growth (Duetler, 1926), requiring matching ossicles at the time of tagging with ambitus ossicles at the time of collec- tion (Ebert, 1988). Fluorescence tagged urchins were identified when ex- posed to an ultraviolet epi-illuminator (Lite-Mite) on a dissecting scope. Growth increments were determined by using the Confocal Microscope (BioRad MRC-600, BioRad Industries, Hercules, CA) with a BHS fluorescence filter (blue wavelength) and the COMOS software package (BioRad Industries, Hercules, CA). Growth was measured from the fluorescent band (indicating size at tagging) to the esophageal edge of the jaw (final size). Growth was also recorded from a second growth zone at the labial tip of the jaw, represented by a glowing arc when present. Initial jaw size (J,) equals jaw size after one year (J,^i) minus the ' Kalvass, P. 1997. Personal commun. Calif. Dep. Fish and Game, 19160 S. Harbor Dr., Fort Bragg, CA. 95437. 616 Fishery Bulletin 101(3) Table 1 Tests for homogeneity of slopes for In {diameter) compared with In (Jau ) for shallow and deep samples of red sea urchins from Salt Point, California: SS: sum of squares df degrees of freedom; MS: mean square. Treatment (depth 1 is significant P= =0.017 when adjusted for covariate (test diameter). SS df MS F-ratio P Homogeneity of slopes Sample 0.013 1 0.013 3.643 0.058 In (jaw) 21.216 1 21.216 5760.0 0.000 Sample x \n(jaw) 0.010 1 0.010 0.723 0.101 Error 0.718 195 0.004 Adjusted means Sample 0.021 1 0.021 5.747 0.017 In (jaw) 29.333 1 29.333 7894.0 0.000 Error 0.728 196 0.004 sum of the esophageal and labial growth. Urchin jaws do not wear or erode as teeth do. Calculating test growth from changes in jaw size may yield a conservative estimate for sublegal red sea urchins (Kalvass et al., 1998). In the PIT-tagged urchins, growth was measured as the change in test diameter after one year. Juvenile urchins less than 30 mm are too small to survive PIT tag implan- tation. Large adults (>100 mm) may grow too little in one year to allow growth in test diameter to be measured. Standard errors in measures of test diameter with calipers range from 1-2 mm which may be greater than the growth increment in adults. Jaw size versus test size The relationship between jaw length and test diameter was determined from a large sample {n=384) of red sea urchins (sample independent of this study) ranging in size from newly settled individuals to large adults. From this sample we obtained an allometric equation relating jaw and test size. Using this equation we converted all the measures of growth (from fluorescent and PIT tagged urchins) into initial and final jaw size (one year later). Jaw size is a plastic trait that can vary spatially (Ebert, 1980b; Rogers-Bennett et al., 1995). Food availability has been correlated with the size of Aristotle's lantern (com- posed of ten jaws and five teeth) such that lanterns are large when food is scarce. Therefore, we examined the relationship between jaw size and test size, segregating the data from the shallow and deep Salt Point sites. To do this we used an ANCOVA (Table 1) with the natural log of test diameter as the covariate. Measurements of the jaw size at tagging from the fluorescent marks allowed for estimates of the test size at tagging using the allometric relationship (Eq. 1). As a control to test for bias in the conversion of jaw size to test diameter with Equation 1, we compared the measured test size at the time of recapture to the predicted test size at the time of recapture using the allometric relationship (Eq. 1). Results indicated that, although there was error in the pre- dicted test size from jaw size, there was no bias. Results Red urchin growth We present growth data from a total of 211 red sea urchins that were tagged internally and recaptured after one year in northern California. Recaptured urchins ranged in test diameter from 7 to 131 mm at the time of tagging. We recov- ered 161 out of 609 (26.4%) tetracycline-tagged wild urchins from the two sites at Salt Point. In addition, 38 of the 240 (15.8%) stocked juvenile urchins tagged with calcien were also recovered. It is unknown whether untagged urchins included tagged adults which were not growing and therefore not taking up the tetracycline stain. In the Caspar Reserve 12 of 53 (22.6%) PIT-tagged wild urchins were recovered. We examined spatial variation in growth and found that the change in size (AJ) was not significantly different for urchins in the shallow, as compared to the deep Salt Point sites (ANCOVA F=0.246, n = 199, P=0.62) with initial jaw size (J,) as the covariate (independent variable). Similarly, growth rates were not significantly different between juveniles recovered in shallow and deep sites (ANCOVA F=0.387, 11=38, P=0.54). Richards function parameter esti- mates (J^, K, n ) generated from the shallow site were sta- tistically identical to those for the deep site. Size-frequency distributions of urchins recovered from the shallow site were not significantly different than those at the deep site (K-S mean difference=0.162, P=0.67). Therefore, growth data from the shallow and deep sites were pooled. Urchin density at the shallow site (4.2 urchins/m'-^) was greater than at the deep site (0.75 urchins/m^). In addition, drift algae abundance was twice as great in the shallow (2.7 g/m^) as in the deep site (1.4 g/m'-) at the time of urchin harvest ( 18 September 1993 ). This resulted in less algae per urchin in the shallow site compared with the deep site (0.63 g/urchin and 1.85 g/urchin respectively) for that sampling date. Guts of urchins from the deep habitats contained more optimal food (fleshy red and brown algae) than guts of urchins in the shallow sites «=2.79, df=19, P=0.012). Gut fullness was generally uniform, roughly 50 mL/urchin. Rogers-Bennett et al,: Modeling growth of Strongylocentrotus franciscanus 617 Jaw size versus test size ANCOVA analysis indicated that the slopes of the natural log of test diameter as a function of the natural log of the jaw size are homogenous (P=0.101 ), but that the adjusted means are signifi- cantly different (P=0.017) — urchins in the deeper habitat having larger jaws (Table 1). Therefore, we constructed two allometric equations, one for urchins from the shallow Salt Point site and a second for urchins from the deep Salt Point site. However, the two equations were so similar that they generated identical test diameters for a given jaw size; therefore we pooled our data from the shallow and the deep sites. We used a larger independent data set of /i=384 from wild and cultured urchins to generate the allometric equation relating jaw size to test size. There is a strong relationship (r^=0.989, df=382) between test diameter (D) and jaw length (J) de- scribed by £> = 3.31Jii5 (1) where D = test diameter (mm); and J = jaw length (mm). Equation 1 predicts that urchins of legal size in northern California (test diameters £89 mm) have jaw lengths alT.S mm. A comparison (using the allometric relationship [Eq. 1|) of the measured test size at the time of recapture with the predicted test size revealed no bias in the conversion. Although individual values of measured and predicted test diameters are not identical, the sum of the differences between the two reveals no strong directional bias. The sum of the differences between the measured and predicted values equals 41 mm for 139 urchins, resulting in an average discrepancy of 0.30 mm per urchin. This discrepancy is smaller than the initial error in the measurement of test size (see "Materials and methods" section). The von Bertalanffy model For many organisms, annual growth rate decreases as size (age) increases. This process is frequently modeled by using the von Bertalanffy equation (von Bertalanffy, 1938) 10-1 A 8 • \. 6 ■ \^^ 4 • 2 - ' \. •\^» n U 1 1 1 1 ] 0 5 10 15 20 25 ~3 <1 10 n B 8 ■ 6 - 4 • • ^""^•v. MS _ • *^*v.« • • • •• *^\^^^ 2 ■ ^^^-,^^ ( 0 5 10 15 20 25 J Figure 1 (A) Linear function fitted to the middle portion of the urchin growth data set. (B) Linear von Bertalanffy function superimposed on the entire data set. c7=jaw length (mm); 4J = ch£uige in jaw length (mm). J,^i =J,+ J J 1 - e'^) - J,( 1-e-''') J = JA1- -Kl ), (2) (2a) which leads to a linear decrease in growth rate as a func- tion of size. We make the point here, that J,^j and J, refer to a discrete data set, whereas J is a smooth, continuous function of ^ Our data and, quite possibly, much of the data collected in similar studies, are not well represented by the von Bertalanffy equation. How is it then that the deficiencies of this well used equation have not come to light? The answer, not surprisingly, lies in the cancellation of errors within data sets that only incompletely cover the critical growth period. Our data ( Figs. IB and 2) show three features of sea ur- chin growth that are inconsistent with the von Bertalanffy model: 1) annual growth, AJ = J,^, -J, , for juveniles that is lower than anticipated from the model; 2) a maximum or plateau in the growth function, AJ = fiJf), for urchins near jaw size J( = ^ mm (test diameter 20 mm); and 3) an asymptotic approach of 4 J to zero (Figs. IB and 2), which may be ascribed to indeterminate growth for adults of all sizes or to dispersion of final adult urchin sizes (Sainsbury, 1980). There is a good deal of individual variation in growth rate as a function of J^ , which prevents unequivocal selec- 618 Fishery Bulletin 101(3) 6 - • A 5 - • • • • \ • \ • 4 - 3 ■ •• • \ 2 - \ ••• 1 - V4^ • 10 15 20 25 30 6n ■ B 5 - •■••.:■ 4 ■ / * • * • \ \ • \ • ^a 3 - 2 ■ 1 ■ n ■ • • \4^ 15 20 25 6 -| c 5 - 4 - / * • • • \ • • • 3 - / •* 2 ^ \« 1 - ^% ^ i*-r^ 10 20 25 Figure 2 Annual fjrowth as a function of jaw size for six models: lopislic dose- response, Gaussian, Tanaka, Ricker, Richards, and von Birtalanffy models. J=jaw length (mm); 4J = change in jaw length (mm). tion of one model. Nevertheless, it is clear that the von Bertalanffy model does not represent the data well over the full range of urchin sizes iJf). To investigate this point further, we divided our data set into three groups over the range of of J, . The groups are 1 Juveniles (J, < 8 mm) that do not fall on the linear descent of 4 J versus J, characteristic of von Bertalanffy growth. 2 Sublegal, actively growing adults (8 mm< J,<16 mm) that do follow von Bertalanffy kinetics. 3 Adults ( 16 mm< J, < 24 mm) that appear to grow to large J^ but only very slowly, and do not conform to the von Bertalanffy model. If the data were fitted to the von Bertalanffy equation, all three groups should give the same slope AJ IJf because three segments of the same straight line all have the same slope. Instead, group 1 gives a small positive slope, group 2 gives a negative slope that leads to unrealistic conclusions for early growth rate and time-to- fishery estimates shown in (Fig. lA), and group 3, excluding growth information from sublegal urchins, yields a plausible mean final jaw size of 22.6 mm but gives a growth rate constant that indicates very slow growth for adults and many decades for time-to-fishery. In the present study we fitted a decreasing, linear von Bertalanffy function only to the sublegal (group 2) urchins (Fig. lA) which did conform to von Bertalanffy growth. The von Bertalanffy function for the partial data set of actively growing urchins in (Fig. 1) has a slope of -0.504/yr, a AJ intercept of 8.7 mm/yr, and a Jf intercept of 17.3 mm. These results lead one to predict that final grow to 90'7f of their final size in 3.5 years and that mean final size will be less than the legal size (89 mm test diameter), which is obviously false. We also show the same function superimposed on the entire data set (Fig. IB) where discrepancies between the von Bertalanffy function and data groups 1 and 3 above are evident. For our data set (Fig. 1) and, we suggest, for urchin growth in general, the von Bertalanffy curve does not represent early growth, and a transition curve or a peaked function reflects actual growth better For our data set, the von Bertalanffy model gives an overestimate of the rate of ur- chin growth and an underestimate of the time to enter the fishery. The slopes of these three line segments give an indication how the von Bertalanffy model, despite its implausible fit to the complete data set, can give plausible growth parameters. Er- rors in fitting a von Bertalanffy curve to a data set resembling ours lie in opposite directions Rogers-Bennett et al.: Modeling growth of Strongylocentrotus franciscanus 619 for groups 1 and 3 of the growth; consequently they cancel, in whole or in part. In fact, all re- ported data sets have many more observations falling into group 3 than into group 1, which is either swamped out by group 3 or does not appear at all. This leaves gi'oups 2 and most or all of group 3 to determine the slope of the von Bertalanffy linear function. The average of these two erroneous slopes may or may not be a realistic approximation for urchin growth, depending on the number of measurements in each group. Alternative growth models Curves that rise to a maximum and then decay asymptotically are very common in the physical sciences and have been successfully modeled for more than a century (e.g. Wien, 1896). Any rising function multiplied into an exponential decay, e.g. (.v) exp(-.x:), models such a curve more or less well. The problem is not in find- ing a model but in selecting from among many possibilities. We compared several models in our study and included a Gaussian model for this data set because it has a small sum of squared residuals and because it has well- defined parameters in the arithmetic mean and standard deviation. Here the arithmetic mean merely serves to fix the position of the maxi- mum on the Jf axis and the standard deviation from f.1 gives the range, in units of J^, of actively growing animals. The model is descriptive only; it does not imply a mechanism of growth. We present results from six growth models, the logistic dose-response, Gaussian, Ricker, Tanaka, Richards, and von Bertalanffy mod- els, in order of quality of fit (Fig. 2). Each model is characterized by a different AJ = fiJf), where f(Jf) is a function of annual growth AJ versus size at tagging, J^. Equa- tions 3-8 were input as user-defined functions into a curve-fitting program (TableCurve, Jandel Scientific, now SPSS, Chicago, IL), either as f^Jf) or the equivalent J^ j^-^ - J^. In certain cases, additive parameters that make a negligible contribution to the final fit were dropped. This curve-fitting program uses the Levenburg-Marquardt procedure for find- ing the minimum of the squared sum of devia- tions. During the least-squares minimization, local minima are occasionally found and must be discarded in favor of the global minimum. Matrix inversion is performed by the Gauss- Jordan method (Carnahan et al., 1969). We present these models ranked by the fit- ting criterion of the sum of squared residuals, called "Error Sum of Squares" in the output from the TableCurve fitting program, which we have given the abbreviation RSS. Several 6 -| • D 5 - ••.• 4 - / * * * • • • 3 - / \ / * * v 2 - / \^ 1 - A / ^:<^i. U C 1 5 1 r » ♦ T* ■■•• *~n 10 15 20 25 8 - E 6 - \ ^ 4- \ * • .' V*-' • > • • • ^?. 2 - >^- r\ '• wbfcfifc^-- . U 1 } 5 10 15 20 25 6 - 5 - • F 4 - • • • \. • • • • 3 - • • x* 2 - •\^ * • 1 - •-.Mju* . 0 ■ Tmm^m* • • 1 0 5 10 15 20 25 J Figure 2 (continued) 620 Fishery Bulletin 101 (3) Table 2 A comparison of the fitt ng criteria for six red sea urchin growth functions. r2 = the coefficient of determination; RSS = the error | sum of squares; AIC = Akaike s information en terion; SBC = Schwartz-Bayesian criterion. r2 SE RSS AIC SBC No. of parameters Logistic 0.946 0.392 31.9 -393 -383 3 Gaussian 0.945 0.397 32.8 -387 -377 3 Tanaka 0.933 0.436 39.6 -347 -337 3 Ricker 0.918 0.483 48.7 -305 -299 2 Richards 0.900 0.534 59.4 -262 -251 3 von Bertalanffy 0.895 0.545 62.1 -254 -247 2 other fitting criteria are also given in Table 2. We used both the AIC information criterion, A/C = KlniRSS)- K\nK+2m, and the Schwartz-Bayesian criterion BIC = KlniRSS)- (K-m )\n{K), where K is the number of data points, and m is the number of parameters in the fitting equation (Akaike, 1979). These tests of curve-fitting quality were used to bring out any substantive difference between the 2-parameter and 3-parameter equations. The results show that differences between the 2- and 3-parameter cases are swamped out by the data, as might have been anticipated from the disparity between the number of points (iiL'=211) and the number of parameters. For the present data set, in applying either of these criteria, one is essentially seeking the smallest RSS. Individual models Logistic dose-response The logistic dose-response curve (time-to-fishery estimate: 6.6 yr) f\Ji^=a/a+{J,/br) (3) (Hastings, 1997) fits our data the best of all the models examined here. The curve fit (a=4.4, 6=12.9, c=6.8) with RSS = 31.9, is a sigmoidal transition function (TableCurve Windows, vers. 1.0, Jandel Scientific Corp., SPSS, Chicago, IL). There is a transition between a fast-growing group of sea urchins, which maintain a constant growth rate {f{Ji)=annua\ AJbAA mm/yr up to about J(=8 mm), to sea urchins growing .slowly at a rate that diminishes as Jf increases beyond 16 mm. The inflection point is at J^ ^ 13 mm. There is considerable individual variation in both data groups, but more in the fast-growing group than in the larger slow-growing group. Gaussian The Gaussian function (time-to-fishery esti- mate: 6.9 yr), although rarely if ever used in this context. f(J,) = Ae -{j,-nyil 4.6 -i- 5.7 = 10.3 mm represents urchins that are one standard devia- tion (a) above the mean of the growth curve, i. e., 16% of the growing population. Urchins with J, > 17.0 mm, are greater than 2(7 above maximal growth, i. e. 2.5% of the growing population. Therefore, a growth curve fit only to adult ur- chins with J, > 17.0 mm represents only small subset of the total growing population and is not representative of the total population. This demonstrates that data from a lim- ited size range can generate erroneous growth parameters and shorten estimates of time to enter the fishery. Variation in growth The plateau in growth rate implied by the logistic dose- response curve or the maximum in growth rate for juve- nile urchins well after settlement implied by the Gaussian curve suggests that urchin growth is not at its maximum when sea urchins first settle. It is realistic to imagine that a sea urchin will be at its maximum growth rate sometime after the first year or two. In this study we found high individual variation in sea urchin growth. Growth in juveniles was especially variable, despite the fact that the juvenile urchins that were stocked were full siblings. We found no evidence for an increase in dispersion as sea urchins grow larger. Data from many sources suggest individual variation in juvenile growth is high. Full sibling red urchins (rt=200) reared in the labora- tory under identical food, temperature, and light regimes varied in test diameter from 4 to 44 mm at one year ( Rogers- Bennett, unpubl. data). Similarly, cultured purple urchins (S. purpuratus). varied from 10 to 30 mm at one year (Pearse and Cameron, 1991) — a trend observed in other commercially cultured marine invertebrates (Beaumont, 1994) and fishes (Allendorf et al. 1987). Our data contain broad distribution in the region of the small size classes, which is consistent with high individual variation in growth iK). Varying the growth constant, K, e.g. in the Ricker model (cf Sainsbury, 1980), produces dis- persion at the smaller size classes. Our urchin growth data also show a wide array of large sizes as well. Models have been used to examine the impact of this type of individual growth variation. In the von Bertalanffy model, if final size, J^, is varied 10%, this results in a broad distribution of the largest size classes (Botsford et al., 1994). We see a broad distribution in the largest size classes in our data, with animals larger and smaller than the estimated final size J^. Many of the animals smaller than J^ could be at their final size. The biological interpretation of this broad distri- bution at the largest sizes is an open question. There may be a wide array of final sizes because of independent values of /iTand J, (cf Sainsbury, 1980) and each individual hits its own final size abruptly or at an asymptotic approach to final size (cf Beverton, 1992) also known as indeterminate growth (cf Sebens, 1987). We suggest that the composite model presented in the present study (Eq. 9) may be useful for a wide array of invertebrates and fishes especially those with a broad ar- ray of final sizes. Spatial patterns in growth In our study, we found no evidence for spatial patterns in growth. To observe spatial patterns this would have to be detectable above the background of individual variation. Sea urchins from the shallow and deep sites at Salt Point had measurable differences in gut contents, food availabil- ity, and oceanographic conditions; however these did not translate into significant differences in growth between 624 Fishery Bulletin 101(3) the depths over the year examined. Similarly, no latitudi- nal differences in red sea urchin growth were found in a large-scale growth study at 18 sites ranging from Alaska to southern California where growth varied between neighboring sites as between much as distant sites (Ebert etal., 1999). Future studies could be longitudinal and examine tem- poral patterns in sea urchin growth, for example during and after warm water El Niiio years, as has been examined for abalone in southern California (Haaker et al., 1998); however these temporal patterns too would have to be greater than individual variation to be detectable. Implications for fishery management Large old sea urchins (>125 mm test diameter) are fished in California despite fishermen receiving lower prices for these sea urchins compared with mid-size animals (Rudie^). Many of the large, old urchins have high gonadal weights (>100 g) (Carney, 1991; Rogers-Bennett et al., 1995), thereby potentially contributing more to repro- duction than smaller urchins (Tegner and Levin, 1983; Tegner, 1989; Kalvass and Hendrix, 1997). Similarly, large coral-reef fish also have the potential to contribute more to reproduction than smaller fish (Bohnsack, 1993). In fished areas, size-frequency distributions are heav- ily skewed to smaller urchins indicating that the larger size classes are absent (Kalvass and Hendrix, 1997). If the abundance and density of red sea urchins is decreased during fishing, this will decrease the chances of fertiliza- tion success significantly (Levitan et al., 1992). Sufficient numbers of large broodstock are critical because recruit- ment does not appear to be successful every year (Ebert, 1983; Pearse and Hines, 1987; Sloan et al., 1987). In addi- tion, fishing can impact recruitment success because the spines of large urchins provide canopy shelter for juveniles; therefore an Allee effect may be present (Tegner and Day- ton, 1977; Sloan et al., 1987; Rogers-Bennett et al., 1995). Size-structured red sea urchin models that include variable recruitment or an Allee effect (positive density dependence) resulted in a >50'7f decrease in estimated population size even at low fishing mortality levels (Pfister and Bradbury, 1996). Harvest experiments conducted in northern California have shown that management strategies that protect large urchins (upper size limits and harvest reserves) improve recovery and recruitment after six years compared with strategies in which large urchins are harvested (lower size limits only) (Rogers-Bennett et al., 1998). Upper size limits and reserves have been used in the management of the sea urchin fishery in Washington state (Bradbury, 2000) and are currently being considered for California's red sea urchin fishery (Taniguchi^). 2 Rudie, n. 1994. Personal commun. Catalina Offshore Prod- ucts Inc., 5202 Lovelock St., San Diego, CA 92110. ' Taniguchi, I. 2002. Personal commun. Calif. Dep. Fish and Game, 4665 Lampson Ave., Los Alamitos, CA. 90720. In conclusion, our work and that of others (Ebert and Russell, 1992, 1993; Ebert et. al., 1999) suggest that red sea urchins are slow growing, long-lived animals. Intense harvest rates may have serious consequences because red sea urchins require seven years to reach harvestable size in northern California. Declines in red sea urchin landings in northern California of more than 80% from the peak of 13,800 t in 1988 (Kalvass, 2000) demonstrate that harvest rates are high. Our growth results suggest that proposed alternative management strategies that would protect large, slow growing broodstock inside reserves or upper size limits for the fishery could be beneficial, in addition to existing regulations, for sustaining the fishery. Acknowledgments Special thanks to H. C. Fastenau, D. Canestro and the U. C. Santa Cruz research dive class (1992) for help tag- ging and measuring red sea urchins. D. Cornelius and the "Down Under" helped harvest urchins. P. Kalvass shared his growth data from PIT tagged sea urchins. F. McLafferty discussed models and "the most probable sea urchin." W. Clark, S. Wang, and F. Griffin gave access to and instruction on the confocal microscope. C. Dewees, H. Blethrow, S. Bennett, and K. Rogers all contributed. This research was funded in part by the California Department Fish and Game, the PADI Foundation, UC. Davis Natu- ral Reserve System, and the Bodega Marine Laboratory. Comments from M. Lamare and M. Mangel improved the manuscript. Literature cited Akaike, H. 1979. A Bayesian extension of the minimum AIC procedure of autoregressive model fitting. Biometrika 66:237-242. Allendorf, F. W„ N. Ryman, N., and F. M. Utter. 1987. Genetics and fishery management: past, present and fiiture. In Population genetics and fishery management, p. 1-19. Washington Sea Grant, Seattle, WA. Atkins, P. W. 1994. Physical chemistry, 5* ed., 1031 p. W.H. Freeman, New York, NY. Barrante, J. R. 1998. Applied mathematics for physical chemistry, 2"'' ed., 227 p. Prentice Hall, Upper Saddle River, NJ. Beverton, R. J. H. 1992. Patterns of reproductive strategy parameters in some marine teleost fishes. J. Fish Biol. 41:137-160. Beaumont, A. R. 1 994 . Genetics and aquaculture. In Genetics and evolution of marine organisms, p. 467-486. Cambridge Univ. Press, Cambridge, England. Bohnsack J.A. 1993. Marine reserves: they enhance fisheries, reduce con- flicts and protect resources. Oceanus 36:63-71. Botsford, L. W., B. D. Smith, and J. F Quinn. 1994. Bimodality in size distributions: the red sea urchin {Strongylncentrotiis franciscanus) as an example. Ecol. Appl. 4:42-50. Rogers-Bennett et al.: Modeling growth of Strongylocentrotus franciscanus 625 Bradbury, A. 2000. Stock assessment and management of red sea urchins [Strongylocentrotus franciscanus) inVilashington. J. Shell- fish Res. 19:618-619. Brody, S. 1927. Growth rates. Univ. Missouri Agri. Exp. Sta. Bull. 97. Burham, K. P., and D. R. Anderson 1998. Model selection and inference: a practical informa- tion theoretic approach, 353 p. Springer Verlag. New York, NY. Camahan, B., H. A. Luthier, and J. O. Wilkes. 1969. Applied numerical methods, 604 p. Wiley Press, New York, NY. Carney, D. 1991. A comparison of densities, size distribution, gonad and total-gut indices and the relative movements of red sea ur- chins Strongylocentrotus franciscanus in two depth regimes. M.S. thesis, 43 p. Univ California, Santa Cruz, CA. Dugan, J. E., and G. E. Davis. 1993. Applications of marine refugia to coastal fisheries management. Can J. Fish. Aquat. Sci. 50:2029-2042. Duetler, F. 1926. Uber das Wachstum des Seeigelskeletts. Zool. Jb. (Abt. Anat. Ontag. Tiere) 48:119-200. Ebert, T. A. 1980a. Estimating parameters in a flexible growth equa- tion, the Richards function Can. J. Fish. Aquat. Sci. 37: 687-692. 1980b. Relative growth of sea urchin jaws: an example of plastic resource allocation Bull. Mar. Sci. 30:467-474. 1982. Longevity, life history, and relative body wall size in sea urchins. Ecol. Monogr. 52:353-394. 1983. Recruitment in echinoderms. Echinoderm Studies 1:169-203. 1988. Calibration of natural gi'owth lines in ossicles of two sea urchins Strongylocentrotus purpurtaus and Echinome- tra mathaei. using tetracycline. In Echinoderm biology (R. D. Burke, P. V. Mladenov, P. Lambert, and R. L. Parsley, eds. ); proceedings of the sixth international echinoderm confer- ence, p. 435-443. A.A. Balkema, Rotterdam. 1999. Plant and animal populations: methods in demogra- phy, 312 p. Academic Press, San Diego, CA. Ebert, T. A., and M. P Russell. 1992. Growth and mortality estimates for red sea urchins Strongylocentrotus franciscanus from San Nicolas Island, California. Mar. Ecol. Prog. Ser. 81:31-41. 1993. Growth and mortality of subtidal red sea urchins Strongylocentrotus franciscanus at San Nicolas Island, California, USA: problems with models. Mar. Biol. 117: 79-89. Ebert, T. A., Dixon, J. D., Schroeter, S. C. Kalvass, P. E. Richmond, N. T. Bradbury W. A. and D. A. Woodby 1999. Growth and mortality of red sea urchin Strongylo- centrotus franciscanus across a latitudinal gradient. Mar Ecol. Prog. Ser 190:189-209. Gage, J. D. 1992. Natural growth bands and growth variability in the sea urchin Echinus esculentus: results from tetracycline tagging. Mar Biol. 114:607-616. Haaker, P. L., D. O. Parker, K. C. Barsky, and C. Chun. 1998. Growth of red abalone, Haliotis rufescens (Swainson), at Johnson's Lee Santa Rosa Island, California. J. Shell. Res. 17:747-753. Harrold, C, and D. C. Reed. 1985. Food availability, sea urchin grazing, and kelp forest community structure. Ecol. 66:1160-1169. Hastings, A. 1997. Population biology concepts and models, 227 p. Springer- Verlag, New York, NY. Jackson, J. B. C, M. X. Kirby, W. H. Berger, K. A. Bjorndal, L. W. Botsford, B. J. Bourque, R. H. Bradbury, R. Cooke, J. Eriandson, J. A. Estes, T. P. Hughes, S. Kidwell, C. B. Lange, H. S. Lenihan, J. M. Pandolfi, C. H. Peterson, R. S. Steneck, M. J. Tegner, and R. R. Warner. 2001. Historical overfishing and the recent collapse of coastal ecosystems. Sci. 293:629-638. Kalvass, P. E. 2000. Riding the roUercoaster: Boom and decline in the California red sea urchin fishery. J. Shellfish Res. 19: 621-622. Kalvass, P. E., and J. M. Hendrix. 1997. The California red sea urchin, Stronglyocentrotus fransiscanus. fishery: catch, effort, and management trends. Mar. Fish. Rev 59:1-17. Kalvass, P. E.; J. M. Hendrix, and P. M. Law. 1998. Experimental analysis of 3 internal marking methods for red sea urchins. Cahf Fish Game 84:88-99. Karpov, K.A., P. L. Haaker, I. K. Taniguchi, and L. Rogers-Bennett. 2000. Serial depletion and the collapse of the California aba- lone (Haliotis spp.) fishery. Can. Spec. Publ. Fish. Aquat. Sci. 130:11-24. Kato, S., and S.C. Schroeter 1985. Biology of the red sea urchin Strongylocentrotus fran- ciscanus, and its fishery in California. Mar. Fish. Rev. 47: 1-19. Keesing, J. K., and K. C. Hall. 1998. Review of harvests and status of world's sea urchin fisheries points to opportunities for aquaculture. J. Shell- fish Res. 17:1597-1604. Levitan, D. R. M. A. Sewell, and F-S. Chia. 1992. How distribution and abundance influence fertiliza- tion success in the sea urchin, Strongylocentrotus francis- canus. Ecol. 73:248-254. Pearse, J. S., and R. A. Cameron. 1991. Echinodermata: echinoidea. In Reproduction of marine invertebrates, vol VI( A. C. Giese, J. S. Pearse and V. B. Pearse, eds.), p. 513-662. Academic Press, New York, NY. Pearse, J. S., and A. H. Hines. 1987. Long-term population dynamics of sea urchins in a central California kelp forest: rare recruitment and rapid decline. Mar. Ecol. Prog. Ser. 39:275-283. Pearse, J. S., and V. B. Pearse. 1975. Growth zones in the echinoid skeleton. Am. Zool. 15: 731-753. Pfister, C. A., and A. Bradbury. 1996. Harvesting red sea urchins: Recent effects and future predictions. Ecol. Appl. 6:29-310. Quinn, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ Press, Oxford, England. Richards, F J. 1 959. A flexible growth curve for empirical use. J. Exp. Bot. 10:290-300. Ricker, W. E. 1954. Stock and recruitment. J. Fish Res. Board Can. 11: 559-623. Rogers, D. W. 1983. BASIC microcomputing and biostatistics, 274 p. The Humana Press Inc., Clifton, NJ. Rogers-Bennett, L. 2001. Evaluating stocking as an enhancement strategy for the red sea urchin, Strongylocentrotus franciscanus: depth- 626 Fishery Bulletin 101(3) specific recoveries. In Proceedings of the 10* international echinoderni conference; Dunedin, New Zealand, p. 527-531. A.A. Balkema, Rotterdam. 2001. Review of some California fisheries for 2000: Market squid, sea urchin, prawn, white abalone, groundfishes, ocean salmon, Pacific sardine. Pacific herring, Pacific mackerel, nearshore live-fishes, halibut, yellowfin tuna, white seabass, and kelp. Calif Coop. Oceanic Fish. Invest. Rep. 42:12-28 Rogers-Bennett, L., H. C. Fastenau, and C. M. Dewees. 1998. Recovery of red sea urchin beds following experimen- tal harvest. In Proceedings of the 9"^ internation echi- noderm conference, San Francisco, CA, p. 805-809. A.A. Balkema, Rotterdam. Rogers-Bennett, L., W. A. Bennett, H. C. Fastenau, and C. M. Dewees. 1995. Spatial variation in red sea urchin reproduction and morphology: implications for harvest refugia. Ecological Applications 5:1171-1180. Rowley, R. J. 1990. Newly settled sea urchins in a kelp bed and urchin barren ground: a comparison of growth and mortality. Mar Ecol. Prog. Sen 62:229-240. Safina, C. 1998. Song for the blue ocean, 458 p. Henry Holt, New York, NY. Sainsbury, K. J. 1980. Effect of individual variability on the von Bertalanffy growth equation. Can. J. Fish. Aquat. Sci. 37:241-247. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140 Sebens, K. P. 1987. The ecology of indeterminate growth in animals. Ann. Rev Ecol. Syst. 18:371-407. Sloan, N. A., C. P. Lauridsen, and R. M. Harbo. 1987. Recruitment characteristics of the commercially har- vested red sea urchin Strongylocentrotus frauciscanus in southern British Columbia, Canada. Fish. Res. 5:55-69. Smith, B. D., L. W. Botsford, and S. R. Wing. 1998. Estimation of growth and mortality parameters from size frequency distributions lacking age patterns: the red sea urchins (Strongylocentrotus franciscanus) as an example. Can J. Fish. Aquatic Sci. 55:1236-1247. Tanaka, M. 1982. A new growth curve which expresses infinite increase. Publ. Amakusa Mar. Biol. Lab. 6:167-177. Tegner, M. J. 1989. The feasibility of enhancing red sea urchin, Strongy- locentrotus franciscanus, stocks in California: an analysis of the options. Fish. Bull. 51:1-22. Tegner, M. J, and P. K. Dayton. 1977. Sea urchin recruitment patterns and implications of commercial fishing. Science 196:324-326. Tegner, M. J. and L. A. Levin. 1983. Spiny lobsters and sea urchins: analysis of a predator- prey interaction. J. Exp. Mar Biol. Ecol. 73:125-150. Troynikov, V. S., and H. K. Gorflne 1998. Alternative approach for establishing legal minimum lengths for abalone based on stochastic growth models for length increment data. J. Shellfish Res. 17:827-831. von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquires on growth laws II). Human Biol. 10:181-213. Walford, L. A. 1946. A new graphic method of describing the growth of animals. Biol. Bull. 90:141-147. Wien, W. 1896. Uber die Energieverteilung im Emissionspectrum eines Schwarzen Korpers. Annalen der Physik 1896: 662-669. In The conceptual development of quan- tum mechanics (M. Jammer, author), 2"'' ed. 1989, p 8-10. Tomask Publishers, American Institute of Physics, Woodby New York, NY. Wilson, C. W., D. W. Beckman, J. D. Dean, and J. Mark. 1987. Calcein as a florescent marker of otoliths of larval and juvenile fish. Trans, Am. Fish. Soc. 1 16:668-670. Yamaguchi, G. 1975. Estimating growth parameters from growth rate data. Problems with marine sedentary invertebrates. Oecologia 20:321-332. 627 Abstract— Age and growth estimates for the blue shark [Prionace glauca) were derived from 411 vertebral centra and 43 tag-recaptured blue sharks col- lected in the North Atlantic, ranging in length from 49 to 312 cm fork length (FL). The vertebrae of two oxytetracy- cline-injected recaptured blue sharks support an annual spring deposition of growth bands in the vertebrae in sharks up to 192 cm FL. Males and females were aged to 16 and 15 years, respectively, and full maturity is attained by 5 years of age in both sexes. Both sexes grew similarly to age seven, when growth rates decreased in males and remained constant in females. Growth rates from tag-recaptured individuals agreed with those derived from vertebral annuli for smaller sharks but appeared overestimated for larger sharks. Von Bertalanffy growth parameters derived from vertebral length-at-age data are L^ = 282 cm FL, A' = 0.18, and t„ = -1.35 for males, and L, = 310 cm FL, K = 0.13, and ig = -1.77 for females. The species grows faster and has a shorter life span than previ- ously reported for these waters. Age and growth of the blue shark (Prionace glauca) in the North Atlantic Ocean* Gregory B. Skomal Massachusetts Division of Marine Fisheries Marthas Vineyard Research Station PO Box 68 Vineyard Haven, Massachusetts 02568 E-mail address: Gregory.SkomaiiSlstale.ma.us Lisa J. Natanson National Manne Fisheries Service, NOAA 28 Tarzwell Dr. Narragansett, Rhode Island 02882 Manuscript approved for publication 5 March 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Pubhcations Office. Fish Bull. 101:627-639 (2003). The blue shark {Prionace glauca) is a large pelagic carcharhinid that is widely distributed in the world's oceans. Throughout its range, it is considered the most abundant species of large shark (McKenzie and Tibbo, 1964; Casey, 1982). In the Atlantic, the blue shark is distributed from Newfoimdland to Argentina in the west and Norway to South Africa, including the Mediter- ranean, in the east (Compagno, 1984). There is strong evidence from tagging data and catch records that blue sharks in the North Atlantic constitute a single stock (Kohler et al., 2002). Moreover, mitochondrial DNA d-loop sequence and nuclear microsatellite analyses indicate no differences between blue sharks from the eastern and western North Atlantic (ShiyjiM. Distribution and movements of the blue shark are strongly influenced by seasonal variations in water tem- perature, reproductive condition, and availability of prey (Kohler et al., 2002). Blue sharks make frequent trans-Atlantic movements between the western and eastern regions, utilizing the major North Atlantic current sys- tems (Stevens, 1976, 1990; Casey, 1982, 1985; Kohler et al., 2002). Temporal and geographic patterns of size and sexual segregation have been described for this species, and mating areas and pupping areas are reported to be in the western and eastern regions of the North Atlan- tic, respectively (Casey, 1982; Kohler et al., 2002). Pregnant females are rare in the western North Atlantic, which is dominated by juveniles of both sexes, adult males, and subadult females (Pratt, 1979; Casey, 1982; Kohler et al., 2002). Catch records from the eastern North Atlantic largely comprised neo- nates and juveniles of both sexes and adult females (Aasen, 1966; Stevens, 1975, 1976; Connett, 1987; Silva et al., 1996; Kohler et al., 2002). Although subjected to a number of fisheries, the blue shark is primarily taken as bycatch in longline fisheries throughout the North Atlantic (ICCAT, 2002). Most blue sharks are discarded or only their fins are harvested because of the low palatability of their flesh (Castro et al., 1999). Although incom- plete, blue shark landings estimates in the North Atlantic reported to the International Commission for the Conservation of Atlantic Tunas were 25.1 and 24.2 thousand metric tons (t) in 1998 and 1999, respectively (ICCAT, 2002). Domestic longline fisheries in the western North Atlantic rarely land blue sharks, but it was estimated that annu- al dead discards ranged from 2.8 to 29.3 thousand blue sharks (99.0-1136.3 t) during the period 1987-2000 (Cortes, 2002). The major source of landings in the U.S. has been the recreational fish- ery, which landed 6.8 thousand blue sharks in 2000 (Cortes, 2002). * Contribution 8 of the Massachusetts Divi- sion of Marine Fisheries, P.O. Box 68, Vine- yard Haven, MA 02568. ' Shivji, M. 2002. Personal commun. Nova Southeastern University, 8000 North Ocean Dr., Dania Beach, FL 33004. 628 Fishery Bulletin 101(3) Ecologically, the blue shark is an apex predator of im- portant teleosts and cephalopods (Stevens, 1973; Tricas, 1978; Kohler, 1987). Historical fisheries have shown that sharks are intrinsically sensitive to sustained exploitation (see review by Castro et al., 1999). Slow growth, late ages at maturity, and low fecundities reflect the life history strate- gies of /^-selected species; stock size is closely linked to re- cruitment (Hoenig and Gruber, 1990). Although the current Fishery Management Plan for Atlantic Tunas, Swordfish, and Sharks has established limits on the U.S. commercial and recreational fisheries that impact blue sharks (NMFS, 1999), no international management is currently in place. Given a single North Atlantic stock for this species, any fisheries exploitation, regardless of its coastal origin, may impact the population. Accurate age determinations are necessary for both the assessment and management of the blue shark because they form the basis for calculations of growth and mortality rates, age at maturity, age at recruit- ment, and estimates of longevity. Age and growth of the blue shark have been described by a number of studies to varying degrees. In the North Pa- cific, Cailliet et al. ( 1983 ) and Tanaka et al. ( 1990) used ver- tebral growth rings and Nakano ( 1994) used both vertebrae and length-frequency modes to establish growth curves for the blue shark. In the North Atlantic, Aasen (1966) aged the species by assigning ages to length-frequency modes. Later, Stevens (1975), Silva et al. (1996), and Henderson et al. (2001) established growth curves from vertebral growth rings of juvenile blue sharks sampled in the eastern North Atlantic. Low sample sizes, inadequate size ranges, and the lack of age validation limit the utility of these studies for the North Atlantic blue shark population. Skomal (1990) generated growth curves for the blue shark from vertebral growth rings, tag-recaptures, and length-frequency data. In that study, vertebrae from oxytetracycline (OTC) in- jected recaptured blue sharks were used to validate age estimates. The purpose of the current study is to augment the work of Skomal (1990) with additional tag-recapture data, with corroborative vertebral readings of a different vertebral processing technique, and with more rigorous growth analyses. Materials and methods Interpretation of vertebrae Vertebrae were obtained from blue sharks caught on research cruises, commercial, and recreational fishing ves- sels, and at sport fishing tournaments between 1966 and 2001. Primary sampling took place between Cape Hatteras, NC, and the Gulf of Maine (NE coast of the United States). To adequately represent the entire size range of the species, small sharks were obtained from the eastern Atlantic from cooperative fishermen and research scientists. When pos- sible, the 15'^ through 20''' vertebrae were excised for the study. When such precision was not possible, this section of backbone was approximated by cutting at the branchial region adjacent to the fifth gill arch. Excess muscle and connective tissue were removed from the vertebrae with a knife. Vertebrae were stored either frozen or preserved in 10% buffered formalin or 70% ethanol. Only samples that had measured fork length (FL — tip of the snout to the fork in the tail, over the body curva- ture), total length (TL — tip of the snout to a point on the horizontal axis intersecting a perpendicular line extending downward from the tip of the upper caudal lobe to form a right angle), or precaudal length (PCL — tip of the snout to the precaudal pit, over the body curvature) were used (Kohler et al., 1995). All lengths reported are in FL unless otherwise noted. TL can be converted to FL by using the regression (Kohler et al., 1995): FL = 0.8313 (TL) + 1.39 [«=572 r-=0.991. PCL can be converted to FL using the regression (NMFS'^) PCL = 0.9075 (FL) - 0.3956 [n = 106 r2=0.99]. One vertebra from each sample was removed for pro- cessing. The centrum was sectioned by using a Ray Tech Gem Saw with two diamond blades separated by a 0.6-mm spacer Each centrum was cut through the middle along the sagittal plane; the resulting bow-tie sections were stored in individual capsules in 70% ethanol. Each section was digitally photographed with a MTI CCD 72 video camera attached to a SZX9 Olympus stereo microscope by using reflected light. All samples were photographed at a mag- nification of 4x. Band pairs (consisting of one opaque and one translucent band) were counted and measured from the images by using Image Pro 4 software (Media Cyber- netics, Silver Spring, MD). Measurements were made from the midpoint of the notochordal remnant of the full bow-tie to the opaque growth bands at points along the internal corpus calcareum. The radius of each vertebral centrum ( VR) was measured from the midpoint of the notochordal remnant to the distal margin of the intermedialia along the same diagonal as the band measurements. Specimens previously processed histologically (Skomal, 1990) were used for counts when whole samples for those specimens were not available for reprocessing. Because of the differ- ent processing method, histological sections were not used for measurements. The criterion for what constitutes a band pair (annulus) was based on the contouring of the corpus calcareum in relation to the strength of the band. A clear indentation of the corpus calcareum at the position of an opaque band constituted the consummation of a growth layer within the vertebra and was considered the annulus (Fig. 1). Each lay- er was considered a temporal growth zone. The first opaque band distal to the focus was defined as the birth mark (BR) and a slight angle change in the corpus calcareum coincided with this mark. In addition, identification of the birth band was confirmed with back-calculation and by comparison of the radius of this band with the radius of vertebrae from young of the year (YOY) and full-term embryos. •^ NMFS (National Marine Fisheries Service). 2001. Unpubl. data. Apex Predators Program, 28 Tarzwell Dr, Narragansett, RI 02882. Skomal and Natanson: Age and growth of Pnonace glauca 629 The relationship between VR and FL was calculated to determine the best method for back-calculation of size-at-age data and to confirm the interpretation of the birth band. Age was calculated for each fish based on a birth date of June 1 (Pratt, 1979), corrected for date of capture. Regressions were fitted to the male and female size-at-age data and an ANCOVA was used to test for difference between the two relationships. The relation- ship between FL and VR was best described by a polynomial equation; therefore the data were In-transformed before linear regression. The Fraser-Lee equation of the In-transformed data was derived for back calculation: liUFLJ = b + (ln[FLJ+b) (InradiusJ (InradwsJ'K where a = age; b = intercept from the regression; and c = capture. Validation To evaluate the periodicity of band pair formation, vertebrae from OTC-injected and measured tag-recaptured sharks were exam- ined. Over 350 blue sharks of various sizes were measured, tagged, and injected with a 25 mg/kg body weight dose of OTC by scientific personnel aboard research and commer- cial vessels in the North Atlantic. Upon recapture, vertebrae were removed from injected specimens and stored in 70% ethanol or were frozen. Vertebrae from these sharks were processed, digitally photographed as previously described, and examined for the OTC mark with reflected UV light. The number of band pairs distal to the OTC mark was then compared with the number of years at liberty and expressed as the proportion of the previous complete growth zone. Data analysis Aging bias and precision of bands counts were examined by using age-bias plots and the coefficient of variation (Cam- pana et al., 1995). Reader 2 counted 98 sections previously counted by reader 1 (Skomal, 1990). Pairwise comparisons were generated from these data. Von Bertalanffy growth functions (VBGF) were fitted to length-at-age data by using the following equation (von Bertalanffy, 1938): Z., = Z..(l-e-''"-'"'), where L, = predicted length at time t; L^= mean asymptotic length; K = a. growth rate parameter (yrM; and Iq = the theoretical age at which the fish would have been zero length. The VBGF was calculated by using the nonlinear regression function in Statgraphics (Manugistics, Inc., Rockville, MD). Figure 1 Photograph of a vertebral section from a male blue shark (Prionace glauca) estimated to be 14 years old. Tagging data From 1963 through 1999, members of the NMFS Coopera- tive Shark Tagging Program tagged 88,899 and recaptured 4967 blue sharks. Only those sharks reliably measured by biologists or fishermen at both tagging and recapture were used in the analyses. All measurements were converted to FL by using the relationships of Kohler et al. (1995). The Gulland and Holt ( 1959) and Francis ( 1988a) models were used to generate VBGFs from the tag-recapture data. The Gulland and Holt ( 1959) method uses graphical inter- pretation of the recapture data to produce estimates of L, and K. Specifically, annualized growth rate (cm/yr ) was plot- ted against average FL (cm) between tagging and recapture to calculate linear regression coefficients. The slope of the line is equal to -K and the x-axis intercept is equal to L^. The Francis ( 1988a) method (GROTAG) uses maximum likelihood techniques to estimate growth parameters and variability from tagging data. A coefficient of variation of growth variability (v), measurement errors (m and s) and outlier contamination (p) are estimated, as well as growth rates at two user selected lengths (« and p). The reference lengths, a and p, were chosen to lie within the range of tagged individuals. The form of the von Bertalanffy equa- tion becomes AZ.= Su-gp 1-1 + The simplest model, a linear fit with minimal parameters (o and s) was used initially and additional parameters 630 Fishery Bulletin 101 (3) were added to successively increase the model complexity. Significant improvement in the model results were deter- mined by using log likelihood ratio tests in accordance with Francis {1988a). Bootstrapping was used to calculate the 95% confidence intervals for the final parameter estimates. The modeling and bootstrapping were carried out by using a Solver based spreadsheet in MS Excel (Microsoft Corp., Redmond, WA) (Simpfendorfer^). The value oft^ cannot be estimated from tagging data alone, it requires an estimate of absolute size at age, such as size at birth, and was calcu- lated with the VBGF by solving for t^ such that t„=l + {\/ K)[\n\L^-L,/ Lj] where L, = known length at age (size at birth); K = the von Bertalanffy growth constant; and L^ = the theoretical maximum attainable length from the VBGF. The 99% ofL^ using the equation > 99% = 5 (ln2) Results Interpretation of vertebrae Vertebral samples from 411 blue sharks were used in our study: 287 males, 119 females, and five of unknown sex. These samples comprised free-living sharks ranging from 49 cm to 312 cm FL. In addition, vertebrae from seven late- term embryos ranging from 36 cm to 43 cm FL were exam- ined. Blue shark vertebrae did not have consistent prebirth marks; thus, the first distinct opaque band was generally considered the birth mark. The location of the birth band coincided with a slight angle change (Fig. 1). The FL-VR relationship was slightly curvilinear and the In-transformed data provided a better linear fit (Fig. 2), In(FL) = 0.89*ln(VR) + 3.10 [r!=392r2=0.97]. ■* Simpfendorfcr, C. 2000. Personal commun. Mote Marine Laboratory, 1600 City Island Park, Sarasota, FL 33577. There was no significant difference between the sexes (ANCOVA,P>0.10). Confirmation of the birth band was made by comparison of the BR of all individuals to the VR of YOY and late-term embryos ( Fig 2 ). The VR of seven late-term embryos ( mean VR ±95% CI=2.04 ±0.25) was slightly less than the BR value of the total sample (mean BR ±95% CI=2.70 ±0.03; n=351); the mean VR of 11 early YOY was slightly higher than the BR of the entire sample (49-58 cm FL; mean VR ±95% CI =2.97 ±0.18) (Fig. 2). The location of the birth ring between the VR of both the late-term embryos and the YOY indicated that the birth ring was identified correctly. Validation OTC-injected recaptured blue sharks provided evidence for the use of vertebral band pairs as age indicators. Vertebrae from two OTC-injected sharks were returned after 0.7 and 1.5 years at liberty (Table 1). OTC injection produced strong fluorescent marks in the vertebral centra of both these sharks (Fig. 3) and the number of annuli past the OTC mark coincided with the number predicted from time at liberty (Table 1 ). In OTC-injected recaptured shark (B536), an opaque growth band was deposited just after tagging in May (Fig. 3). In recaptured shark B 116452, an opaque growth band was deposited just prior to tagging in June (Fig. 3). These results suggest an annual spring deposition of growth zones within the vertebrae. Thus, ver- tebral annuli were validated in these two sharks, which were up to 4* years of age; the older of these fish (B536) corresponded to this age. Beyond this age, bands were assumed to be annual on the basis of the similar nature of band deposition. Comparison of counts between two readers indicated no appreciable bias (Fig. 4). The coefficient of variation fiuc- tuated around 15%. This level of precision was considered acceptable; thus, counts generated by both readers and preparation methods were combined for the analyses. The reader maintained quality control by periodically recount- ing earlier samples and by cross-checking the readings. Length-at-age data indicated that males and females grow at roughly the same rate. The overlap in observed size-at-age data, as well as the graphical representation of the VBGF curves, indicated that there is little differ- ence in growth for the sexes (Fig. 5). However, the LOW- ESS (locally weighted regression smoothing) derived curves as well as the VBGF parameters indicated that, theoretically, females grow slower and to a larger overall size than males (Table 2, Fig. 6). The LOWESS curves clearly showed minor differences in growth beginning at approximately seven years of age (Fig. 6), but this was likely an artifact of low female sample size at older ages. Subsequent analyses are presented for each sex and for sexes combined for ease of comparison with previously published studies. Skomal and Natanson: Age and growth of Prionace glauca 631 300 1 Mean radius of birth mark (n = 351 ) °° o » o - 250 • 200 ■ d __^ Q nffflMJflP^ 1 E nxfiiBBH^' o ^^StP^ ^ s: ^X^ "> P 150 ■ to O U- ■) o oO O Male WNA(n = 273) Full term a Female WNA(n= 112) 100- embryos \ A MaleENA(n = 7) \ ■ Female ENA (n ^ 2) 50- \ f • \ Average size at birtti Smallest free-tiving (n = 11) 0 5 10 15 20 Vertebral radius (mm) Figure 2 Relationship between vertebral radius and fork length in the blue shark ^Prionace glauca). Tag-recapture recapture. data for OTC Table 1 -injected recaptured blue shark (Prionace glauca). TFL = fork length at tagging. RFL = fork length at Sample number Sex TFL (cm) RFL (cm) Date tagged Date recaptured Years at liberty Growth (cm) No. of bands after OTC mark B 116452 B536 F M 116 172 162' 192 18Jun 1987 9 May 1985 21 Dec 1988 16 Jan 1986 1.5 0.7 33 29 1.20 0.68 ' Calculated from precaudel length. Tagging data A total of 43 blue sharks was recaptured with sufficient information for tag-recapture analysis. Data from 18 sharks at liberty >0.9 years were used for Gulland and Holts (1959) method and all the recaptured sharks were used for the Francis ( 1988a) method (GROTAG). The results of the likelihood ratio tests with GROTAG (Francis, 1988a) showed that the more complex nonlinear model with all six parameters was the best fit for these data (Table 3, model 3). The mean annual growth rates are ggg = 44.2 cm/yr and gjg^ = 25.5 cm/yr, corresponding to growth rates at a FL= 90 cm and 180 cm, respectively (Fig. 7). Von Bertalanffy estimates from the Gulland and Holt (1959) and GROTAG (Francis, 1988a) methods produced similar von Bertalanffy curves (Table 4, Fig. 8A). Longevity The maximum age determined from vertebral band pair counts was 16 and 15 years for males and females, respec- tively. These ages are likely to be an underestimate of longevity, given the history of fisheries exploitation of this species. Using Taylor's (1958) method, we determined that the age at which 95% and 99% of the L^ is reached was 16.5 and 26.1 years, respectively. Fabens (1965) method for >99% longevity produced an estimate of 20.7 years. 632 Fishery Bulletin 101(3) Figure 3 Vertebral sections from two OTC-injected blue sharks [Prionace glauca). Annuli and birth marks are indicated. Discussion Several methods have been employed to vali- date or verify (or both) age estimates derived from vertebral banding patterns (Cailliet, 1990). Although corroborative verification often comes from the interpretation of length-frequency data, laboratory and field growth studies, and centrum edge analyses, direct age validation for sharks is limited to the interpretation of vertebral banding patterns in OTC-injected fish. In his review of elasmobranch age and growth studies, Cailliet (1990) found validated growth curves for only six species, which included three carcharhinids: the lemon (Negaprion brevirostris); the sandbar (Carcharhinus plutnbeus); and the Atlantic sharpnose [Rhizoprionodon terraenovae) sharks. Although more than ten years have trans- pired since this review, validated growth curves for sharks are still lacking. In lamnids, direct valida- tion of annual band deposition with the use of OTC has been reported in a single species, the porbeagle shark, Larnna nasus (Natanson et al., 2002). Al- though age estimates from vertebral banding pat- terns have been reported for several carcharhinids, including the oceanic whitetip shark, Carcharhin us longimanus (Seki et al., 1998; Lessa et al., 1999), the dusky shark, C. obscurus (Natanson et al., 1995; Natanson and Kohler, 1996; Simpfendorfer, 2000), the blacktip shark, C. limbatus (Wintner and Cliff, 1995 ), and the bronze whaler, C. brachyurus (Walter and Ebcrt, 1991 ), ago interpretations were not validated and vertebral bands were assumed to be annual. More recently, Simpfendorfer ot al. (2002) validated the annual formation of vertebral banding patterns in C. obscurus from Western Australian waters. In the current study, we have validated annual band pair deposition in Prionace glauca up to 4* years in age using 18 -| CV = 0.15 3 4 1 1 % 16 - N = 98 , 3 \ 14 • 7 1 Al y 12 - 13 ,1 \ y f 10 - 8 • 26 15 7 y V ■k y\ r^ 1 6 ■ k>\ 4 - ^ 2 y^ 0 ■ ^ Age (yr) of reader 1 Figure 4 Age bias graph for pair-wise comparison of 98 blue shark i.Prio- nace glauca) vertebral counts from two independent readers. Each error bar represents the dSVr confidence interval for the mean age assigned by reader 2 to all fish assigned a given age by reader 1. The one-to-one equivalence line is also presented. vertebrae from two OTC-injected fish. These data indicate that annulus formation occurs in the spring. This seasonal formation is further supported by the marginal increment analysis of Skomal ( 1990 ), which shows that one band pair is formed annually. However, the low sample size and the lack of OTC-injected recaptured fish over the entire size range of the species do not allow for full age and growth validation. Clearly, the study requires OTC-injected recaptured blue sharks over a broader size range and greater time at lib- erty— a requirement that is not atypical of age and growth Skomal and Natanson: Age and growth of Prionace glauca 633 studies on large highly migratory elasmobranchs. Wintner and Dudley (2000) used two OTC-injected recaptured individuals to conclude that growth band deposition is annual in the tiger shark {Ga- leocerdo ciivier). Moreover, Natanson et al. (2002) validated annuli in the porbeagle shark up to 11 years of age by using only two OTC-injected and six YOY recaptured individuals, although the species was aged to 25 years. The processes that govern vertebral growth have yet to be described in elasmobranchs. The pattern varies from one ring per year in most carcharhinids (Cailliet, 1990), and two rings per year in some lamnids (Parker and Stott, 1965; Pratt and Casey, 1983) to the complete absence of periodicity (Natanson, 1984). Some research- ers feel that temperature plays a major role in this process (Stevens, 1975; Ferreira and Vooren, 1991). The blue shark, however, remains within a discrete temperature range year-round (Stevens, 1975; Sciarrota and Nelson, 1977; Casey, 1982). Moreover, acoustic tracking has shown that blue sharks experience large changes in body tempera- ture (up to 7°C) as they routinely pass through the thermocline in their daily periodic dives from the surface to depths of 200-600 m (Carey and Scharold, 1990). The ecology of this species may provide a more likely explanation of annulus formation. Kohler (1987) found a seasonal cycle for energy storage that correlated with the migratory patterns of the blue shark. In general, blue shark condition was found to be at an annual low in the winter and spring. Blue sharks use energy stores during this time for extensive north-south and trans-Atlantic migrations (Casey, 1985; Kohler, 1987) and periodic deep dives (Carey and Scharold, 1990). It is logical that growth may be depressed during these months, thereby causing a check or annulus in the vertebrae. Tag-recapture data provide verification of the growth curves derived from vertebral banding. Francis (1988b) suggested that growth curves generated from age-length and length-increment (tagging) data are not directly comparable and that the comparison of growth rates at length was more appropriate. Although VBGF parameters derived from tagging data are noticeably higher, growth rates were similar for both methods (Fig. 7). The higher L^ and K can be attributed to the different derivation of the VBGF parameters and the absence of older recaptured sharks in the sample. Pratt (1979) proposed that maturity in the male blue shark occurs at 183 cm FL and this would coincide with an age of 4-5 years based on the results of the present study. Females enter a distinct subadult phase (Pratt, 1979) at 145 cm FL and 2* years of age. Full maturity in females is attained at 185 cm FL (Pratt, 1979), which corresponds to about 5 years of age. Previous estimates of age and growth of the blue shark in the Atlantic have been determined from vertebral banding patterns, and verification has been made from the interpre- tation of length-frequency and tagging data (Stevens, 1975; 350 - 300 - • aogft^-^O 'ork length (cm) o o o J^*^^ 100 ■ Jr^ O Males A^ * Females M —0— Female VBGF | 50 i f Male VBGF 0 2 4 6 8 10 12 14 16 18 Age (years) Figure 5 Prionace glauca growth curves and size-at-age data based on verte- bral band counts. Von Bertalanffy growth function curves have been fitted to the data by sex. Table 2 Von Bertalanffy growth function parameters and 95% \ confidence intervals calculatec by using vertebral and tag- recapture methods for the blue shark (Prionace glauca). | n = number of sharks in sample. Method L, K to n Vertebral Combined 286.8 0.17 -1.43 411 CI ±7.32 0.01 0.20 Male 282.3 0.18 -1.35 287 CI ±7.15 0.02 0.23 Female 310.8 0.13 -1.77 119 CI ±34.8 0.03 0.50 Tag-recapture GROTAG Combined 302.4 0.23 -0.69 43 Gulland and Combined 331.7 0.19 -0.77 18 Holt (1959) CI ±80.0 0.12 Silva et al., 1996; Henderson et al., 2001) (Table 5, Fig. 8). The eastern Atlantic vertebral sample of Stevens (1975) comprised largely females (89%), ranging from 34 cm to 227 cm FL. The resulting growth curve, therefore, largely reflects female growth (Fig. 8C). His use of whole silver- stained centra coupled with the lack of maximum-size fish allowed for the interpretation of only six annuli. From only mean back-calculated lengths at ages two through five, Ste- vens extrapolated growth of the species with a VBGF to an age of 20 years. Similarly, Silva et al. (1996) and Henderson et al. (2001) investigated age and growth in this species with whole vertebrae from sharks sampled in the eastern North Atlantic. In the former study, vertebral SEunples from 634 Fishery Bulletin 101(3) 350 A 300 250 200 150 r °V o o ^^- 100 r r />" 50 Lf E o 0 -, . . . _l L 8 10 12 14 16 o Li. 350 300 r B o 250 r o i^-^-^^^ <> 200 r oo o o 150 100 L^ ^ 50 r ' 0 ~ 1 . . . 1 ... 1 ... 1 . . . 1 6 8 10 12 14 Age (years) 16 Figure 6 Prionace glauca growth curves and size-at-age data based on vertebral band counts. LOWESS (locally weighted regression smoothing) curves have been fitted to the data by sex: (A) males and (B) females. 308 juvenile blue sharks collected in the Azores were used to model early growth in this species. Silva et al. (1996) calculated an annual growth rate of 30 cm/yr for the first five years of life and aged the samples to seven years. More recently, Henderson et al. (2001) used 159 vertebrae sampled from blue sharks taken from oceanic waters off Ireland. Like the previous two studies, the size range of samples was limited to juvenile fish less than 191 cm FL and the estimated ages ranged from 1 to 6 years. Stevens (1975), Silva et al. (1996), and Henderson et al. (2001) modeled blue shark growth with the VBGF. These curves are similar to each other (Silva et al., 1996, Hender- son et al., 200 1 ), yet show slower growth than the current study (Fig. 8) despite the fact that we used criteria similar to those of Stevens ( 1975) for vertebral interpretation. This result is not surprising in light of the fact that these three studies share common methods and sample biases. All three of the previous studies were performed on juvenile sharks from the eastern North Atlantic, the vast majority of which were between 100 and 184 cm FL. Because of the lack of samples from very small fish, one study (Silva et al., 1996) included vertebral readings from full-term embryos in the growth curve. It is well documented that embryonic growth is not comparable to postnatal gi'owth (Casey et al., 1985; Pratt and Casey, 1990) and, therefore, embryos should not be included in a postnatal growth curve. The lack of large and small specimens in the calculations of these growth curves is particularly problematic because validation of the first growth increment is essential as it forms the basis of further counts. Moreover, the smallest and largest of the specimens are the most influential in the estimation of growth (Campana, 2001). All three of the previous studies used similar whole centrum vertebral processing techniques and band count criteria, which would lead to corroborating counts, yet not necessarily to accurate counts (Campana, 2001). Wliole vertebrae simply do not allow for high band resolution in older slower growing fish. Therefore, counts from whole Skomal and Natanson: Age and growth of Prionace glauca 635 Log-likelihood function values i GROTAG (Francis 1988a). For a by at least 1.92 (Francis 1988a). Table 3 ind parameter estimates for three growth models fitted to Prionace significant (P<0.05) improvement in fit, the introduction of one extra * indicates fixed parameters. Model 3 shows 95% confidence intervals glauca tagging data using parameter must increase X Parameter Symbol (unit) Model 1 2 3 Log likelihood Mean growth rates Growth variability Measurement error Outliers g90 (cm/yr) gl80 (cm/yr) V s (cm) »! (cm) P -197.29 21.53 10.92 0* 1.06 0* 0.83 -176.91 39.04 21.90 0.46 1.37 0* 0.28 -174.61 44.18 (35.37-54.33) 25.46(19.29-33.41) 0.27 (0.06-0.44) 5.39(2.25-7.40) -2.03 (-5.37-2.10) 0.18 vertebrae generally underestimate ages in larger indi- viduals. The counts obtained in the three eastern Atlantic studies may be accurate because they are from juvenile sharks where vertebral bands are not compressed. In fact, juvenile growth from our size-at-age data overlaps the growth cui-ves from these studies. However, the VBGF growth curves and resulting estimates of growth rate and age at maturity from the eastern Atlantic studies are suspect because of the lack of fish at the lower and upper end of the curve. The general lack of maximum-size fish in these studies resulted in the estimation of an artificially inflated L^ and, therefore, a lower growth rate iK) for this species (Table 5). Vertebral band deposition was assumed to be annual in these studies, but low sample sizes, sample bias, and lack of validation limits the utility of this previ- ous work. In the current study, the use of sections and the adequate representation of the entire size range for both sexes yielded more accurate age estimates of 16 and 15 years for males and females, respectively. Age and growth estimates of the blue shark in the North Pacific have been determined by using vertebral bands and length-frequency data (Cailliet et al., 1983; Tanaka et al., 1990; Nakano, 1994 ). Although the VBGF was used to mod- el growth based on vertebral interpretation, the resulting parameters differed greatly among studies (Table 5). In general, Cailliet et al. (1983) reported a male growth rate similar to that in our present study, but a much smaller L^ (Table 5). For females, the latter holds true, but the growth coefficient is much higher (0.25) than reported in our study. Tanaka et al. ( 1990 ) found a similar growth trend in the western North Pacific with females growing faster than males, but the VBGF parameters were very differ- ent with higher L^ and lower K values. When compared to our study, the VBGF parameters of Tanaka et al. ( 1990) yield slower growth and a greater maximum size for males and a similar growth rate and smaller maximum size for females. Tanaka et al. ( 1990 ) attributed these intra- and in- ter-oceanic differences to the different methods used. More recently, Nakano (1994) sampled blue sharks across the North Pacific and derived VBGF growth parameters that Table 4 Size at age (cm) for the blue shark (Prionace glauca) calcu- lated from von Bertalanffy equations based on tag-recap- ture and vertebral methods. Age (yr) Vertebral method Tag-recapture GROTAG method combined male female 0 61.0 60.9 66.1 45 1 95.8 97.4 97.0 99 2 125.2 127.8 124.0 141 3 150.1 153.3 147.6 175 4 171.2 174.5 168.2 201 5 189.0 192.3 186.2 222 6 204.1 207.1 201.9 239 7 216.8 219.5 215.7 252 8 227.6 229.8 227.7 263 9 236.7 238.5 238.2 10 244.4 245.7 247.4 11 251.0 251.7 255.4 12 256.5 256.8 262.4 13 261.1 261.0 268.5 14 265.1 264.5 273.9 15 268.4 267.4 278.5 16 271.3 269.9 were similar to those of Tanaka et al. ( 1990), but estimated growth rate to be slower than that of our present study. It is difficult to ascertain whether interoceanic differences in growth are real or are an artifact of method. Although Tanaka et al. (1990) presented data to support the latter within the North Pacific, the much larger maximum size attained by this species in the North Atlantic (Strasburg, 1958; Tanaka, 1984) cannot be overlooked in relation to interoceanic growth differences. 636 Fishery Bulletin 101(3) Longevity estimates for the blue shark indicate that they may Hve for 26 years when Taylor's ( 1958) method is employed. On the other hand, Fabens" (1965) method for >99% longevity produced an estimate of 20.7 years, which may be more realistic. The maximum age determined from vertebral band-pair counts was 16 and 15 years for males and females, respectively. An analysis of maximum times at liberty for tagged blue sharks supports the notion that this species does not live as long as previously reported for the North Atlantic. Of the 4967 blue sharks recaptured to date. 70 T ♦ Growth/yr T/R Vertebral 60 - ♦ — ♦- Tag/recapture (Francis 1988) ♦ ♦ 50 ^ ^ ♦ ♦ ♦ ♦ ■H- 40 *♦♦ .c I ♦ 2 30- ♦ <♦ O) M ^"^^ * i 20- ♦ ^^-...^^^ c < 10 ■ ♦ ^^--^ * ♦ ^""~<^^^ 0 ■ 50 100 150 200 250 300 Fork length (cm) Figure 7 Comparison of the annual growth rates of the blue shark (Prionace glauca) derived from multiple aging methods. 99% were at liberty for less than five years. The maximum times at liberty are 9.1 and 8.5 years, despite the 39-year history of the tagging program. The shark at liberty for 9.1 years was a male tagged at an estimated 122 cm FL; size at recapture was not reported. According to our growth curve, the shark was tagged at 1* years of age, which would cor- respond to a maximum age of 10+ years at recapture. The shark at liberty for 8.5 years, also a male, was estimated to be 198 cm FL at tagging, which would correspond to 5* years of age. Therefore, at recapture, this fish would be a maximum age of 13.5 years, although its measured FL at recapture actually corresponds to 11 years on our growth curve. The largest long-term recap- ture was a male, 244 cm FL at tagging and 266 cm FL at recapture 6 years later. This would cor- respond to an estimated age of 10 years at tagging and 16 years at recapture, which falls well within the values of directly aged vertebrae (Fig. 5). The occurrence of sexual differences in growth is well documented in elasmobranchs; females usually grow larger than males (Cortes, 2000). Although the largest blue shark in our study was a 312-cm-FL female, there is little evidence that large females are highly abundant in the North At- lantic. Maximum size male and female specimens in our study, 284 cm FL and 312 cm FL, respective- ly, represented the largest reliably measured blue sharks from the North Atlantic, with the exception of a 320-cm-FL specimen (sex unspecified) exam- ined by Bigelow and Schroeder (1953). Indeed, a thorough review of the literature reveals that although 288-cm-FL and 279-cm-FL females were reported by Gubanov and Grigoryev (1975) from the Indian Ocean, males are consistently cited as being very much larger than females in the world's Table 5 Van Bertalanffy growth function parameters and maximum age derived from vertebral bands in the blue shark (Prionace glauca) \ separated by location and sex. Sex Ocean Region n Z., A' T, Max. age Authors Male North Atlantic All 287 282.3 0.18 -1.35 16 Current studv East 112 309.0 0.12 -1.07 5 Silvaetal. (1996) North Pacific East 38 246.7 0.18 -1.11 9 Caillietetal. (1983) West 43 308.1 0.10 -1.38 7 Tanakaet al.(1990) All 148 319.5 0.13 -0.76 10 Nakano(1994) Female North Atlantic All 119 286.8 0.16 -1.56 15 Current study East 82 353.0 0.11 -1.04 6 Stevens (1975) East 170 382.0 0.09 -1.19 5 Silvaetal. (1996) North Pacific East 88 202.6 0.25 -0.80 9 Caillietetal. (1983) West 152 254.1 0.16 -1.01 8 Tanakaetal. (1990) All 123 268.9 0.14 -0.85 10 Nakano(1994) Combined North Atlantic All 411 285.4 0.17 -1.41 16 Current study East 336 284,0 0.14 -1.08 5 Silvaetal. (1996) East 159 314.4 0.12 -1.33 6 Henderson etal. (2001) North Pacific East 130 222.1 0.22 -0.80 9 Caillietetal. (1983) Skomal and Natanson; Age and growth of Prionace glauca 637 oceans (Suda, 1953; Tucker and Newnham, 1957; Aasen, 1966; McKenzie andTibbo, 1964; Dragonik and Pelzarski. 1983; Stevens, 1984; Francis et al,, 2001 ). Although the largest blue shark reported from the North Pacific was only 254 cm FL (Stras- burg, 1958; Cailliet et al., 1983), individuals up to 331 cm FL have been reported from the South Pa- cific and the largest sharks were all males (Francis et al., 2001). The paucity of females exceeding 225 cm FL in the current study and the complete lack of these specimens in the Stevens (1975), Silva et al. (1996), and Henderson et al. (2001) samples indicate that these fish are rare, inhabit unknown or unfished areas of the Atlantic, or possibly avoid fishing gear In our study, the VBGF parameters (Table 5) show that females theoretically attain larger sizes than males. However, the low number of large females in this and previous studies may indicate that natural mortality prevents them from attaining these lengths. The occurrence of severe lacerations on female blue sharks incurred during courtship is well documented (Stevens, 1974; Pratt, 1979). Although highly speculative, the long-term cumulative effects of such behavior may act as a source of increased mortality in females of the species, shortening their life-span and limiting the number that reach the larger sizes. Through an integrated approach incorporating vertebral banding, OTC injection, and tagging data, it has been shown that the blue shark grows faster and lives a shorter life than previously thought in the North Atlantic. We believe that the validated vertebral interpretations generated during this study for the first four years of growth, combined with the vertebral counts and longevity estimates from tag-recapture data, provide vigor- ous estimates of age and growth for a large pelagic carcharhinid, the blue shark. Acknowledgments We thank the many people who contributed to the success of this study on research vessels, at recreational fishing tournaments, and on board private, chartered, and commercial fishing vessels. This study would not have been possible without the staff of the NMFS Apex Predators Program (Narragansett, RI) including Nancy Kohler, Pat Turner, and Ruth Briggs. We especially thank retired NMFS researchers Jack Casey and H. Wes Pratt for giving the senior author the tools necessary to initiate and complete this work. We are grateful to shark-aging pioneer Gregor Cailliet for his moral support and relentless pursuit of this publication. We are indebted to the thousands of fishermen who voluntarily tagged and returned sharks for the NMFS Cooperative Shark Tagging Program. This study was partially funded with support from the Sportfish Restoration Act. 300 A Tag/recapture-GROTAG Tag/recapture-Gultand method Vertebrae-current study Aasen (1966)-LF modes ' Henderson el al (2001) Silva etal. (1996) 0 2 4 6 8 10 12 14 16 18 B Vertebrae-current study / Silva etal (1996) 12 14 16 18 250 c ^__^ ^.^ Vertebrae-current study OTC y^ y^ ■/■ Stevens (1975) /^ / y Silva etal (1996) // / 0 2 4 6 8 10 12 14 16 Age (years) Figure 8 Von Bertalanfly growth curves generated from vertebral and recap- ture data for (A) sexes combined, (B) male, and (C) female Prionace glauca, as compared to OTC-injected recaptured blue sharks; included for comparison are the von Bertalanffy growth curves of other North Atlantic studies. 638 Fishery Bulletin 101 (3) Literature cited Aasen, O. 1966. Blahaien, Prionace^Zauca (Linnaeus, 1758). Fisken ogHavet 1:1-15. Bigelow, H.. B., and W. C. Schroeder. 1953. Fishes oftheGulfofMaine, 577 p. U.S. Dep. Int., Fish and Wildl. Serv., Fish. Bull. 53. Cailliet, G. M. 1990. Elasmobranch age determination and verification: an updated review. In Elasmobranchs as living resources: advances in the biology, ecology, systematics, and status of the fisheries (H. L. Pratt Jr., S. H. Gruber, and T. Taniuchi, eds.), p. 157-165. U.S. Dep. Commer., NOAA Tech. Rep. 90. Cailliet, G. M., L. K. Martin, J. T. Harvey, D. Kusher, and B. A. Welden. 1983. Preliminary studies on the age and growth of blue, Prionace glauca, common thresher, Alopias vulpinus, and shortfin mako, Isurus oxyrinchus, sharks from California waters. In Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, billfishes. and sharks (E. D. Prince and L. M. Pulos, eds.), p. 179-188. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 8. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Campana, S. E., M. C. Annand, and J. I. McMillan. 1995. Graphical and statistical methods for determining the consistency of age determinations. Trans. Amer. Fish. See. 124:131-138. Carey, F. G., and J. Scharold. 1990. Movements of blue sharks (Prionace glauca) in depth and course. Mar Biol. 106:329-342. Casey, J. G. 1982. Blue shark, Prionace glauca. Species synopsis. In Ecology of the Middle Atlantic Bight fish and shellfish — Monograph 15 — fish distribution (M.D. Grosslein and T. Azarovitz, eds.), p. 45-^8. MESA New York Bight Atlas, NY Sea Grant, Albany, NY. 1985. Trans-Atlantic migrations of the blue shark: a case history of cooperative shark tagging. In World angling resources and challenges; proceedings of the first world angling conference (R. H. Stroud, ed.), p. 253-267. Int. Game Fish Assoc, Ft. Lauderdale, FL. Casey, J. G., H. L. Pratt Jn, and C. E. Stillwell. 1985. Age and growth of the sandbar shark (Carcharhinus plumbeus) from the western North Atlantic. Can. J. Fish. Aquat. Sci. 42(5):963-975. Castro, J. L, C. M. Woodley, and R. L. Brudek. 1999. A preliminary evaluation of the status of shark species. FAG Fisheries Technical Paper 380, 72 p. FAG, Rome. Compagno, L. J. V. 1984. FAG species catalogue. Sharks of the world. An anno- tated and illustrated catalogue of shark species known to date. Part 1. Hexanchiformes to Lamniformes. FAO Fish Synop. 125, vol. 4, 250 p. FAO, Rome. Connett, S. 1987. Blue sharks studied in the eastern Atlantic. In The shark tagger: 1987 summary (J. Casey, H. L. Pratt Jr, N. E. Kohler, and C. E. Stillwell, eds. ), p. 8-10. Newsletter of the Cooperative Shark Tagging Program. U.S. Dep. Commer., NOAA, NMFS, Narragansett, RL Cortes, E. 2000. Life history patterns and correlations in sharks. Rev. Fish. Sci. 8(4):299-344. 2002. Catches and catch rates of pelagic sharks from the northwestern Atlantic, Gulf of Mexico, and Caribbean. Col. Vol. Sci. Pap. ICCAT 54(4): 1164-1 181. Dragonik, B, and W. Pelzarski. 1983. The occurrence of the blue shark, Prionace glauca (L.), in the North Atlantic. Rep. Sea Fish. Inst. 19:63-77. Fabens. A. J. 1965. Properties and fitting of the von BertalanfTy growth curve. Growth 29:265-289. Ferreira, B. P., and C. M. Vooren. 1988a. Age, growth, and structure of the vertebrae in the school shark Galeorhinus galeus (Linnaeus, 1758) from southern Brazil. Fish. Bull. 89:19-31. Francis, R. I. C. C. 1988a. Maximum likelihood estimation of growth and growth variability from tagging data. NZ J. Mar Fresh- water Res. 22:43-51. 1988b. Are growth parameters estimated from tagging and age-length data comparable? Can. J. Fish. Aquat. Sci. 45: 936-942. Francis, M. P., L. H. Griggs, and S. J. Baird. 2001. Pelagic shark by catch in the New Zealand tvma long- line fishery. Mar. Freshwater Res. 52(2):165-178. Gubanov, Ye. P., and V. N. Grigor'yev. 1975. Observations on the distribution and biology of the blue shark Prionace glauca (Carcharhinidae) of the Indian Ocean. J. Ichthyol. 15:37-43. Gulland, J. A., and S. J. Holt. 1959. Estimation of growth parameters for data at unequal time intervals. J. Cons. Int. Explor Mer 25:47-49. Henderson, A. C, K. Flannery, and J. Dunne. 2001. Observations on the biology and ecology of the blue shark in the North-east Atlantic. J. Fish. Biol. 58: 1347-1358. Hoenig, J. M., and S. H. Gruber. 1990. Life history patterns in the elasmobranchs: implica- tions for fisheries management. In Elasmobranchs as living resources: advances in the biology, ecology, systemat- ics, and status of the fisheries (H. L. Pratt Jr, S. H. Gruber, and T Taniuchi, eds), p. 1-16. U.S. Dep. Commer, NOAA Tech. Rep. 90. ICCAT (International Commission for the Conservation of Atlantic Tunas). 2002. ICCAT data preparatory meeting for Atlantic shark stock assessment. Col. Vol. Sci. Pap. ICCAT, 54(4): 1064-1106. Kohler, N. E. 1987. Aspects of the feeding ecology of the blue shark in the western North Atlantic. Ph.D. diss., 163 p. Univ Rhode Island, Kingston, RI. Kohler, N. E., J. G. Casey, and P A. Turner 1987. Length-weight relationships for 13 species of sharks from the western North Atlantic. Fish. Bull. 93:412-418. Kohler, N. E., P. A. Turner, J. J. Hoey, L. J. Natanson, and R. Briggs. 2002. Tag and recapture data for three pelagic shark species: blue shark (Prionace glauca ), shortfin mako (Isurus oxyrin- chus), and porbeagle (Lamna nasus) in the North Atlantic Ocean. Col. Vol. Sci. Pap. ICCAT 54(4):1231-1260. Lessa, R., F M. Santana, and R. Paglerani. 1999. Age, growth and stock structure of the oceanic whitetip shark, Carcharhinus longnnanus. from the southwestern equatorial Atlantic. Fish. Res. 42:21-30. Skomal and Natanson: Age and growth of Phonace glauca 639 McKenzie, R. A., and S. N. Tibbo. 1964. A morphometric description of the blue shark i.Prio- nace glauca ) from the Canadian Atlantic waters. J. Fish. Res. Board Can. 21:865-866. Nakano, H. 1994. Age. reproduction and migration of blue shark in the North Pacific. Bull. Nat. Res. Inst. Far Seas Fish. 31:141-256. Natanson, L. J. 1984. Aspects of age, growth, and reproduction of the Pacific angel shark, Squatina californica, off Santa Barbara, California. M.A. thesis, 71 p. San Jose State Univ., San Jose, CA. Natanson, L. J., J. G. Casey, and N. E. Kohler. 1995. Age and growth of the dusky shark, Carcharhinus obscunis, in the western North Atlantic. Fish. Bull. 93: 116-126. Natanson, L. J., and N. E. Kohler 1996. A preliminary estimate of age and growth of the dusky shark Carcharhinus obscurus from the south-west Indian Ocean, with comparisons to the western North Atlantic. S. Afr. J. Mar. Sci. 17:217-224. Natanson, L. J., J. J. Mello, and S. E. Campana. 2002. Validated age and growth of the porbeagle shark, Lamna nasus, in the western North Atlantic. Fish. Bull. 100:266-278. NMFS (National Marine Fisheries Service). 1999. Final fishery management plan for Atlantic tunas, swordfish, and sharks., 1162 p. Highly Migratory Species Management Division, Silver Spring, MD. Parker, H. W., and F C. Stott. 1965. Age, size and vertebral calcification in the basking shark, Cctorhinus maximus (Gunnerus). Zool. Meded. 40(34);305-319. Pratt, H. L., Jr. 1979. Reproduction in the blue shark, Prionace glauca. Fish. Bull. 77:445-470. Pratt, H. L., Jr., and J. G. Casey 1983. Age and growth of the shortfin mako, Isurus oxyrin- chus. using four methods. Can. J. Fish. Aquat. Sci. 40(11): 1944-1957. 1990. Shark reproductive strategies as limiting factors in directed fisheries, with a review of Holden's method of esti- mating growth parameters. In Elasmobranchs as living resources: advances in the biology, ecology, systematics, and status of the fisheries (H. L. Pratt Jr., S. H. Gruber, and T. Taniuchi, eds), p. 97-110. NOAATech. Rep. 90. Sciarrotta, T. C, and D. Nelson. 1977. Diel behavior of the blue shark, Prionace glauca, near Santa Catalina Island, California. Fish. Bull 73:519-528. Seki, T., T. Taniuchi, H. Nakano, and M. Shimizu. 1998. Age, growth and reproduction of the oceanic whitetip shark from the Pacific Ocean. Fish. Sci. 64(1): 14-20. Silva, A. A., H. M. Silva, and K. Erzini. 1996. Some results on the biology of the blue shark, Prio- nace glauca. in the North Atlantic based on data from a re- search cruise of the R/V Arquipelago in Azorean waters: a summary paper, 9 p. Universidade dos Acores, Horta, Acores, Portugal. Simpfendorfer, C. A. 2000. Growth rates of juvenile dusky sharks, Carcharhinus obscurus (Lesueur, 1818), from southwestern Australia esti- mated from tag-recapture data. Fish. Bull. 98:811-822. Simpfendorfer, C. A., R. B. McAuley, J. Chidlow, and P. Unsworth. 2002. Validated age and growth of the dusky shark, Carcha- rhinus obscurus, from Western Australian waters. Mar. Freshwater Res. 53:567-573. Skomal, G. B. 1990. Age and growth of the blue shark, Prionace glauca, in the North Atlantic. M.S. thesis, 82 p. Univ. Rhode Island, Kingston, RI. Stevens, J. D. 1973. Stomach contents of the blue shark (Prionace glauca L.) of southwest England. J. Mar Biol. Assoc. U.K. 53: 357-361. 1974. The occurrence and significance of tooth cuts on the blue shark (Prionace glauca L. ). J. Mar Biol. Assoc. U.K. 54:373-378. 1975. Vertebral rings as a means of age determination in the blue shark (Prionace glauca L.). J. Mar. Biol. Assoc. U.K. 55:657-665. 1976. First results of shark tagging in the northeast Atlan- tic, 1972-1975. J. Mar. Biol. Assoc. U.K. 56: 929-937. 1984. Biological observations on sharks caught by sport- fishermen off New South Wales. Aust. J. Mar. Freshwater Res. 35:573-590. 1990. Further results from a tagging study of pelagic sharks in the Northeast Atlantic. J. Mar Biol. Assoc. U.K. 70: 707-720. Strasburg, D. W. 1958. Distribution, abundance, and habits of pelagic sharks in the central Pacific Ocean. Fish. Bull. 58:335-361. Suda, A. 1953. Ecological study of the blue shark (Prionace glauca Linne'). South Sea Area Fish Res. Lab. Rep. 26:1-11. Tanaka, S. 1984. Present status of fisheries biology, /n Elasmobranchs as fishery resources (T. Taniuchi and M. Suyama, eds.), p. 46-59. Jpn. Soc. Sci. Fish., Fish. Ser. 49. Tanaka, S., G. M. Cailliet, and K. G. Yudin. 1990. Differences in growth of the blue shark, Prionace glauca: technique or population? In Elasmobranchs as living resources: advances in the biology, ecology, systemat- ics, and status of the fisheries (H. L. Pratt Jr., S. H. Gruber, and T. Taniuchi, eds.), p. 177-187. U.S. Dep. Commer., NOAA Tech. Rep. 90. Taylor, C. C. 1958. Cod growth and temperature. J. Cons. Int. Explor Mer 23:366-370. Tricas, T. 1978. Relationships of the blue shark, Prionace glauca, and its prey species near Santa Catalina Island, California. Fish. Bull. 77:175-182. Tucker, D. W., and C. T. Newnham. 1957. The blue shark Prionace glauca breeds in British seas. Ann. Mag. Nat. Hist., Series 12, 10:673-688. von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquiries on growth laws II). Hum. Biol. 10:181-213. Walter, J. P, and D. A. Ebert. 1991. Preliminary estimates of age of the bronze whaler Carcharhinus brachyurus (Chondrichthyes: Carcharhini- dae) from southern Africa, with a review of some life history parameters. S. Afr. J. Mar. Sci 10:37-44. Wintner,S. P, and G. Cliff. 1995. Age and growth determination of the blacktip shark, Carcharhinus lintbatus, from the east coast of South Africa. Fish. Bull. 94:135-144. Wintner, S. P, and S. F J. Dudley 2000. Age and growth estimates for the tiger shark, Galeo- cerdo cuvier, from the east coast of Africa. Mar Freshwa- ter Res. 51:43-53. 640 Abstract — Little is known about the ocean distributions of wild juvenile coho salmon off the Oregon- Washington coast. In this study we report tag recov- eries and genetic mixed-stock estimates of juvenile fish caught in coastal waters near the Columbia River plume. To sup- port the genetic estimates, we report an allozyme-frequency baseline for 89 wild and hatchery-reared coho salmon spawning populations, extending from northern California to southern Brit- ish Columbia. The products of 59 allo- zyme-encoding loci were examined with starch-gel electrophoresis. Of these, 56 loci were polymorphic, and 29 loci had ^0 9.5 levels of polymorphism. Average heterozygosities within populations ranged from 0.021 to 0.046 and aver- aged 0.033. Multidimensional scaling of chord genetic distances between sam- ples resolved nine regional groups that were sufficiently distinct for genetic mixed-stock analysis. About 2.9% of the total gene diversity was due to differ- ences among populations within these regions, and 2.6% was due to differences among the nine regions. This allele-fre- quency data base was used to estimate the stock proportions of 730 juvenile coho salmon in offshore samples col- lected from central Oregon to northern Washington in June and September- October 1998-2000. Genetic mixed- stock analysis, together with recoveries of tagged or fin-clipped fish, indicates that about one half of the juveniles came from Columbia River hatcheries. Only 22% of the ocean-caught juveniles were wild fish, originating largely from coastal Oregon and Washington rivers (about 20%). Unlike previous studies of tagged juveniles, both tag recoveries and genetic estimates indicate the pres- ence offish from British Columbia and Puget Sound in southern waters. The most salient feature of genetic mixed stock estimates was the paucity of wild juveniles from natural populations in the Columbia River Basin. This result reflects the large decrease in the abun- dances of these populations in the last few decades. Genetic analysis of juvenile coho salmon iOncorhynchus kisutch) off Oregon and Washington reveals few Columbia River wild fish' David J. Teel Donald tA. Van Doornik David R. Kuligowsl a. 88 1 89 La Push v/ ®^ . 83 84 82 63 ^^"^■' +++++ 59-62 64 '^^-'^* 57-58 65 70-72 -46° Location of coastal pelagic trawl surveys 66-68 69 +++++:■- . *-5 C +++++ .53-56 +++++ 50-52 Washington ^''-^^ 38-39 +++++ 32-33 , , , , ^, 40-42 Columbia **^^ 31 43-49 ++++ 30 ++++ 26-29 Cape Perpetua 23-25 ^^^ 21-22 -42° 17-20 11-16 Oregon \ V • Rogue 5 — £■ River . ■/ ....... ^ 6 ' Xallfornia -38° N i 1 km 0 100 200 Figure 1 Locations of ocean sampling transect lines ( + ) and 89 coho salmon populations in California, Oregon, Washington, and British Colum- bia. Numbers correspond to population names in Table 1. Genotypic frequencies of polymorphic loci for each base- line sample were examined for departures from expected Hardy-Weinberg proportions with a Fisher's exact test (Guo and Thompson, 1992) by using GENEPOP version 3.1 (Raymond and Rousset, 1995). Hardy-Weinberg tests were performed on isoloci (comigrating protein products of duplicated loci) following Waples (1988). We estimated allelic frequencies for each sample. Allelic frequencies for isoloci were calculated as mean frequen- 642 Fishery Bulletin 101(3) Table 1 Sample information and indices of genetic variability for coho salmon from the Pacific Northwest and Cahfomia. Map codes refer | to Figure 1. Indices of genetic variability are '^tP^g^ = percentage of Pog^ loci and H = heterozygosity. Source Number of Region and map code Year sampled fish '-'''^95 H California coast 1 Scott Creek 1994 21 12.5 0.039 2 Little River 1994 27 14.3 0.040 3 Warm Springs Hatchery 1994, 1994 160 16.1 0.041 4 Mad River Hatchery 1994 120 17.9 0.040 Klamath River to Cape Blanco 5 Iron Gate Hatchery 1994 120 9.0 0.021 6 Trinity Hatchery 1984, 1994 218 9.0 0.028 7 Rogue River (Illinois River, Greyback Creek) 1993 40 7.2 0.022 8 Cole Rivers Hatchery, stock no. 52 (Rogue River) 1993 100 9.0 0.030 9 North Fork Elk and Elk Rivers 1993 32 7.2 0.021 Oregon coast 10 Sixes River (Crystal and Edson Creeks) 1993 44 7.2 0.026 11 New River (Bether and Morton Creeks) 1993 62 10.7 0.034 12 Butte Falls Hatchery, stock no. 44 (Coquille River) 1993 100 9.0 0.036 13 Cole Rivers Hatchery, stock no. 37 (South Fork Coos River) 1993 129 10.7 0.034 14 Coos River (MilHcoma River and Marlow Creek) 1993, 1997 50 12.5 0.033 15 Butte Falls Hatchery, Eel River stock no. 63 1993 100 7.2 0.032 16 Ten Mile Lake 1992 56 7.2 0.030 17 Rock Creek Hatchery, stock no. 55 (Umpqua River) 1993 100 7.2 0.029 18 North Umpqua River (Williams Creek) 1993, 1997 67 7.2 0.025 19 Butte Falls Hatchery, stock no. 18 ( Umpqua River) 1993 100 7.2 0.027 20 Smith River (Halfway Creek) 1993 40 10.7 0.034 21 Fall Creek Hatchery, stock no. 113 (Talikenitch River) 1993 100 7.2 0.030 22 Siuslaw River 1996 51 9.0 0.029 23 Fall Creek Hatchery, stock no. 31 (Alsea River) 1993 100 14.3 0.040 24 Fall Creek Hatchery, stock no. 43 (Alsea River) 1993 95 9.0 0.037 25 Alsea River 1996 62 10.7 0.031 26 Beaver Creek 1993 62 9.0 0.035 27 Yaquina River 1996 54 12.5 0.043 28 Salmon River Hatchery, stock no. 33 (Siletz River) 1993 100 12.5 0.041 29 Siletz River (Forth of July, Sunshine, and Buck Creeks) 1993 50 10.7 0.033 30 Salmon River Hatchery, stock no. 36 (Salmon River) 1993 100 10.7 0.037 31 Trask River Hatchery, stock no. 34 (Trask River) 1992, 1993 220 16.1 0.039 32 Nehalem River Hatchery, stock no. 99 (Nehalem River) 1992 80 12.5 0.045 33 Nehalem River Hatchery, stock no. 32 (Nehalem River) 1993 100 14.3 0.044 Columbia River 34 Lewis and Clark River 1991, 1993 36 12.5 0.038 35 Big Creek Hatchery 1991 80 12.5 0.040 36 Grays River Hatchery 1987, 1991 200 7.2 0.033 37 Clatskanie River (Carcus Creek) 1991, 1992, 1996 113 10.7 0.033 38 Cowlitz Hatchery early-run 1991 80 9.0 0.027 39 Cowlitz Hatchery late-run 1991, 1992 180 7.2 0.031 40 Scappoose River (Siercks, Raymond, and Milton Creeks) 1991 44 14.3 0.041 41 Lewis River Hatchery early-run 1991 80 5.4 0.027 42 Lewis River Hatchery late-run 1991 80 12.5 0.032 43 North Fork Clackamas River early-run 1998« 48 16.1 0.036 44 North Fork Clackamas River late-run 1999» 45 14.3 0.028 45 Eagle Creek Hatchery 1991, 1992 180 7.2 0.037 continued Teel et al.: Genetic analysis of juvenile Oncorhynchus kisutch 643 Table 1 (continued) Source Number of Region and map code Year samplet fish ■^"^0.95 H 46 Sandy River Hatchery 1991, 1992 180 10.7 0.046 47 Sandy River 1991, 1992, 1996 124 10.7 0.043 48 Bonneville Hatchery 1991, 1992 180 10.7 0.043 49 Willard Hatchery 1991 80 7.2 0.032 South Washington coast 50 Naselle River Hatchery 1991 100 9.0 0.029 51 Nemah River Hatchery 1991 100 10.7 0.029 52 Willapa River Hatchery 1991 100 9.0 0.031 53 Chehalis River (Stillman Creek) 1995 71 9.0 0.026 54 Chehalis River (Satsop River, Bingham Creek) 1995 98 10.7 0.028 55 Bingham Creek Hatchery 1991,' 1992,' 1995 180 9.0 0.027 56 Chehalis River (Hope Creek) 1994, 1995, 1996 171 9.0 0.030 North Washington coast 57 Queets River 1995 99 9.0 0.028 58 Clearwater River 1995 100 7.2 0.029 59 Bogachiel River 1987 80 10.7 0.030 60 Sol Due Hatchery Summer Run 1994' 80 7.2 0.030 61 Sol Due River Summer Run 1995 120 10.7 0.030 62 Sol Due Hatchery Fall Run 1995' 80 9.0 0.032 63 Hoko River 1987 96 9.0 0.033 Puget Sound and Hood Canal 64 Dungeness Hatchery 1987 80 12.5 0.037 65 Quilcene Hatchery 1994' 100 9.0 0.025 66 North Fork Skokomish River 1994,' 1995' 126 7.2 0.030 67 Dewatto River 1994,' 1995,' 1996' 169 9.0 0.028 68 Minter Creek Hatchery 1992,' 1995' 80 9.0 0.035 69 Soos Creek Hatchery 1994,' 1995, 1996 680 9.0 0.034 70 Snoqualmie River (Harris Creek) 1987 120 7.2 0.034 71 Snoqualmie River (Grizzly Creek) 1994,' 1995,' 1996' 215 7.2 0.030 72 North Fork Skykomish River (Lewis Creek) 1995' 102 9.0 0.032 73 North Fork Stillaguamish River (Fortson Creek) 1987, 1989 ' 200 9.0 0.031 74 North Fork Stillaguamish River (Mcgovem Creek) 1987 40 10.7 0.032 75 Upper Skagit River 1993 127 9.0 0.033 76 Skagit River (Carpenter Creek) 1993 139 9.0 0.032 77 Skagit River (West Fork Nookachamps Creek) 1987, 1993 220 9.0 0.035 78 Skagit River (Baker River) 1992' 303 10.7 0.036 79 Skagit River (Suiattle River, All Creek) 1987, 1993 200 10.7 0.032 80 Skagit River (Upper Sauk River) 1992, 1993 200 9.0 0.034 81 Skagit River (Upper Cascade River) 1992. 1993 224 9.0 0.031 82 Samish River (Ennis Creek) 1994, ' 1995, ' 1996 ' 167 9.0 0.035 British Columbia 83 Chilliwack River Hatchery 1984 100 10.7 0.034 84 Cowichan River Hatchery 1984 80 9.0 0.036 85 Big Qualicum Hatchery 1989,' 1991 180 10.7 0.037 86 Robertson Creek Hatchery 1984 100 9.0 0.030 87 Capilano Hatchery 1989, ' 1991 200 12.5 0.038 88 Squamish River Hatchery 1988' 98 7.2 0.035 Upper Fraser River 89 Spius River Hatchery 1987 200 10.7 0.035 Meai 10.0 0.033 ' Sample taken from adult fish. All other samples were from juvenile fish. 644 Fishery Bulletin 101 (3) Table 2 Enzymes and study results for 59 loci in samples of 89 coho salmon populations from the Pacific Northwest and California. Number of Range of Enzyme or Enzyme commission Locus populations common allele protein name number abbrev. polymorphic frequency Aspartate aminotransferase 2.6.1.1 sAAT-1,2* 36 1.000-0.966 sAAT-3* 1 1.000-0.956 sAAT-4* 71 1.000-0.839 Adenosine deaminase 3.5.4.4 ADA-1* 34 1.000-0.924 ADA-2* 15 1.000-0.929 Aconitate hydratase 4.2.1.3 mAH-1* 2 1.000-0.992 mAH-2* 22 1.000-0.919 mAH-3* 3 1.000-0.944 sAH* 60 1.000-0.849 Adenylate kinase 2.7.4.3 AK* 4 1.000-0.993 Alanine aminotransferase 2.6.1.2 ALAT* 12 1.000-0.958 Creatine kinase 2.7.3.2 CK-Al* 8 1.000-0.971 CK-A2* 22 1.000-0.919 CK-Cl* 4 1.000-0.983 CK-C2* 9 1.000-0.972 CK-B* 1 1.000-0.999 Esterase 3.1.-.- EST-1* 85 1.000-0.652 Fructose-bisphosphate aldolase 4.2.1.13 FBALD-3* 1 1.000-0.996 FBALD-4* 14 1.000-0.962 Formaldehyde dehydrogenase (glutathione) 1.2.1.1 FDHG* 33 1.000-0.954 Fumarate hydratase 4.2.1.2 FH* 43 1.000-0.835 b-N-Acetylgalactosaminidase 3.2.1.53 bGALA* 89 0.889-0.357 Glyceraldehyde-3-phosphate dehydrogenase 1.2.1.12 GAPDH-2* 64 1.000-0.713 GAPDH-3* 26 1.000-0.867 GAPDH-4* 9 1.000-0.975 GAPDH-5* 0 1.000-1.000 Glucose-6-phosphate isomerase 5.3.1.9 GPI-A* 17 1.000-0.906 GPI-Bl* 1 1.000-0.962 GPI-B2* 47 1.000-0.815 (ilutathione reductase 1.6.4.2 GR* 6 1.000-0.988 continued cies over both loci and treated as a single tetrasomic locus. Following the recommendations of Waples (1990), allelic frequencies of samples taken in different years from the same location were combined. In general, little temporal allele-frequency variation was detected in coho salmon populations sampled over years (Van Doornik et a!., 2002; present study). Levels and patterns of genetic variation within and between populations were estimated with 56 polymorphic loci (Table 2). Average expected heterozygos- ity per locus (isoloci excluded) for each population was calculated by using an unbiased estimator (Nei, 1978). The proportion of P,, (,r, loci was computed for each popula- tion, in which a locus was considered to be polymorphic if the frequency of the most common allele was sO.95. Chord distances (Cavalli-Sforza and Edwards, 1967) were computed between all pairs of populations with BIOSYS (Swofibrd and Selander, 1981), and relationships among populations were depicted with multidimensional scaling (MDS, NTSYS-PC, Exeter Software, NY). Allele-frequency variation among baseline populations was partitioned (Chakraborty et al., 19821 into two geographic levels: 1) populations within regions; and 2) among regions (Table 1). These regions were delimited by geography and by genetic groupings in the MDS analyses. We used the maximum likelihood procedures of Pella and Milner (1987) and the Statistical Package for Analyzing Mixtures (SPAM; Debevec et al., 2000) to estimate stock contributions to simulated and actual mixtures of coho salmon. Estimates were made by using 56 polymorphic loci (Table 2) and 89 baseline populations, except for analysis of marked (hatchery) fish where only hatchery populations were used (Table 1). Allocations to individual baseline populations were then summed to estimate contributions of regional stock groups (Pella and Milner, 1987). Average mix- Teel et al : Genetic analysis of juvenile Oncorhynchus kisutch 645 Table 2 (continued) Number of Range of Enzyme or Enzjone commission Locus populations common allele protein name number abbrev. polymorphic frequency Isocitrate dehydrogenase 1.1.1.42 mIDHP-1* 4 1.000-0.964 mIDHP-2* 11 1.000-0.799 sIDHP-1* 11 1.000-0.948 sIDHP-2* 29 1.000-0.851 Lactate dehydrogenase 1.1.1.27 LDH-Al* 7 1.000-0.700 LDH-A2* 2 1.000-0.995 LDH-Bl* 18 1.000-0.942 LDH-B2* 20 1.000-0.956 LDH-C* 0 1.000-1.000 Malate dehydrogenase 1.1.1.37 sMDH-Al,2* 35 1.000-0.976 sMDH-Bl,2* 21 1.000-0.947 Mannose-6-phosphate isomerase 5.3.1.8 MPI* 41 1.000-0.897 o-Mannosidase 3.2.1.24 MAN* 5 1.000-0.981 Dipeptidase 3.4.-.- PEPA* 63 1.000-0.895 Tripeptide amino peptidase 3.4.-.- PEPB-1* 10 1.000-0.979 Peptidase-C 3.4.-.- PEPC* 89 0.903-0.391 Proline dipeptidase 3.4.-.- PEPD-2* 56 1.000-0.798 Leucyl-L-tyrosine peptidase 3.4.-.- PEPLT* 19 1.000-0.953 Phosphogluconate dehydrogenase 1.1.1.44 PGDH* 7 1.000-0.967 Phosphoglycerate kinase 2.7.2.3 PGK-1* 14 1.000-0.930 PGK-2* 13 1.000-0.975 Phosphoglucomutase 5.4.2.2 PGM-1* 72 1.000-0.600 PGM-2* 32 1.000-0.958 Purine-nucleoside phosphorylase 2.4.2.1 PNP-1* 87 1.000-0.614 Pyruvate kinase 2.7.1.40 PK-2* 14 1.000-0.980 Triose-phosphate isomerase 5.3.1.1 TPI-1* 5 1.000-0.986 TPI-2* 0 1.000-1.000 TPI-3* 27 1.000-0.930 TPI-4* 2 1.000-0.994 ture estimates derived from 100 simulated mixtures were used to evaluate the accuracy of estimated contributions to each region with mixture sizes of 100, 300, and 500 fish. We analyzed mixtures composed of fish entirely from each region and also mixtures that excluded fish from regions south and north of our marine sampling area. Precisions of the stock composition estimates for the actual mixtures were estimated by bootstrapping baseline and mixture genetic data 100 times as described in Pella and Milner ( 1987). Stock compositions were estimated for June and September-October. We also combined samples over sur- veys and made separate estimates from samples of marked ( fin-clipped and tagged hatchery fish ) and unmarked fish to examine hatchery and wild stock compositions. However, because not all hatchery fish are marked, unmarked fish are a mixture of wild and hatchery fish. We therefore es- timated the proportion of hatchery fish for a region in the sample of unmarked fish (Pyf^) by where P MH RvIRm ^U^^M = the proportion of hatchery fish from a par- ticular region in the sample of marked fish; = the ratio of unmarked to marked releases in a region; and = the ratio of unmarked to marked fish in our ocean samples. The R,,/Rf^f for 1997 and 1998 brood years varied consider- ably among regions; California coast 1.0, Klamath River to Cape Blanco 0.01, Oregon coast 0.12, Columbia River 0.12, southern Washington coast 0.03, northern Washington coast 0.69, Puget Sound 0.43, southern British Columbia 0.09, and Upper Eraser River 0.80 (Lavoy'; PSMFC^). We PuH =(Pmh(Ru'Rm^)/(S„/SJ, (1) Lavoy, L. 2001. Personal commun. Washington Department of Fish and Wildlife, Olympia, WA. 98501. ■ PSMFC (Pacific States Marine Fisheries Commission). 2001. Regional Mark Information System (RMIS) coded-wire tag on-line database. [Available from Pacific States Marine Fisheries Commission, 45 SE 82°'' Dr., Suite 100, Gladstone, OR 97027-2522.1 646 Fishery Bulletin 101(3) then subtracted P^,^ for each region from the genetic esti- mate of the region's contribution to the sample of unmarked fish. The sum of the remaining values estimated the pro- portion of wild fish in the sample of unmarked fish. When Pijfj for a region was greater than the genetic estimate of the region's contribution to the sample of unmarked fish, the percentage of wild fish from that region was considered to be zero. We estimated regional proportions of hatchery and wild coho salmon in the all-fish marine sample that included both marked and unmarked coho salmon. Regional hatch- ery contributions to the all-fish sample were made by sum- ming each region's estimated contribution to the sampled marked and unmarked fish, weighted by the proportion of each of these sample types in the total sample. Regional proportions of wild coho salmon in the all-fish sample were made by multiplying a region's estimated proportion of wild coho salmon in the unmarked sample by the propor- tion of unmarked fish in the total sample. Results Baseline genetic data and population structure Although coho salmon generally have low levels of genetic variability in relation to other Pacific salmon, a sufficient number of polymorphic loci were detected to distinguish many populations and regional population groups. Of 59 loci screened in all 89 populations, 56 were polymorphic, and 29 of these were at the Pq 95 level of polymorphism in at least one population (Table 2). Allelic frequencies are reported in an appendix that can be retrieved at the Northwest Fisheries Science Center website [http: //www.nwfsc. noaa.gov]. Twenty of the 56 polymorphic loci had two alleles per locus, 24 had three alleles per locus, nine had four alleles, two had five alleles, and one had six alleles. Two loci (BGALA* and PEPC* ) varied in all popula- tions studied. Three loci (GAPDH-5*, LDH-C*, and TPI-2*) were monomorphic in all populations. Observed genotypic proportions for polymorphic loci in 128 samples departed significantly (P<0.05) from expected Hardy- Weinberg pro- portions in 75 of 1476 tests (5.1% ). There were no consis- tent trends by population or locus. Because the number of significant tests is close to the number expected by chance for this rejection level, we did not attach any biological significance to these departures. The percentages of Ppgr^ loci and average heterozygosi- ties over 56 loci for each population appear in Table 1. The percentage of P^^^ loci ranged from only 5.4% in Lewis River hatchery early run (population 41) to 17.9% in the Mad River hatchery (4). Average heterozygosities ranged from 0.021 in Iron Gate hatchery (5) and Elk River (9) to 0.046 in Sandy River hatchery (46 ). Gene diversity analysis of the 89 populations resulted in a total gene diversity (//•,.) of 0.035 and an average sample diversity (H^) of 0.033. Thus, 94.5% of the total genetic diversity was attributable to within-sample variability and 5.5% was attributable to variability among samples. About 2.9% of the total gene diversity was due to variability among populations within regions, and 2.6% was due to variability among the nine regions. Genetic relationships among populations of coho salmon as revealed by two-dimensional MDS analysis showed that genetic differences among populations were geographically structured (Fig. 2). The first axis in the plot separated pop- ulations in coastal Oregon and California from northern populations. Several populations, including two from the Rogue River in southern Oregon (numbers 7 and 8) and Big Qualicum hatchery (85) on Vancouver Island, were po- sitioned near the convergence of the southern and northern population groups. The Iron Gate hatchery sample (5) from the Klamath River, California, clustered with the northern population group. Several genetically discrete groups ap- peared on smaller geographical scales. However, samples from Iron Gate hatchery (5), Yaquina River (27), Nehalem hatchery (33), Willapa Bay area (50, 51, and 52), Dungeness hatchery (64), McGovern Creek (74), upper Cascade River (81), and Ennis Creek (82) did not cluster with nearby populations. The single population in our study from the upper Fraser River region — Spius hatchery (89) of the Thompson River-was the most genetically distinct in the MDS analysis (.r=-2.3,y=-0.9) and was positioned beyond the scaling shown in Figure 2. The Little River (2) popula- tion also fell outside the area of the plot {.v=5.1,y=2.0), but was genetically most similar to other California coastal populations (1, 3, and 4). Genetic estimates of simulated stock mixtures One demonstration of discreteness among regional groups is the correct allocation in a mixed-stock analysis of simu- lated samples from baseline populations to their stock of origin. We used simulated sample sizes of 100, 300, and 500 taken from one region at a time; therefore the results represent the accuracy of reallocation back to the region of origin. Table 3 presents the average values of 100 bootstrap resamplings of both the baseline and the mixture samples. For simulated sample sizes of 100, reallocation accuracy ranged from 81% (coastal northern Washington) to 98% (upper Fraser River population) and averaged 88.7% over the nine regions. Average accuracy increased to 92.9% with an increase in the size of the simulated sample to 300. Only marginal improvement (93.6%r accuracy) was achieved by increasing the simulated sample size to 500. We also used mixed-stock analysis of simulated samples to examine the accuracy of composition estimates for Cali- fornia, Puget Sound, and British Columbia regions when fish from these areas were not present in mixtures. Average values for sample sizes of 100 ranged from 0%^ (California coast, upper Fraser River) to 4% (Oregon coast) and aver- aged 1.8% over the five regions (Table 4). Increased sample sizes of 300 and 500 resulted in small improvements in average accuracy (1.4%^ and 1.0 %). Stock compositions of ocean-caught coho salmon Genotypes for 56 loci were scored for 730 juvenile coho salmon captured in ocean trawls in 1998-2000 (Table 5). About 65% of the 455 fish in June trawls were sampled Teel et al.: Genetic analysis of juvenile Oncorhynchus kisutch 647 E Q British Columbia upper Cascaae R _ ^ 84 ^^K,^© 64 Dongoness H >.j O _?4 83 ^ai „K 87 o Q27 YaquBiaR California coast V 033 NehalemH Puget Sound north Washington coast Oregon coast Klamath River to Cape Blanco south Washington Coast ■1 .0 -0.3 0 4 10 1 7 Dimension 1 Figure 2 Multidimensional scaling (MDS) of Cavalli-Sforza and Edwards ( 1967) chord distances based on 56 allozyme loci between samples of 89 populations of coho salmon extending from northern California to southern British Columbia. Location numbers are given in Table 1 and Figure 1. Populations within regions are identified with polygons where possible. Open circles indicate populations that did not cluster closely with nearby populations. Populations 2 from the California coast and 89 from the upper Fraser River fall beyond the scale of the plot. Table 3 Mean estimated percentage contributions (± standard deviations) of 100 bootstrap resamplings of mixtures composed of fish from only one region. Population numbers are explained in Table 1. 71=100 7^=300 n=500 Region of largest Region (populations) Estimate Estimate Estimate misallocation California coast (1-4) 95 ±4 97 ±3 97 ±3 Oregon coast Klamath River to Cape Blanco (5- -9) 94 ±5 96 ±3 96 ±3 Columbia River Oregon coast (10-33) 86 ±7 91 ±4 92 ±3 Klamath River to Cape Blanco Columbia River (34-49) 84 ±8 92 ±3 93 ±3 Oregon coast South Washington coast (50-56) 88 ±8 95 ±3 95 ±3 North Washington coast North Washington coast (57-63) 81 ±10 88 ±5 90 ±4 Puget Sound Puget Sound (64-82) 85 ±7 90 ±5 90 ±4 British Columbia British Columbia (83-88) 83 ±9 88 ±6 90 ±5 Puget Sound Upper Fraser River (89) 98 ±2 99 ±1 99 ±1 Columbia River in the two northern most transects along the Washington coast, 24% in three transects closest to the Columbia River, and 10% in the four most southern transects along the Oregon coast. Samples from these three areas comprised 43%, 23%, and 33%, respectively, of the 275 fish caught in September trawls. The numbers of offshore juveniles caught in 1998 were too small to provide accurate mixed- stock estimates; therefore the ocean samples collected in 1999 and 2000, and a sample pooled over 1998-2000, were analyzed separately. In the 1998-2000 pooled sample, Columbia River populations were estimated to be the major contributing regional group in June (47%, SD=6%) 648 Fishery Bulletin 101(3) Table 4 Actual percentage composition and mean estimated percentage contributions (± standard deviations) of 100 bootstrap resamplings of mixtures composed of 100, 300, and 500 fish Popul ation numbers are explained in Table 1. « = 100 n=300 n=500 Region (populations) Actual Estimate Estimate Estimate California coast ( 1-4) 0 0±1 0±0 0±0 Klamath River to Cape Blanco (5-9) 0 3 ±4 2 ±2 1±2 Oregon coast (10-33) 20 21 ±8 20+5 20 ±4 Columbia River (34-49) 50 44 ±11 46 ±6 48 ±5 South Washington coast (50-56) 15 14 ±8 14 ±4 14 ±3 North Washington coast (57-63) 15 10 ±7 12 ±5 12 ±4 Puget Sound (64-82) 0 4 ±5 4 ±3 3 ±2 British Columbia (83-88) 0 2 ±3 1±2 1 ±1 Upper Fraser River (89) 0 0±1 0±1 0±0 and September (32%, SD=9%). The Oregon coastal region contributed about 18% (SD=5%') to the June mixture and 21% (SD=7% ) to the September sample. The estimated con- tribution of Puget Sound fish to the pooled ocean samples was much greater in September ( 17%, SD=7%) than it was in June(3%, SD=2%'). Genetic mixed-stock analysis of ocean-caught hatchery fish with CWTs provided a direct comparison of genetic estimates and a mixed-stock sample of known origins (Brodziak et al. 1992). Only 41 fish had CWTs (Table 6). No fish with CWTs appeared in the 1998 sample. Most of the fish with CWTs in 1999 and 2000 originated from Co- lumbia River (68%, n=28) and Oregon coastal (12%, n=5) hatcheries. In the genetic analysis of the 41 fish, Columbia River hatcheries were estimated to contribute about 22 fish (53% , SD=2 1% ). Approximately 7 fish ( 16% , SD= 17% ) were estimated to originate from Oregon coastal hatcheries. Of the 730 juveniles sampled during the study, 501 (69% ) bore hatchery marks ( clipped adipose fins ). The per- centage of unmarked fish in the September sample (35%) was greater than that in June (29% ). Genetic mixed-stock estimates for hatchery-marked fish alone indicated that 69% (SD=6%) originated from the Columbia River and 14% (SD=4%) from Oregon coastal hatcheries (Table 7). The sample of unmarked fish, which contained a mixture of wild and unmarked hatchery fish, was estimated to have a much smaller proportion of Columbia River fish (20%, SD=8% ) but a larger proportion of coastal Oregon (36%, SD=9%) and northern Washington (25%, SD=7%) fish (Table 7). About 30% of unmarked fish in the pooled ocean sample originated from hatcheries (Eq. 1) and 70% from wild populations. Estimated contributions from hatchery and wild populations of all ocean juveniles sampled (marked and unmarked) were 78% and 22%, re- spectively. Coho salmon originating in the Columbia River were estimated to comprise 54% of the total sample, but only 1% consisted of wild fish. Oregon coastal rivers con- tributed 21% to the total ocean sample, and nearly equal proportions were contributed from hatcheries and wild populations. Discussion Usefulness of coho salmon allozyme data for mixed-stock analysis Although the gene diversity analysis indicated that the level of allele-frequency differentiation among populations within regions was similar to that between regions, further analyses showed that the magnitude of regional differen- tiation in the baseline was sufficient to provide accurate mixed-stock estimates. First, we found several genetically discrete population groups of coho salmon over an area extending from California to southern British Columbia. Most of the samples in the MDS plot clustered with nearby samples, and the north-south arrangement of neighboring population groups indicated that isolation by distance is an important component of genetic population structure on this geographic scale. As with other species of Pacific salmon, natal homing to spawning areas is an important isolating mechanism between populations of coho salmon. Second, the analysis of simulated stock mixtures also demonstrated that regional differences were sufficient to provide reliable estimates of coho salmon stock composi- tions. Accurate estimates were obtained from simulated sample sets composed of 100% contributions from each re- gion (Table 3). Third, a more rigorous test of the adequacy of the baseline was made by comparing genetic estimates with direct determinations based on CWTs. These esti- mates were reasonably accurate, especially for the largest contributing regions (Table 6), given the small sample of only 41 fish bearing CWTs. Both the simulation and CWT mixture results are consistent with the findings of Wood et al. ( 1987 ) that estimation accuracy decreases substantially when mixture sample sizes are small and when genetic separation among stocks is limited. Lastly, the analyses of ocean-caught mixture samples themselves appeared to provide reasonable composition estimates (Table 5). Ad- ditionally, estimates for samples pooled over years tended to be intermediate between the two annual estimates, as would be expected from pooling. Teel et al : Genetic analysis of juvenile Oncorhynchus kisutch 649 Table 5 Estimated percentage stock compositions (standard deviations), samp e sizes (n ). and recoveries of coded wire tags (CWT) for coho 1 salmon sampled in trawl surveys along the Oregon and Washington coasts in June and September 1998, 1999, and 2000 . Stock compositions were not estimated for June [n =43) and September (n-- = 18) 1998 because of small sample sizes. None of the 1998 | samples contained coded wire tags. Region June September Est. CWT Est. CWT 1999 n California coast 278 152 0±1 0 0±0 0 Mamath River to Cape Blanco 6 ±6 0 0±0 0 Oregon coast 25 ±7 5 25 ±8 0 Columbia River 46 ±9 8 20 ±14 4 South Washington coast 11 ±4 2 9 ±5 0 North Washington coast 10 ±5 2 18 ±15 0 Puget Sound 3 ±4 0 25 ±9 1 British Columbia 0±1 0 3 ±3 0 Upper Fraser River 0±0 0 0±0 0 2000 n California coast 134 105 0±0 0 1±3 0 Klamath River to Cape Blanco 1±7 0 0±0 0 Oregon coast 11 ±7 0 17 ±8 0 Columbia River 40 ±11 11 48 ±16 5 South Washington coast 17 ±7 0 6 ±9 0 North Washington coast 21 ±11 0 10 ±16 0 Puget Sound 11 ±8 0 14 ±7 1 British Columbia 0±0 0 3 ±8 2 Upper Fraser River 0±0 0 0±0 0 1998, 1999, and 2000 combined n California coast 455 275 0±0 0 0±0 0 Klamath River to Cape Blanco 7 ±4 0 0±0 0 Oregon coast 18 ±5 5 21 ±7 0 Columbia River 47 ±6 19 32 ±9 9 South Washington coast 11 ±3 2 9 ±4 0 North Washington coast 13 ±4 2 19 ±11 0 Puget Sound 3 ±2 0 17 ±7 2 British Columbia 0±0 0 2 ±2 2 Upper Fraser River 0±0 0 0±0 0 Nonetheless, the usefulness of the allozyme baseline that we compiled for coho salmon is limited by two factors. First, few samples in the baseline are from California and British Columbia populations. Although the baseline appears to be adequate to analyze stock mixtures of juvenile coho salmon off Oregon and Washington, mixed stock analyses of sam- ples from other marine areas, particularly to the north, re- quires the sampling of additional populations. Second, our study demonstrated that estimates of stock compositions are not sufficiently accurate to effectively identify stock groups that are absent from mixtures or present in small proportions (Tables 4 and 6). Estimation accuracy can be improved by using additional gene markers. These markers will likely be based on DNA variability because coho salmon minisatellite (Miller et al., 1996; Beacham et al., 1996) and microsatellite (Small et al., 1998a; 1998b; Beacham et al., 2001) loci show much higher levels of polymorphism than do allozyme loci. Recently, variation at eight microsatellite DNA loci and one Mhc locus in coho salmon populations in British Columbia and Washington was used to estimate the stock compositions of fisheries off the west coast of Vancouver Island (Shaklee et al., 1999; Beacham et al., 2001). However, the use of highly polymorphic microsatel- lite loci may not provide increased discrimination among populations on large geographical scales because of allelic convergence from multiple mutations (Nauta and Weiss- 650 Fishery Bulletin 101(3) ing, 1996). Nonetheless, the extension of a DNA baseline to include populations in Oregon and California, may resolve fine-scale (geographic and temporal) differences between coho salmon populations in southern coastal areas. Stock compositions of ocean-caught juvenile coho salmon Studies using large purse seines conducted in 1981-85 revealed that juvenile coho salmon were the most abundant of the Oncorhynchus species in the nearshore areas along the Oregon and Washington coasts ( Pearcy and Fisher, 1988; 1990). Pearcy and Fisher (1988; 1990) captured hatchery- Table 6 Actual composition and estimated contributions ± stan- dard deviations) of a mixture of 41-CWT fish. Region Actual Genetic estimate Number 7c Number % California coast 0 0 1 3 ±4 Klamath River to Cape Blanco 0 0 0 0±0 Oregon coast 5 12 7 16 ±17 Columbia River 28 68 22 53 ±21 South Washington coast 2 5 0 0±0 North Washington coast 3 7 5 11 ±9 Puget Sound 1 2 5 11 ±18 British Columbia 2 5 2 6 ±11 Upper Fraser River 0 0 0 0±0 tagged juvenile coho salmon and concluded they were not highly migratory, often remaining close to their point of sea entry for several months. Our genetic results corroborate that finding. Genetic estimates indicate that about 89% of ocean juveniles caught in June and 81% in September originated from the Columbia River and adjacent coastal rivers. Recoveries of hatchery-tagged fish in=41) also indi- cate that juveniles remain near river mouths in their first few months after ocean entry; only three of these CWT- marked fish came from hatcheries in other regions. However, our genetic results indicate that a change has occurred in the distribution of Washington coastal and Puget Sound juvenile coho salmon. In the 1980s, juvenile coho salmon from Washington coastal hatcheries were not recovered along the Washington and Oregon coasts after mid summer, apparently having migrated northward (Pearcy and Fisher, 1988). Pearcy and Fisher (1990) also found that Puget Sound coho salmon did not migrate along the Washington and Oregon coast until sometime between their first and second summer at sea. However, our genetic results showed that in 1998-2000 fish from Washington coastal streams and hatcheries comprised substantial proportions of the juveniles in nearshore areas along the Washington and Oregon coast in both early and late summer (24% and 28%). We also found that juvenile coho salmon from Puget Sound are present in late summer. Our finding that coho salmon from northern stocks move south along the coast during their first summer was substanti- ated by the catch of CWT-marked fish originating from Puget Sound (n=2) and southern British Columbia {n=2). Recent reductions in the number of coho salmon smolts released from the region's hatcheries have not resulted in a decrease in the proportion of hatchery juveniles along the Oregon and Washington coasts. Annual releases of hatch- Table 7 Estimated percentage stock compositions and sampl 3 sizes for populations of marked (fish with clipped adipose fins) and unmarked coho salmon sampled in trawl surveys along the Oregon and Washington coasts in 1998 1999, and 2000. Samples from June and September were combined. Separate estimates for the contributions of hatchery and wild stocks were made by using estimates of | hatchery marking rates for each region. Marked fish Unmarked fish (hatchery fish) (hatchery and Wild fish) All fish Genetic estimate Genetic estimate (n=501) (n=229) Hatchery Wild Hatchery Wild Total Region (%) (%) (%) (%) (%) (%) (%) California coast 0±0 1±2 0 1 0 0 0 Klamath River to Cape Blanco 1 ±2 1±7 0 1 1 0 1 Oregon coast 14 ±4 36 ±9 4 32 n 10 21 Columbia River 69 ±6 20 ±8 18 2 53 1 54 South Washington coast 4 ±4 9 ±5 0 9 3 3 6 North Washington coast 1±7 25 ±7 2 23 1 7 8 Puget Sound 6 ±5 8 ±5 6 2 6 1 7 British Columbia 5 ±2 0±0 0 0 3 0 3 Upper Fraser River 0±0 0±0 0 0 0 0 0 Total 100 100 30 70 78 22 100 Teel et a\: Genetic analysis of juvenile Oncorhynchus ktsutch 651 ery smolts exceeded 64 million fish during the early 1980s but have decreased to about 39 million in recent years, a 40% reduction (PSMFC^; NRC^). Nonetheless, the propor- tion of hatchery coho salmon in nearshore marine waters has remained high, averaging 74% in 1981-85 (Pearcy and Fisher, 1990) and 78% in 1998-2000 (present study). This result, therefore, leads to the conclusion that the number of naturally produced juveniles in Oregon and Washington coastal waters has also decreased proportionately during this period. If so, wild populations of coho salmon may also have experienced a decline in abundance on the order of 40%. Steep declines in Columbia River wild populations are particularly evident. At the beginning of the 20"^ century, populations in the Columbia River are thought to have been the largest producers of coho salmon in the region (Chapman, 1986; Lichatowich, 1989) and likely contributed a substantial proportion to the nearshore population of ju- venile salmon. At present, Columbia River juveniles pre- dominate along the coast. However, these fish are almost entirely releases from hatchery facilities and Columbia River wild coho salmon are conspicuously absent. Acknowledgments We are grateful to George Milner and Paul Aebersold who developed much of the allozyme baseline for coho salmon. Sewall Young, Laurie Weitkamp, Kathleen Neely, Bill Waknitz, Kathryn Kostow, Orlay Johnson, Ken Currens, Eric Beamer, Scott Chitwood, Doug Cramer, Marc Miller, and Jennifer Nielsen provided baseline samples. We thank Ed Casillas, Ric Brodeur, Bob Emmett, Cindy Bucher, Susan Hinton, Cheryl Morgan, Paul Bentley, and Joe Fisher for providing coho salmon samples and data from their coastal salmon surveys. This study was supported in part by funds from the Bonneville Power Administration and the U.S. GLOBEC program as part of an initiative to understand the effects of ocean dynamics on salmon populations. Literature cited Aebersold. P. B., G. A. Winans, D. J. Teel, G. B. Milner, and F. M. Utter. 1987. Manual for starch gel electrophoresis: a method for the detection of genetic variation. U.S. Dep. Conimer., NOAA Tech. Report NMFS 61, 19 p. Beacham, T. D., J. R. Candy, K. J. Supemault, T. Ming, B. Deale, A. Schulze, D. Tuck, K. H. Kaukauna. J. R. Irvine, K. M. Miller, and R. E. Wither. 2001. Evaluation and application of microsatellite and major histocompatibility complex variation for stock identification of coho salmon in British Columbia. Trans. Am. Fish. Soc. 130;1116-1149. 3 NRC (Natural Resource Consultants, Inc.). 1995. Database of artificially propagated anadromous salmon (database). [Available from Environmental and Technical Services Division, NMFS, 525 N.E., Oregon Street, Portland, OR 97232.1 Beacham, T. D., K. M. Miller, and R. E. Withler 1996. Minisatellite DNA variation and stock identification of coho salmon. J. Fish Biol. 49:411-429. Beamish, R. J., D. McCaughran, J. R. King, R. M. Sweeting, and G. A. McFarlane. 2000. Estimating the abundance of juvenile coho salmon in the Strait of Georgia by means of surface trawls. North Am. J. Fish. Manage. 20:369-375. Brodziak, J., B. Bentley, D. Bartley, G.A.E. Gall, R. Gomulkiewicz, and M. Mangel. 1992. Tests of genetic stock identification using coded wire tagged fish. Can. J. Pish. Aquat. Sci. 49:1507-1517. Cavalli-Sforza, L. L., and A. W. F Edwards. 1967. Phylogenetic analysis: models and estimation pro- cedures. Evolution 21:550-570. Chakraborty, R., M. Hagg, N. Ryman, and G. Stahl. 1982. Hierarchical gene diversity analysis and its applica- tion to brown trout population data. Hereditas 97:17-21. Chapman, D. W. 1986. Salmon and steelhead abundance in the Columbia River in the nineteenth century. Trans. Am. Fish. Soc. 115:662-670. Debevec, E. M., R. B. Gates, M. Masuda, J. Pella, J. Reynolds, and L. W. Seeb. 2000. SPAM (version 3.2): statistics program for analyzing mixtures. J. Hered. 91:509-511. Emmett, R. L., and R. D. Brodeur. 2000. Recent changes in the pelagic nekton community off Oregon and Washington in relation to some physical oceano- graphic conditions. North Pacific Anadr. Fish Comm. Bull. 2:11-20. Godfrey, H. 1965. Coho salmon in offshore waters. In Salmon of the North Pacific Ocean. Part IX. Coho, chinook, and masu salmon in offshore waters, p. 1-39. Int. North Pacific Fish. Comm. Bull. 16. Guo, S. W., and E. A. Thompson. 1992. Performing the exact test of Hardy- Weinberg propor- tions for multiple alleles. Biometrics 48:361-372. Guthrie, C. M., E. V. Fariey Jr., N. M. L. Weemes, and E. C. Martinson. 2000. Genetic stock identification of sockeye salmon cap- tured in the coastal waters of Unalaska island during ApriL/May and August 1998. North Pacific Anadr. Fish Comm. Bull. 2:309-315. Hartt, A. C, and M. B. Dell. 1986. Early oceanic migrations and growth of juvenile Pacific salmon and steelhead trout. Int. N. Pac. Fish. Comm. Rep. 46, 105 p. Johnson, O. W, T A. Flagg, D. J. Maynard, G. B. Milner, and F W. Waknitz. 1991. Status review for lower Columbia River coho salmon. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS F/NWC- 202, 94 p. Lawson, P. W., and R. M. Comstock. 2000. The proportional migration selective fishery model. In Sustainable fisheries management: Pacific salmon (E. E. Knudsen, C. R. Steward, D. D. MacDonald, J. E. WilUams, and D. W Reiser, eds. ), p. 423-433. Lewis Pubhshers, Boca Raton, FL. Lichatowich, J. A. 1989. Habitat alteration and changes in abundance of coho {Oncorhynchus kisutch) and chinook [Oncorhynchus tshaw- ytscha) salmon in Oregon's coastal streams. In Proceed- ings of the national workshop on effects of habitat alteration on salmonid stocks, May 6-8, 1987, Nanaimo, B.C. (C. D. 652 Fishery Bulletin 101 (3) Levings, L. B. Holtby, and M. A. Henderson, eds.), p. 92-99. Can. Spec. Publ. Fish. Aquat. Sci. 105. Miller, K. M., R. E. Withler, and T. D. Beacham. 1996. Stock identification of coho salmon (Oncorhynchus kisutch) using minisatellite DNA variation. Can. J. Fish. Aquat. Sci. 53:181-195. Milner, G. B., D. J. Teel, F M. Utter, and G. A. Winans. 1985. A genetic method of stock identification in mixed populations of Pacific salmon, Oncorhynchus spp. Mar. Fish. Rev. 47:1-8. Nauta, M. J., and F J. Weissing. 1996. Constraints on allele size at microsatellite loci: implica- tions for genetic differentiation. Genetics. 143:1021-1032. Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590. Orsi, J. A., M. V. Sturdevant, J. M. Murphy, D. G. Mortensen, and B. L. Wing. 2000. Seasonal habitat use and early marine ecology of juve- nile Pacific salmon in southeastern Alaska. North Pacific Anadr Fish Coram. Bull. 2:111-122. Pearcy W. G., and J. P Fisher. 1988. Migrations of coho salmon, Oncorhynchus kisutch, during their first summer in the ocean. Fish. Bull. 86: 173-195. 1990. Distribution and abundance of juvenile salmonids off Oregon and Washington, 1981-1985. U.S. Dep. Commerce, NOAA Tech. Report NMFS 93, 83 p. Pella, J., and G. B. Milner 1987. Use of genetic marks in stock composition analyses. In Population genetics and fishery management (N. Ryman and F Utter, eds.), p. 247-276. Washington Sea Grant, Univ. Washington Press, Seattle, WA. Raymond, M., and F Rousset. 1995. GENEPOP (versions 1.2 and 3.1): population genet- ics software for exact tests and ecumenism. Heredity 86: 248-249. Shaklee, J. B., F W. Allendorf, D. C. Morizot, and G. S. Whitt. 1990. Gene nomenclature for protein-coding loci in fish. Trans. Am. Fish. Soc. 119:2-15. Shaklee, J. B., T. D. Beacham, L. Seeb, and B. A. White. 1999. Managing fisheries using genetic data: case studies from four species of Pacific Salmon. Fish. Res. 43:45-78. Small, M. P, R. E. Withler, and T. D. Beacham. 1998a. Population structure and stock identification of Brit- ish Columbia coho salmon iOncorhynchus k!sutch)based on microsatellite DNA variation. Fish. Bull. 96:843-858. Small, M. P, T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998b. Discriminating coho salmon (Oncorhynchus kisutch ) populations within the Eraser River, British Columbia, using microsatellite DNA markers. Mol. Ecol. 7:141-155. 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. Utter, F, P. Aebersold, and G. Winans. 1987. Interpreting genetic variation detected by electro- phoresis. In Population genetics and fishery management (N. Ryman and F Utter, eds.), p. 21^5. Washington Sea Grant, Univ. Washington Press, Seattle, WA. Van Doomik, D. M., M. J. Ford, and D. J. Teel. 2002. Patterns of temporal genetic variation in coho sal- mon: estimates of the effective proportion of 2-year-olds in natural and hatchery populations. Trans. Am. Fish. Soc. 131:1007-1019. Waples, R. S. 1988. Estimation of allele frequencies at isoloci. Genetics 118:371-384. 1990. Temporal changes of allele frequencies in Pacific salmon: implications for mixed-stock fishery analysis. Can. J. Fish. Aquat. Sci. 47:968-976. Weitkamp, L. A., T C. Wainright, G. J. Bryant, G. B. Milner, D. J. Teel, T G. Kope, and R. S. Waples. 1995. Status review of coho salmon from Washington, Ore- gon, and California. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-NWFSC-24, 258 p. Wood, C. C, S. McKinnell, T. J. Mulligan, and D. A. Foumier. 1987. Stock Identification with the maximum likelihood mixture model: sensitivity analysis and application to com- plex problems. Can. J. Fish. Aquat. Sci. 44:866-881. 653 Abstract— Recreational fisheries in the waters off the northeast U.S. target a variety of pelagic and demersal fish species, and catch and effort data sampled from recreational fisheries are a critical component of the information used in resource evaluation and man- agement. Standardized indices of stock abundance developed from recreational fishei-y catch rates are routinely used in stock assessments. The statistical prop- erties of both simulated and empirical recreational fishery catch-rate data such as those collected by the National Marine Fisheries Service (NMFS) Marine Recreational Fishery Statis- tics Survey (MRFSS) are examined, and the potential effects of different assumptions about the error structure of the catch-rate frequency distribu- tions in computing indices of stock abundance are evaluated. Recreational fishery catch distributions sampled by the MRFSS are highly contagious and overdispersed in relation to the normal distribution and are generally best characterized by the Poisson or nega- tive binomial distributions. The model- ing of both the simulated and empirical MRFSS catch rates indicates that one may draw erroneous conclusions about stock trends by assuming the wrong error distribution in procedures used to developed standardized indices of stock abundance. The results demonstrate the importance of considering not only the overall model fit and significance of classification effects, but also the pos- sible effects of model misspecification, when determining the most appropriate model construction. The statistical properties of recreational catch rate data for some fish stocks off the northeast U.S. coast Mark Terceiro Northeast Fisheries Science Center National Marine Fishenes Service, NOAA 166 Water Street Woods Hole, Massachusetts 02543 E-mail address: mtercer@whsunl wh whoiedu Manuscript approved for publication 30 January 2003 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:653-672 (2003). Major recreational fisheries in the wa- ters off the northeast U.S. coast target a wide variety of pelagic and demer- sal fish species (NMFS, 1995, 1996). Fishery data collected in the National Marine Fishery Service ( NMFS ) Marine Recreational Fishery Statistics Survey (MRFSS) are the basis of fishery catch and effort estimates for most of these recreational fisheries and for indices of population abundance used in stock assessments (USDOC, 1992, 2001). For some stocks, reliable fishery-indepen- dent data such as research trawl survey indices are not available, and therefore the recreational fishery data are essen- tial for tracking stock abundance. The intercept (creel sampling) portion of the MRFSS is an interview-type survey of recreational fishing trips and is con- ducted at public fishing sites such as marinas, launching ramps, fishing piers, and beaches. MRFSS catch esti- mates are made by expanding intercept survey sample catch rates in numbers, calculated on a per trip basis, by the estimated total number of recreational fishing trips. The estimated total number of fishing trips is calculated from data collected in a MRFSS tele- phone survey of households located in coastal counties. The U.S. Department of Commerce (USDOC, 1992, 2001) has provided overviews of the MRFSS inter- cept and telephone survey methods and catch estimation procedures. In many cases recreational and com- mercial catch rates used as abundance indices are standardized by using general linear models that assume a lognormal error distribution (Gulland, 1956; Robson, 1966; Gavaris, 1980; Kimura, 1981). Commercial fishery catch-rate data generally meet tests of normality when log-transformed (Gulland, 1956; O'Brien and Mayo, 1988). Because of the efficiency and "integrating" property of commercial fishing gear (including trawls, fixed nets, and longlines), even catch rates on a per tow or per set basis are usu- ally lognormally distributed (Taylor, 1953). An important characteristic of commercial data is that catch rates of zero (tows or sets with no catch of the target species) are rare. With the assumption that there is an underlying lognormal error distribu- tion, general linear models have often been used to standardize recreational fishery catch rates and compute indi- ces of abundance. This approach has been used in the assessments of blue- fin tuna (Brown and Browder, 1994), summer flounder (TerceiroM, black sea bass (NEFSC^), tautog (NEFSC^), winter flounder (NEFSC^), and bluefish ' Terceiro, M. (ed.). 1993. Assessment of summer flounder (Paralichthys denta- tus), 1993: report of the stock assessment workshop summer flounder working group. Northeast Fisheries Science Center refer- ence document 93-14, 72 p. Northeast Fisheries Science Center, Woods Hole, MA 02543. - NEFSC (Northeast Fisheries Science Cen- ter). 1996a. Report of the 20th north- east regional stock assessment workshop (20th SAW): Stock Assessment Review Committee (SARC) consensus summary of assessments. Northeast Fisheries Science Center reference document 95-18, 210 p. Northeast Fisheries Science Center, Woods Hole, MA 02542. 'NEFSC. 1996b. Report of the 21st north- east regional stock assessment workshop (21st SAW): Stock Assessment Review Committee (SARC) consensus summary of assessments. Northeast Fisheries Sci- ence Center reference document 96-05d. 200 p. Northeast Fisheries Science Center, Woods Hole, MA 02543. 654 Fishery Bulletin 101(3) (NEFSC"*; Gibson and Lazar''). However, Banneret and Austin (1983) noted that the sampling distribution of rec- reational catch data is often highly skewed with a longer right-hand tail than might be expected even from a log- normal distribution. Furthermore, depending on the way the catch rate is defined (i.e. catch per trip, day, or hour), recreational fishery catch-rate distributions may contain a high proportion of zero catches. Hilbom ( 1985 ) presented a frequency distribution of num- bers of salmon caught per trip in the British Columbia sport fishery that appears to be best characterized by the nega- tive binomial distribution, with a catch per hour frequency best characterized by the Poisson distribution. Jones et al. (1995) investigated the statistical properties of recreational fishery sampling data collected in angler surveys in Virginia and noted that the non-normality of recreational fishery data may violate assumptions of lognormality in methods used to develop indices of abundance, and especially the validity of confidence intervals. Power and Moser (1999) expressed similar concerns about sampled distributions of fish and plankton collected by research trawl nets, noting that the assumption of an underlying normal or lognormal distribution for these types of data is commonplace, and perhaps in error, and that distributions such as the Pois- son or negative binomial may be more appropriate. Smith (1990, 1996) recommended various nonparametric resam- pling methods (e.g. bootstrap confidence intervals) for char- acterizing the dispersion of highly skewed research trawl survey catch distributions having a large proportion of zero catches. Smith (1999) modeled angling success for salmon, expressed as the catch after the first hour of angling, using a negative binomial distribution model. In addition to the Poisson and negative binomial, al- ternatives to the lognormal error model for recreational fishery catch rates also include the delta-lognormal and delta-Poisson error models. These models are combinations of the delta distribution (Pennington, 1983) and lognormal or Poisson model approaches. The delta distribution has been used in modeling fish and plankton abundance indices from research trawl survey data, which are characterized by highly skewed distributions with a relatively high pro- portion of zero catches (Pennington, 1983). In the combined delta-lognormal and delta-Poisson approaches, indices of abundance are modeled as a product of binomially distrib- uted probabilities of a positive catch and lognormal or Pois- son distributed positive catch rates. The delta-lognormal model has been used in modeling fish-spotter data (Lo et al., 1992) and in the standardization of recreational fishery catch rates forbluefin tuna (Brown and Porch, 1997; Turner ^ NEFSC. 1997. Report of the 23rd northeast regional stock assessment workshop (23'''' SAW): Stock Assessment Review Committee I SAR(^ ) consensus summary of assessments. North- east Fisheries Science Center reference document 97-05. 191 p. Northeast Fisheries Science Center, Woods Hole, MA 03543. ■'■' Gibson, M. R.. and N. Lazar 1998. Assessment and projection of the Atlantic coast bluefish using a biomass dynamic model. A report to the Atlantic States Marine Kisheri(>s Commission Blue- fish Technical Committee and Mid-Atlantic Fishery Manage- ment Council Scientific and Statistics Committee, 29 p. Rhode Island Division of Fish and Wildlife, Jamestown, RI 02835 et al., 1997; Brown, 1999; Ortiz et al., 1999), both character- ized by a highly contagious spatial distribution and a large proportion of zeroes. Bluefin and yellowfin tuna catch rates in the commercial and recreational fisheries have also been standardized by using Poisson (Brown and Porch, 1997), negative binomial (Turner et al., 1997), and delta-Poisson error distributions (Brown, 2001; Brown and Turner, 2001) to address these distributional characteristics. In this study I first examine the statistical properties of recreational fishery catch-rate data as sampled by the MRFSS. Next, I examine the goodness of fit to different statistical distributions of empirical MRFSS catch rates, on both per trip and per hour bases. I then explore the effects of five different assumptions about the error struc- ture of the catch-rate frequency distributions (lognormal, delta-lognormal, Poisson, delta-Poisson, and negative binomial) in deriving standardized indices of abundance with general linear models, using simulated recreational fishery and empirical MRFSS catch per trip (zero catches included) data. Materials and methods Overview of statistical methods This work focuses on catch number per trip sampled in the MRFSS as the index of abundance. The distributional properties of MRFSS catch-per-hour rates are also exam- ined, in order to explore whether the general conclusions reached for catch-per-trip rates are likely to be similar to catch-per-hour rates. Directed trips are defined as those for which interviewed anglers indicated that they were intending to catch a particular species as a primary or secondary target, whether successful or not (zero catches included). In analyses of trips for all species, all trips were used regardless of target or success (zero catches included). Catch rates were expressed as integer (natural) numbers of fish per trip or per hour A value of 1 was added to all observations when applying a lognormal transformation to allow inclusion of the zero catch rate observations (this constant was subtracted upon retransformation to the original scale). Expected sample values for the lognormal distribution were calculated by using the normal distribution and log-transformed catch rates (Sokol and Rohlf 1981). Previous work on MRFSS catch-per-trip data has shown that the value of 1 is the appropriate constant to be added (Terceiro'; NEFSC'^) be- cause it tends to minimize the sum of the absolute value of skew and kurtosis for these distributions (Berry, 1987). The standard logarithmic transform bias correction was applied to express results in the original arithmetic scale (Finney, 1951; Bradu and Mundlak, 1970). No constant was added when data were analyzed under the assumption of bino- mial, Poisson, or negative binomial error distributions. The binomial distribution is a discrete frequency (prob- ability) distribution of the number of times an event oc- curs in a sample in which some proportion of the members possess some variable attribute (Snedecor and Cochran, 1967). Each event is assumed independent of other prior Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 655 Table 1 Descriptive statistics for MRFSS (Marine Recreational Fishery Statistics Survey) 1981, 1988, and 1996 northeast U.S. coast catch per trip, including zero catches. Catch is given in numbers offish. CV is the coefficient of variation (% ). D is the Kolmogorov test statistic for normality. Test statistics significant at the 1% level (P<0.01 ) are shown by **, indicating rejection of the null hypothesis that catch rates follow a normal distribution. Species No. of trips Mean Median Variance CV Skew D 1981 Bluefish 4615 3.80 0.00 155.09 328 27.78 0.380** Summer Flounder 3135 1.88 0.00 14.69 204 4.71 0.312** Atlantic cod 509 2.55 1.00 13.49 144 2.48 0.244** Scup 269 8.44 2.00 275.10 196 4.08 0.305** All species 20,280 3.45 0.00 355.94 547 65.48 0.427** 1988 Bluefish 7294 1.60 0.00 18.65 270 6.42 0.355** Summer Flounder 4779 2.26 0.00 18.48 190 3.70 0.300** Atlantic cod 1558 4.56 2.00 21.55 154 3.68 0.258** Scup 960 9.28 3.00 312.65 190 4.48 0.300** All species 48,423 2.29 0.00 43.66 289 8.94 0.365** 1996 Bluefish 5457 1.20 0.00 13.12 301 8.40 0.370** Summer Flounder 7047 2.33 1.00 13.49 157 3.40 0.263** Atlantic cod 1099 3.97 1.00 43.34 166 3.29 0.273** Scup 643 13.83 4.00 524.60 165 3.44 0.273** All species 81,057 2.57 0.00 47.32 268 10.45 0.354** events in the same sample (Sokal and Rohlf, 1981). In the present study, the binomial distribution was used only to model the probabilities of a positive catch (as opposed to a zero catch; thus the variable attribute of the observation is either catch or no catch ) in the combined delta-lognormal and delta-Poisson models. The Poisson distribution is also a discrete frequency dis- tribution of the number of times an event (such as catching a fish during a trip ) occurs in a sample and is characterized by a small mean value in relation to the observed maxi- mum number of events within the sample (Sokal and Rohlf, 1981). For a Poisson distribution, the expected variance is equal to its mean, and Poisson frequency distributions are more highly skewed than normal or lognormal distribu- tions (Bliss and Fisher, 1953). The negative binomial is a discrete frequency distribu- tion with a higher degree of dispersion than the Poisson distribution, such that the variance is significantly larger than the mean. A negative binomial distribution will con- verge to a Poisson as the variance approaches the mean (Bliss and Fisher, 1953). Although not as widely applied as the Poisson in the analysis of count data, there is a growing literature describing the properties of negative binomial regression methods to be used when analyzing "over-dispersed Poisson" frequency distributions (Manton et al., 1981; Lawless, 1987). The dispersion parameter of the negative binomial distribution, k, is a positive exponent relating the mean and variance of the distribution such that as the variance of a distribution exceeds the mean, the value ofk decreases and the "over-dispersion" of the distri- bution in relation to a Poisson distribution increases. The most efficient estimate of the sample parameter, k', is esti- mated by maximum likelihood (Bliss and Fisher, 1953). Descriptive statistics and frequency distributions of MRFSS catch per trip and catch per hour observations were compiled by using the SAS FREQ and UNIVARIATE procedures (SAS, 2000 ). Tests of normality were made with the Kolmolgorov-Smirnov D-statistic for normality (test significance expressed as probability < D; SAS, 2000). Eval- uation of the most appropriate distributional fit to the data was based on inspection of the frequency distribution plots, the parametric chi-square (x^) and G-statistic goodness- of-fit tests, and the nonparametric Kolmogorov-Smimov (Z3-statistic) goodness-of-fit test for an intrinsic hypothesis (because the expected distributions were calculated from the observed sample moments; Sokol and Rohlf, 1981). For the chi-square and G-tests, when intervals (classes) of catch per trip with fewer than 3 expected instances occurred, expected and observed frequencies for these in- tervals were pooled with the adjacent intervals to obtain a joint class with an expected frequency of occurrence of 3 or more (Sokol and Rohlf, 1981). Because of the large sample sizes involved (»100), the G-test correction suggested by Williams (1976) proved to be very small in a few test calculations and therefore was not routinely applied. Un- realistic (for recreational fishery catch-rate data) negative expected values computed for the lognormal distributions were excluded, and the remaining positive distribution was raised to the observed sample total, so that the expected proportions at each interval summed to 1.0. 656 Fishery Bulletin 101(3) Standardized annual indices of abundance derived from the simulated recreational and empirical MRFSS data were calculated by using maximum likelihood estimation to fit generalized linear models with the SAS GENMOD procedure (SAS, 2000). The SAS (2000) defaults for model specification were generally followed. An identity link func- tion was used under the lognormal distribution assump- tion (catch rates were In-transformed prior to analysis). A logistic link function was used under the binomial distri- bution assumption applied for the probability of positive catch component in the delta-lognormal and delta-Poisson model approaches. A logarithmic link function was used under the Poisson and negative binomial assumptions (SAS, 2000). Type-3 general linear models were fitted in all cases because the results of this type of analysis do not depend on the order in which the terms of the model are specified. The significance of the individual classification effects (factors) in the models was judged by the chi-square statistic (Searle, 1987; SAS, 2000). The overall goodness of fit of the standardization mod- els was evaluated by using the deviance and log-likelihood statistics. The deviance is defined to be twice the difference between the maximum achievable log likelihood and the log likelihood at the maximum likelihood estimates of the model parameters (McCullagh and Nelder, 1989). The devi- ance has a limiting chi-square distribution, and so signifi- cance is judged by comparison to critical values of the chi- square distribution. The scale parameter (i.e. for normal distributions) was held fixed at 1 for all models to facilitate evaluation of goodness of fit and the degree of overdisper- sion for models with different error distribution assump- tions. Holding the scale parameter fixed has no effect on the estimated intercept or model regression coefficients (e.g. in the study, the year coefficients that serve as the annual indices of abundance), but allows equivalent calculation among models of a "dispersion estimate" (SAS, 2000). This "dispersion estimate," measured after model fitting as the deviance divided by the degrees of freedom (deviance/df), is used to judge whether the data are overdispersed or un- derdispersed with respect to the error distribution used in model fitting and is therefore useful in evaluating whether the correct error distribution assumption has been used in the model (McCullagh and Nelder, 1989; SAS, 2000). Descriptive statistics for MRFSS catch rates The descriptive statistics (mean, median, variance, skew- ness, and Kolmolgorov-Smirnov (D) normality test statis- tic) and frequency distributions of MRFSS sample catch rates for 1981, 1988, and 1996 were examined for four species from U.S. Atlantic coast waters (Maine to the east coast of Florida), and in aggregate for all species sampled along the U.S. Atlantic coast. The following individual species were considered: bluefish (Pomatomus saltatrix, an example of a Atlantic coast predatory "gamefish"); summer flounder (Paralichthys dentatus, a Mid Atlantic Bight demersal flatfish); Atlantic cod (Gadus morhua, a New England demersal roundfish); and scup (Stenotomus Table 2 Descriptive statistics for MRFSS (Marine Recreational Fis hery Statistics Survey) 1981, 1988, and 1996 northeast U.S. coast catch 1 per trip, positive catches only. Catch is given in numbers offish. CV is the coefficient of variation (9< . D is the Kolmogorov test statistic for normality. Test statistics significant at the I'X level (P<0.01) shown by **, ndicating rejection of the null hypothesis | that catch rates follow a normal distribution. Species No. of trips Mean Median Variance CV Skew D 1981 Bluefish 2288 7.66 4.00 283.32 220 22.02 0.340** Summer Flounder 1380 4.26 2.67 23.21 113 3.87 0.222** Atlantic cod 298 4.36 3.00 15.19 89 2.26 0.188** Scup 165 13.76 7.33 375.88 141 3.39 0.262** All species 9484 7.33 3.00 732.27 368 47.02 0.395** 1988 Bluefish 2445 4.75 2.33 40.35 133 4.40 0.254** Summer Flounder 2326 4.64 3.00 26.92 112 2.94 0.209** Atlantic cod 1065 6.67 4.00 58.02 114 3.46 0.219** Scup 614 14.52 7.67 413.03 140 3.88 0.245** All species 19,094 5.76 3.00 90.39 165 6.49 0.278** 1996 Bluefish 1666 3.93 2.00 32.26 144 5.61 0.258** Summer Flounder 4196 3.91 2.66 16.46 104 3.13 0.203** Atlantic cod 679 6.43 4.00 54.39 115 2.85 0.210** Scup 438 20.31 12.50 638.90 124 3.05 0.220** All species 39,094 5.30 2.67 83.43 172 8.35 0.286** Tercelro: The statistical properties of recreational catch data off the northeastern U.S. coast 657 Table 3 Descriptive statistics for MRFSS (Marine Recreational Fishery Statistics Survey) 1981, 1988, and 1996 northeast U.S. coast catch per hour, including zero catches. Catch is given in numbers offish. CV is the coefficient of variation (%). D is the Kolmogorov test statistic for normality. Test statistics significant at the 1% level (P<0.01) shown by (**), indicating rejection of the null hypothesis that catch rates follow a normal distribution. Species No. of trips Mean Median Variance CV Skew D 1981 Bluefish 4615 0.77 0.00 5.57 305 12.21 0.371** Summer Flounder 3135 0.36 0.00 0.64 221 6.43 0.325** Atlantic cod 509 0.38 0.17 0.43 172 4.04 0.280** Scup 269 1.59 0.44 7.96 178 2.79 0.287** All species 20,280 0.74 0.00 13.23 491 40.11 0.419** 1988 Bluefish 7294 0.39 0.00 1.26 287 7.26 0.363** Summer Flounder 4779 0.45 0.00 0.82 200 6.11 0.309** Atlantic cod 1558 0.96 0.50 1.99 147 3.32 0.249** Scup 960 2.12 0.67 14.03 177 3.60 0.286** All species 48,423 0.54 0.00 3.07 325 15.18 0.379** 1996 Bluefish 5457 0.34 0.00 1.23 322 7.73 0.378** Summer Flounder 7047 0.52 0.22 0.729 164 4.17 0.271** Atlantic cod 1099 0.86 0.28 1.99 165 3.12 0.272** Scup 643 3.06 1.17 27.78 172 3.79 0.281** All species 81,057 0.62 0.00 4.51 341 35.84 0.385** Table 4 Descriptive statistics for MRFSS (Marine Recreational Fishery Statistics Survey) 1981, 1988, and 1996 northeast U.S. coast catch per hour, positive catches only. Catch is given in numbers offish. CV is the coefficient of variation ('7f ). D is the Kolmogorov test statistic for normality. Test statistics significant at the 1% level (P<0.01) shown by (**), indicating rejection of the null hypothesis that catch rates follow a normal distribution. Species No. of trips Mean Median Variance CV Skew D 1981 Bluefish 2288 1.56 0.75 10.00 203 9.48 0.317** Summer Flounder 1380 0.82 0.50 1.07 126 5.35 0.229** Atlantic cod 298 0.65 0.45 0.56 115 3.64 0.220** Scup 165 2.56 1.67 10.41 125 2.16 0.242** All species 9484 1.58 0.67 26.96 328 29.00 0.381** 1988 Bluefish 2445 1.16 0.63 2.85 145 4.91 0.253** Summer Flounder 2326 0.93 0.60 1.24 120 5.45 0.212** Atlantic cod 1065 1.40 1.00 2.29 108 3.14 0.196** Scup 614 3.31 1.71 17.98 128 3.09 0.223** All species 19,094 1.37 0.67 6.67 188 11.08 0.301** 1996 Bluefish 1666 1.13 0.55 3.15 157 4.80 0.270** Summer Flounder 4196 0.87 0.58 0.90 109 3.93 0.198** Atlantic cod 679 1.39 0.80 2.49 114 2.68 0.205** Scup 438 4.49 2.73 34.38 131 3.38 0.228** All species 39,094 1.29 0.63 8.49 226 28.09 0.331** 658 Fishery Bulletin 101(3) Table 5 Summary of goodness of fit tests for 1996 MRFSS (Marine Recreational Fishery Statistics Si jrvey) catch per trip distributions, | including zero catches for bluefish, summer flounder, Atlantic cod, scup, and all species. Expected number Degrees of X-' G D Species of intervals freedom statistic statistic X 0 01 statistic ^0 01 Bluefish Mean = 1.20 Variance = 13.12 n = 5457 Lognormal 9 6 7314 5722 17 0.462 0.014 Poisson 7 5 5315 4251 15 0.394 0.014 Negative binomial 23 20 68 27 38 0.007 0.014 Summer flounder Mean = 2.33 Variance = 13.49 n = 7047 Lognormal 11 8 8654 4714 20 0.281 0.012 Poisson 10 8 10,902 5772 20 0.307 0.012 Negative binomial 22 19 139 101 36 0.011 0.012 Atlantic cod Mean = 3.97 Variance = 43.34 n = 1099 Lognormal 12 9 2138 1068 22 0.360 0.031 Poisson 12 10 8284 2212 23 0.425 0.031 Negative binomial 25 22 48 22 40 0.015 0.031 Scup Mean = 13.83 Variance = 524.60 n =643 Lognormal 28 25 389,173 3850 44 0.541 0.041 Poisson 25 23 6.67e+07 6391 42 0.544 0.041 Negative bionomial 51 48 305 235 74 0.053 0.041 All species Mean = 2.57 Variance = 47.32 n = 81,057 Lognormal 14 11 180,754 83,230 25 0.382 0.004 Poisson 13 11 306,000 129,928 25 0.440 0.004 Negative binomial 51 48 1577 1146 74 0.020 0.004 chrysops, a Mid-Atlantic demersal schooling roundfish, likely to yield a relatively high catch per trip). These spe- cies were selected as examples because they occur over a broad range along the northeast U.S. coast, are among the most frequently caught by recreational fishermen, and their catch-rate distributions are representative of most species caught by recreational fishermen in the northeast US (USDOC, 1992). Four configurations of catch rate dis- tributions were examined: 1 ) catch per trip distributions including zero catches, 2) catch per trip distributions with positive catches only, 3 ) catch per hour distributions includ- ing zero catches, and 4) catch per hour distributions with positive catches only. Goodness-of-fit statistics for the lognormal, Poisson, and negative binomial distributions were calculated for the four individual species and for all species to help judge which error structure best characterized the MRFSS catch-rate data. A single year ( 1996) is presented because of the similarity of the catch distributions across species and time. Given the results of the Kolmogorov-Smirnov D tests from the descriptive statistics work, which in- dicated that none of the catch rates were normally dis- tributed (see "Results" section), that error structure was not examined further. As with the descriptive statistics analysis, both catch-per-trip and catch-per-hour rates were examined in the goodness-of-fit exercise, both for Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 659 Table 6 Summary of goodness -of-fit tests for 1996 MRFSS (Marine Recreational Fishery Statistics Survey) catch per trip distributions, | positive catches only, for bluefish, summer flounder, Atlantic cod, scup, and all species. Expected number Degrees of X^ G D Species of intervals freedom statistic statistic y2 A 0-01 statistic £'ooi Bluefish Mean = 3.93 Variance = 32.26 n = 1666 Lognormal 9 6 1803 1026 17 0.312 0.025 Poisson 11 9 2091 1211 22 0.312 0.025 Negative binomial 21 18 425 347 35 0.196 0.025 Summer flounder Mean = 3.91 Variance = 16.46 n = 4196 Lognormal 10 7 6068 2863 18 0.270 0.016 Poisson 12 10 3821 2234 23 0.240 0.016 Negative binomial 20 17 699 587 33 0.143 0.016 Atlantic cod Mean = 6.43 Variance = 54.39 n = 679 Lognormal 12 9 3376 962 22 0.379 0.040 Poisson 14 12 3419 925 26 0.365 0.040 Negative binomial 27 24 121 88 43 0.147 0.040 Scup Mean = 20.31 Variance = 638.90 n = 438 Lognormal 30 27 3.74e+ll 6565 47 0.543 0.049 Poisson 32 30 8.09e+07 3477 51 0.475 0.049 Negative binomial 50 47 204 147 72 0.089 0.049 All species Mean = 5.30 Variance = 83.43 n = 39,094 Lognormal 11 8 70,234 24,957 20 0.254 0.001 Poisson 16 14 169,662 59,516 29 0.391 0.001 Negative binomial 50 47 12,293 10,217 72 0.201 0.001 all directed trips including zero catches and for positive catches only. Simulated recreational fishery catch rates To isolate the consequences of possible model misspecifica- tion in deriving standardized indices of abundance, nega- tive binomial distributions with characteristics like those of MRFSS recreational catch-per-trip distributions were simulated by using the SAS RANTBL function (SAS, 2000). The simulated distributions were arranged to provide con- tinuously decreasing, continuously increasing, and peaked (increasing to a peak and then decreasing) trends in an 11-year time series of catch per trip. For the decreasing trend, the simulation procedure began with year 1 set at a mean catch per trip = 3.0, maximum catch per trip of 50 fish per trip, and variance = 81.0, which are characteristic of the MRFSS catch-per-trip distributions for all species (Table 1). For year 1, this combination of mean and vari- ance provided a maximum likelihood estimate of the nega- tive binomial dispersion parameter, k, of 0.23. The vector of expected probabilities of catch per trip for these initial moments, assuming a negative binomial distribution, was then used to randomly generate 1000 660 Fishery Bulletin 101 (3) MRFSS 1996 samples 5OO0 ■ Bluefish 2500 1 0 l-._ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 llAM^ Summer flounder 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 500 1 3k ■ Atlantic C cr 250 I cod III. 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Scup ll«-i — f- - -m- — — » 0 10 20 50 0000 -1 All 5000 species 0 !■■.. 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Catch number per trip (Including zero catches) Figure 1 Marine Recreational Fishery Statistics Survey (MRFSS) 1996 sample data for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast: catch number per trip (including zero catches). The 15 and 50 fish intervals are "plus groups'" because they include totals for larger intervals. Il MRFSS 1996 samples Bluefish 1 lu. 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Summer flounder 0 12 3 4 5 6 7 ll 9 10 11 12 13 14 15 Atlantic cod 1«1 U^ ^A 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Scup llll.li.i-l... I.-I.l ^ 10 20 50 j b All species 1^ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Catch number per trip (positive catches only) Figure 2 Marine Recreational Fishery Statistics Survey (MRFSS) 1996 sample data for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast: catch number per trip ( positive catches only). The 15 and 50 fish intervals are "plus groups" because they include totals for larger intervals. observations of catch per trip (including zeroes) for year 1 (n = 1000). The initial mean for year 2 was then set at 10 percent less than year 1 (i.e. 2.7) and the year 2 set of 1000 observations generated under the negative binomial as- sumption. The dispersion parameter, k, was held constant at the year 1 maximum likelihood estimate of 0.23, result- ing in a decrease in variance, a relatively stable coefficient of variation (CV), and less frequent occurrence of large catch-per-trip values, as the mean decreased. These condi- tions were felt to best reflect the true changes in angler catch per trip as stock abundance declines. The exercise was repeated for years 3 to 11, providing a time series of decreasing simulated recreational fishery catch per trip. The simulated annual distributions, scaled (normalized) to the 1 1-year time series mean of 1.75, were re-ordered to create the increasing and peaked time series. Standardized indices of abundance were then calculated from the simulated, trended series by using lognormal, Pois- son, negative binomial, delta-lognormal, and delta-Poisson models, with year serving as the single classification vari- Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 661 MRFSS 1996 samples LL Bluefish 0123456789 LiL Summer flounder 0123456789 Atlantic cod 0123456789 ilL All species 0123456789 Catch number per hour (including zero catches) Figure 3 Marine Recreational Fishery Statistics Survey (MRFSS) 1996 sample data for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast: catch number per hour (including zero catches). The 9 fish interval is a "plus group" because it includes totals for larger intervals. 0 1 MRFSS 1996 samples LL Bluefish ^ F- : . ^ 0123456789 Ll Summer flounder 0123456789 LL Atlantic cod 0123456789 Scup 5 6 Ll 8 9 All species 0123456789 Catch number per hour (positive catches only) Figure 4 Marine Recreational Fishery Statistics Survey (MRFSS) 1996 sample data for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast: catch number per hour (positive catches only). The 9 flsh interval is a "plus group" because it includes totals for larger intervals. able and index of abundance. Modeled in this way, the negative binomial model is expected to provide year-effect coefficients very close in absolute value to the unstandard- ized, mean simulated catch per trip of the true underlying negative binomial distribution because no other classifica- tion effects are present to account for variance from the unstandardized mean. The deviance of the year coefficients provided by the models, assuming the other error distribu- tions, then provides an indication of the degree of model misspecification because virtually all the estimated vari- ance in this particular exercise is due to model (process) error, except for the small amount generated by the random draw from the starting probability distributions. MRFSS standardized indices of abundance, 1981-98 The potential effect of the assumed error structure on the calculation of standardized indices of abundance was fur- ther explored with empirical examples using the 1981-98 MRFSS time series of catch-per-trip rates (zero catches 662 Fishery Bulletin 101(3) MRFSS 1996 samples Bluefish Observed Lognormal Poisson Neg binomial t -»m , , . 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Catch number per trip (Including zero catches) Rgure 5 Observed and expected catch number per trip (including zero catches) frequency distributions for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast. The 15 and 50 fish intervals are "plus groups" because they include totals for larger intervals. 0 1 2 3 4 5 6 7 e 9 10 11 12 13 14 15 Catch number per trip (positive catches only) Figure 6 Observed and expected catch number per trip (positive catches only) frequency distributions for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast. The 15 and 50 fish intervals are "plus groups" because they include totals for larger intervals. included) for bluefish, summer flounder, Atlantic cod, scup, and for all species. Annual indices of stock abundance were developed from these MRFSS catch rate data following pro- cedures in previous Atlantic coast bluefish and summer flounder stock assessments (Terceiro'; NEFSC'; Gibson and Lazar'). Standardized indices were calculated by applying lognormal, Poisson, negative binomial, delta-lognormal, and delta-Poisson models, using the main effects classifica- tion variables determined in these stock assessments to be statistically significant factors: year, fishing mode (shore, private or rental boat, party or charter boat), state of land- ing (Maine to Florida), fishing wave (two-month sampling period, e.g. Jan-Feb), fishing area (>3 miles from shore, s 3 miles from shore), and daysl2, the angler-reported days of saltwater fishing during the previous 12 months ( a proxy for angler avidity, experience, or skill, or a proxy for all three characteristics). The retransformed, bias-corrected (when necessary) year coefficients serve as the annual indices of stock abundance. Calculation and evaluation of the MRFSS standardized indices followed the general pro- cedures described in the "Overview of statistical methods" in the "Materials and methods" section. Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 663 Table 7 Summary of goodness-of-fit tests for 1996 MRFSS (Marine Recreational Fishery Statistics Survey) catch per hour distributions, including zero catches, for bluefish, summer flounder, Atlantic cod, scup, and all species. Expected number Degrees of x' G D Species of intervals freedom statistic statistic „2 X 001 statistic ^001 Bluefish Mean = 0.35 Variance = 1.23 n = 5457 Lognormal 6 3 2806 3047 11 0.323 0.014 Poisson 5 3 514 41 11 0.028 0.014 Negative binomial 5 2 514 41 9 0.028 0.014 Summer flounder Mean = 0.52 Variance = 0.72 n = 7047 Lognormal 7 4 2022 2430 13 0.245 0.012 Poisson 5 3 1408 1209 11 0.193 0.012 Negative binomial 5 2 1408 1209 9 0.193 0.012 Atlfuitie cod Mean = 0.86 Variance = 2.00 n = 1099 Lognormal 7 4 289 300 13 0.243 0.031 Poisson 5 3 51 11 11 0.051 0.031 Negative binomial 5 2 51 11 9 0.051 0.031 Scup Mean = 3.06 Variance = 27.78 n = 643 Lognormal 11 8 546 346 20 0.312 0.041 Poisson 10 8 1209 583 20 0.347 0.041 Negative binomial 19 16 54 39 32 0.032 0.041 All species Mean = 0.62 Variance = 4.51 n = 81,057 Lognormal 13 10 144,556 126,529 23 0.575 0.004 Poisson 7 5 54,675 72,657 15 0.036 0.004 Negative binomial 7 4 54,675 72,657 13 0.036 0.004 Results Descriptive statistics for MRFSS catch rates Descriptive statistics of MRFSS catch rates for the four catch rate configurations, four individual species, and for all species are presented for the years 1981, 1988, and 1996 (Tables 1-4). These three years are characteristic of the 1981-2002 time series of MRFSS data. Given the similarity among these years, frequency distributions are plotted only for 1996 (Figs. 1-4). Catch rate means, both with and with- out zero catches, are generally much higher than medisms, variances are much larger than the means, skewness is always much larger than zero, and there is a high propor- tion of zero catch and one-fish catch-rate observations. In all cases, the Kolmogorov-Smimov D test statistics were significant at the 1% level. All of these factors indicate that MRFSS catch-rate distributions are highly contagious and overdispersed in relation to the normal distribution (Sokol and Rohlf, 1981). Scup has highest frequency of high catch rates (Figs. 1-4). The scup and Atlantic cod samples exhibit modes at regular intervals of high catch-per trip rates (e.g. 10, 15, 20, 25, and 30 fish per trip) that may indicate some degree of digit bias in the sampling. 664 Fishery Bulletin 101(3) Table 8 Summary of goodness of fit tests for 1996 MRFSS (Marine Recreational Fishery Statistics Survey) catch per hour distributions, positive catches only, for bluefish, summer flounder, Atlantic cod scup, and all species. Expected number Degrees of x'' G D Species of intervals freedom statistic statistic X 001 statistic ^0.01 Bluefish Mean = 1.13 Variance = 3.15 n = 1666 Lognormal 6 3 1508 1462 11 0.446 0.025 Poisson 5 3 590 590 11 0.269 0.025 Negative binomial 5 2 590 590 9 0.269 0.025 Summer flounder Mean = 0.87 Variance = 0.90 n = 4196 Lognormal 6 3 3005 3146 11 0.416 0.016 Poisson 5 3 838 921 11 0.210 0.016 Negative binomial 5 2 838 921 9 0.210 0.016 Atlantic cod Mean = 1.39 Variance = 2.49 n = 679 Lognormal 6 3 414 368 11 0.356 0.040 Poisson 5 3 145 113 11 0.213 0.040 Negative binomial 5 2 145 113 9 0.213 0.040 Scup Mean = 4.49 Variance = 34.38 n = 438 Lognormal 10 7 601 292 18 0.270 0.049 Poisson 11 9 725 324 22 0.280 0.049 Negative binomial 17 14 127 99 29 0.166 0.049 All species Mean = 1.29 Variance = 8.49 n = 39,094 Lognormal 7 4 38,171 33,641 13 0.434 0.001 Poisson 8 6 31,475 16,429 17 0.270 0.001 Negative binomial 8 5 31,475 16,429 15 0.270 0.001 For the catch-per-trip configurations, catch rates were best characterized by the negative binomial di.stribution (Table.s 5-6, Figs. 5-6). Note that the calculated ehi-square, G-. and D-test statistics were generally significant at the 1% level, so that based on strict interpretation of these results, the null hypothesis that the observed distributions come from one of the theoretical distributions was rejected in all cases. How- ever, the calculated test statistics for the negative binomial distributions were at least an order of magnitude smaller than those for the Poisson and lognormal distributions, sug- gesting that an underlying negative binomial distribution was much more likely. The distributions of the catch-per- hour rates generally had a truncated range compared to the catch-per-trip rate configurations (Figs. 1-4). For most of the catch-per-hour distributions, the maximum likelihood solution for the negative binomial k parameter occurred at very large values (>1000). The expected frequencies for the negative binomial distribution therefore converged to those expected for a Poisson distribution, resulting in identical test statistic values and indicating that the catch-per-hour rates are best characterized by the Poisson distribution (Tables 7-8, Figs. 7-8). Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 665 MRFSS 1996 samples Bluefish Observed Lognormal Poisson Neg. binomial Summer Observed flounder --a- Lognoimal — •* — Poisson ■ Neg. binomial Observed — a — Lognormal — ■• — Poisson — •— Neg, binomial 12 3 4 5 6 7 Catch number per hour (including zero catches) Figure 7 Observed and expected catch number per hour (including zero catches) frequency distributions for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast. The 9 fish intervals is a "plus group" because it includes totals for larger intervals. MRFSS 1996 samples Bluefish Observed - ^ — Lognofmal - -• — Potsson — ■ — Meg binomial Scup Observed - -A — Lognormal - -• — Poisson -•— Neg. binomial — Obsen/ed — Lognormal — Poisson Neg. binomial 0123456789 Catch number per hour (positive catches only) Figure 8 Observed and expected catch number per hour (positive catches only) frequency distributions for bluefish, summer flounder, Atlantic cod, scup, and all species, Maine to the Florida east coast. The 9 fish intervals is a "plus group" because it includes totals for larger intervals. Simulated recreational fishery catch rates The eleven simulated distributions of catch per trip had means ranging from 2.80 to 0.98 fish per trip, variances ranging from 31.81 to 4.39, and CVs of about 200%. Simu- lated variance decreased as the simulated mean decreased because the negative binomial dispersion parameter, k, was held constant at 0.23. The resulting unstandardized, simu- lated index of abundance declined by 65'7( over the 11 year series (Table 9). All standardization model fits were highly significant (P<0.001), as characterized by the chi-square statistics for the year effect (Table 10). The three different time series trends had no effect on the results, and therefore only the results for the decreasing series are reported. The Poisson and negative binomial models generated year coefificients as standardized indices of abundance that were very simi- lar to each other and, as expected, virtually identical to the unstandardized annual means, indicating a 65% dechne over the time series (Fig. 9). Interestingly, the diagnostic 666 Fishery Bulletin 101(3) Table 9 Summary statistics for the simulated recreational fishery catch per trip assuming a negative binomial distribution, configured to decline by ID'S in successive time periods (years). For year 1 , starting maximum catch per trip was 50 fish pe r trip, mean was 3.0, variance was 81.00, coefficient of variation (CV) of 300%, and the dispersion parameter of the negative binomial distribution, k. was | 0.23. In years 2-11, k was held constant at the year-1 value of 0.23, a llowing the variance to decrease as the mean catch declined. Annual simulated means were scaled to the 11 year time series mean (1.75) for comparability with standardized indices calculated | for decreasing increasing, and peaked time series trends. Simulated Simulated Scaled mean catch maximum catch Simulated Simulated simulated catch Year per trip per trip variance CV (%) per trip 1 2.80 47 31.81 201 1.60 2 2.49 39 24.75 200 1.42 3 2.27 37 20.97 202 1.30 4 2.05 34 17.23 203 1.17 5 1.85 31 14.24 204 1.06 6 1.67 29 11.84 206 0.95 7 1.49 26 9.65 209 0.85 8 1.32 21 7.36 206 0.75 9 1.21 21 6.47 211 0.69 10 1.09 19 5.38 213 0.62 11 0.98 17 4.39 215 0.56 £ 2.0 1.5 Simulated catch per trip scales to series means 123456789 10 11 1.0 0.5 12 3 4 5 6 7 Year 9 10 11 Figure 9 Simulated recreational fishery indices of abundance mod- eled under different error distribution assumptions. statistics indicated a better determined year effect (more precise year coefficients) for the Poisson than for the nega- tive binomial. However, the dispersion estimate (deviance/ df) for the Poisson model was much greater than 1.0, indi- cating that the input data were overdispersed with respect to the Poisson distribution (Table 10). The latter was the expected result, given that the variance of the annual simulated data sets was much larger than the mean. The results indicated that the negative binomial was a more appropriate model, with a dispersion estimate closer to 1.0, which was also the expected result given the true negative binomial distribution of the simulated data (SAS, 2000). The consequence of assuming a lognormal model for the true underlying negative binomial distribution was a more extreme smoothing of the true time series trends than with the other model assumptions, with a decline of only 28% over the time series (Fig. 9). The diagnostic statistics for the lognormal model indicated a significant model fit, but the dispersion estimate was much less than 1.0, indicating that the input data were underdispersed with respect to the lognormal distribution (Table 10). This finding is reflective of the large number of 0 and 1 catch-per-trip observations, and a lack of observations near the mean of the input prob- ability distribution (SAS, 2000). In this simulation exercise, therefore, the lognormal model dispersion estimate of much less than 1.0 is indicative of model misspecification. As noted in the "Materials and methods" section, the in- dices of abundance from the delta models are calculated as the product of the year-effect coefficients from the two com- ponent models. The interaction of the year coefficients from the binomial proportion positive catches and lognormal or Poisson positive catches components of the delta models Terceiro The statistical properties of recreational catch data off the northeastern U.S. coast 667 Table 10 Summary of model fits for simulated recreational fishery catch per trip (including zero catches) with a decreasing time series trend. Total model degrees of freedom were 10,989; for the positive catches component of the delta models, degrees of freedom were 4,184. Year-model-effect degrees of freedom were 10, and the year-model effect was highly significant (P<0.0001) in all five models. Criterion Value Dispersion estimate (value/df) Lognormal model Deviance 7330 0.6670 Log-likelihood -13,773 Year chi-square 183 Poisson model Deviance 51,719 4.7064 Log-likelihood -7483 Year chi-square 2049 Negative binomial model Deviance 10,699 0.9736 Log-hkelihood Year chi-square 7524 239 Delta models: binomial proportion positive catch Deviance 14,546 1.3237 Log-likelihood Year chi-square -7273 78 Delta-lognormal model: lognormal positive catches Deviance 3474 0.8303 Log-likelihood -5557 Year chi-square 119 Delta-Poisson model: Poisson positive catches Deviance 15,822 3.7815 Log-likelihood 10,466 Year chi-square 936 Table 11 Summary of model fits for estimating indices of abundance from empirical MRFSS (Marine Recreational Fishery Statistics Survey) bluefish catch per trip (including zero catches), 1981-98. Total model degrees of freedom (df) were 130,300; for the positive catches component of the delta models, degrees of freedom were 48,447. All model fits and classification effects were highly significant (P<0.001). Dispersion estimate Criterion Value (value/df) Lognormal model Deviance 84,150 0.6458 Log-likelihood -156,444 Year chi-square 1835 Poisson model Deviance 675,791 5.1864 Log-likelihood -19,680 Year chi-square 20,604 Negative binomial model Deviance 99,393 0.7628 Log-likelihood 190,140 Year chi-square 2104 Delta models: binomial proportion positive catch Deviance 157,674 1.2101 Log-likelihood -78,837 Year chi-square 854 Delta-lognormal model: lognormal positive catches Deviance 39,963 0.8249 Log-likelihood -64,129 Yera chi-square 1240 Delta-Poisson model: Poisson positive catches Deviance 249,112 5.1419 Log-likelihood 193,660 Year chi-square 10,501 provided some interesting results in this simulation exer- cise. The binomial model component, common to both delta models, provided a highly significant year effect and indi- cated a 41% decline in abundance over the time series. The dispersion estimate indicated some overdispersion of the data with respect to the binomial distribution (Table 10). The lognormal positive catches component of the delta- lognormal model also provided a highly significant year effect and indicated a 39% decline in abundance over the time series, producing a smoothing effect similar to that observed for the lognormal model of catch per trip including zeroes. The dispersion estimate indicated some underdispersion of the data with respect to the lognormal distribution (Table 10). The product of the annual year coef- ficients from the two delta-lognormal model components, which individually indicated less decline than the unstan- dardized indices, provided final indices of abundance that declined 64% over the time series (due to the product of two positive fractional values <1 providing a even smaller value <1) — nearly identical to the unstandardized, Poisson, and negative binomial series (Fig. 9). The Poisson positive catches component of the delta- Poisson model provided a highly significant year effect and indicated a 51% decline in abundance over the time series. The dispersion estimate was much greater than 1.0, indicat- ing overdispersion of the data with respect to the Poisson model (Table 10). The product of the annual year coefficients from the two delta-Poisson model components provided in- dices of abimdance that dechned 71% over the time series, a slightly greater decrease than for the other models (Fig. 9). Note again that the delta-lognormal and delta-Poisson models share the same binomial proportion positive catch model components, and therefore annual year coefficients for this component. The decrease estimated by the delta- Poisson model was greater than that for the delta-lognor- mal because the year coefficients from the Poisson positive catch model were all smaller, and more closely matching the unstandardized positive catch series, than the comparable 668 Fishery Bulletin 101(3) Table 12 Summary of model fits for estimating indices of abundance from empirical MRFSS (Marine Recreational Fishery Sta- tistics Survey) summer flounder catch per trip (including zero catches), 1981-98. Total model degrees of freedom (df) were 102,162; for the positive catches component of the delta models, degrees of freedom were 52,507. All model fits and classification effects were highly significant (P<0.001). Dispersion estimate Criterion Value (value/df) Lognormal model Deviance 66,452 0.6505 Log-likelihood -122,989 Year chi-square 2663 Poisson model Deviance 444,657 4.3525 Log-likelihood -14,827 Year chi-square 14,053 Negative binomial model Deviance 96,698 0.9465 Log-likelihood 97,777 Year chi-square 2560 Delta models: binomisil proportion positive catch Deviance 130,341 1.2758 Log-likelihood -65,171 Year chi-square 2498 Delta-lognormal model: lognormal positive catches Deviance 36,780 0.7005 Log-likelihood -65,202 Year chi-square 1203 Delta-Poisson model: Poisson positive catches Deviance 183,019 3.4856 Log-likelihood 115,991 Year chi-square 5675 Table 13 Summary of model fits for estimating indices of abundance from empirical MRFSS (Marine Recreational Fishery Statistics Survey) Atlantic cod catch per trip (including zero catches), 1981-98. Total model degrees of freedom (df) were 20,629; for the positive catches component of the delta models, degrees of freedom were 13,160. All model fits and classification effects were highly significant (P<0.001). Dispersion estimate Criterion Value (value/df) Lognormal model Deviance 19,425 0.9416 Log-likelihood -28,697 Year chi-square 380 Poisson model Deviance 142,834 6.9239 Log-likelihood 54,501 Year chi-square 4090 Negative binomial model Deviance 21,824 1.0579 Log-likelihood 98,335 Year chi-square 323 Delta models: binomial proportion positive catch Deviance 24,997 1.2117 Log-likelihood -78,837 Year chi-square 191 Delta-lognormal model: lognormal positive catches Deviance 11,657 0.8858 Log-likelihood -17,920 Year chi-square 353 Delta-Poisson model; Poisson positive catches Deviance 75,359 5.7264 Log-likehhood 88,239 Year chi-square 2805 lognormal positive catch year coefficients over the course of the time series. For example, the year- 11 coefficient from the binomial proportion positive catches model was 0.59; the year-11 lognormal positive catches coefficient was 0.61, providing a product for the year-11 index of 0.36. In con- trast, the year-11 Poisson positive catches coefficient was 0.49, providing a product for the year-1 1 index of 0.29. When these and the other annual coefficients were scaled to the respective series means, the delta-Poisson model indicated a slightly greater decline over the time series. MRFSS standardized indices of abundance, 1981-98 All standardization models of the MRFSS catch per trip (including zero catches), for the four individual species and for all species, fitted well. In part because of the large number of observations, the overall model fits and the indi- vidual classification effects (year, mode, state, wave, and days 12 ) were all highly significant. Only the year effect chi- square statistics are tabulated because the year effect coef- ficients serve as the annual indices of abundance (Tables 11-15). The year effect was generally the second or third most important effect in the models, after mode and state. The dispersion estimates (deviance/df) for the lognormal models indicated the data were generally underdispersed with respect to the lognormal; the dispersion estimates for the Poisson models indicated overdispersion with respect to that distribution. The dispersion estimates for the nega- tive binomial models and binomial components of the delta models were generally close to 1.0, indicating appropriate model specification (Tables 11-15). As in the simulated catch-rate exercise, the lognormal standardized abundance indices generally show lower Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 669 Table 14 Summai'v of model fits for estimating indices of abundance from empirical MRFSS (Marine Recreational Fishery Sta- tistics Survey) scup catch per trip (including zero catches), 1981-98. Total model degrees of freedom (df) were 17,604; for the positive catches component of the delta models, degrees of freedom were 11,124. All model fits and classifi- cation effects were highly significant (P<0.001). Dispersion estimate Criterion Value (value/df) Lognormal model Deviance 32,270 1.8331 Log-likelihood -30,346 Year chi-square 332 Poisson model Deviance 375,924 21.3545 Log-likelihood 309,490 Year chi-square 12,094 Negative binomial model Deviance 18,668 1.0604 Log-likelihood 466,529 Year chi-square 369 Delta models: binomial proportion positive catch Deviance 22,027 1.2512 Log-likelihood -11,013 Year chi-square 174 Delta-lognormal model: lognormal positive catches Deviance 14,340 1.2891 Log-likelihood -17,225 Year chi-square 350 Delta-Poisson model: Poisson positive catches Deviance 212,250 19.0804 Log-likelihood 391,327 Year chi-square 8793 Table 15 Summary of model fits for estimating indices o f abundance from empirical MRFSS (Marine Recreational Fishery Sta- tistics Survey catch per trip for all species (including zero catches), 1981-98. Total model degrees of freedom (df) were 1,033,367; for the positive catches component of the delta models, degrees of freedom were 457,598. All model fits and classification effects were highly significant (P<0.001). Dispersion estimate Criterion Value (value/df) Lognormal model Deviance 861,881 0.8341 Log-likelihood -1,372,576 Year chi-square 7246 Poisson model Deviance 8,048,246 7.7884 Log-likelihood 277,042 Year chi-square 28,734 Negative binomial model Deviance 870,357 0.8422 Log-likelihood 3,118,822 Year chi-square 2243 Delta models: binomial proportion positive catch Deviance 1,351,532 1.3079 Log-likelihood -675,766 Year chi-square 11,867 Delta-lognormal model: lognormal positive catches Deviance 466,644 1.0198 Log-likelihood 653,838 Year chi-square 655 Delta-Poisson model: Poisson positive catches Deviance 3,773,909 8.2472 Log-likelihood 2,414,210 Year chi-square 10,785 rates of change in abundance than do the unstandardized, Poisson, or negative binomial indices, with the CV of the lognormal series about 25-50'7f of the CV of the unstan- dardized indices (Figs. 10-14). In effect, the lognormal standardization of MRFSS per trip catch rates had an unintended (and undesirable) smoothing effect on the independent annual indices abundance. The Poisson and negative binomial models generally provided interpreta- tions of the trend and annual changes in abundance very similar to those of the unstandardized indices. For bluefish, the delta-lognormal, and delta-Poisson models provided time series of indices with about the same variability and trend, but slightly different annual changes, as those from the unstandardized, Poisson, and negative binomial models. For summer flounder, Atlantic cod, scup, and all species, the delta-lognormal and delta- Poisson models provided time series of abundance indices that were more variable, with slightly different trends and annual changes, than the unstandardized, Poisson, and negative binomial series. This last result is comparable to that observed for the delta models used with the simulated data and is therefore likely due in part to model misspecifi- cation of the positive catch-per-trip component (recall that catch per trip for these examples is best characterized by the negative binomial distribution) and a comparable interaction of the binomial, lognormal, and Poisson model year coefficients. Discussion The frequency distributions of recreational fishery catch- rate data as sampled by the MRFSS are highly skewed, often with a significant proportion of zero catch observa- tions. The present study indicates that MRFSS catch rates generally are not normally or lognormally distributed 670 Fishery Bulletin 101 (3) Bluefish catch per trip indices scaled to series means • Unstandardtzed ■ Lognormal - Passon - Neg BirwfTiial 82 84 86 88 90 92 94 96 98 86 88 90 Year 96 98 Rgure 10 Indices of bluefish abundance (catch number per trip including zero catches) modeled under different error distribution assumptions, 1981-98. Atlantic cod catch per trip indices scaled to series means 2.0 1.5 1.0 0.5 • Unstandardized - Lognormal -#~ Posson X -•- Heq. Binomial ^V flj ' , ' ■ ■ E a 2.0 82 84 86 88 90 92 94 1 96 98 o ^^ Unstarwjardized 1 1.5 1.0 0.5 0.0 -V- Delta-Lognornial -O- Delta-Poisson ^^'^ J \l V / 82 84 86 88 90 Year 92 94 96 98 Figure 12 Indices of Atlantic cod abundance (catch number per trip including zero catches) modeled under different error distribution assumptions, 1981-98. Summer flounder catch per trip indices scaled to series means 2.0 15 J!-^ yA ID U C 5 ■»rA/'\^ H- \..*-^ 0.5 0.0 ' V -T- Lognormal -•- Poisson — •— Nog Biriomial ' 3 82 84 86 88 90 92 94 96 98 (0 O X 2.0 ''^ h /k K / 10 0.5 0.0 /vy ^ '\^ -O- Della-Potsson , 82 84 86 88 90 92 94 96 98 Year Figure 11 Ind ces of summer flounder abundance (catch number per trip including zero catches) modeled under differ- | ent error distribution assumptions, 1981-98. 2.5 1 2.0 1.5 1.0 g 05 1 0.0 0} c 2.0 1.5 1.0 0.5 0.0 Indices includinj distribut Scup catch per trip indices scaled to series means ^"^ Unstandardized T Lognormal -^- Poisson -■- Neg- Binomial \^^ 82 84 86 88 90 92 94 96 98 gx2 •^^ UnstarKJardaed -V- Delta-Logrxxmal -O- Defta-Poisson vA-^ ^ 82 84 86 88 90 92 94 96 9 Year Figure 13 i( scup abundance (catch number per tri ; zero catches) modeled under different errc ion assumptions, 1981-98. 3 P r Terceiro: The statistical properties of recreational catch data off the northeastern U.S. coast 671 All species catch per trip indices scaled to series means 1.5 1.0 ^HAv*"-**^^ 0.5 ■ Unslandafdlzed -▼- Lognormal -•- Poisson -•- Neg. Binomial u c ■o 82 84 86 88 90 92 94 96 98 Index of abu b by h^^^^^A>^ 05 0.0 b-^ ^^ Unstandanllzed -V- Della-Lognormal -Q- Delta -Poisson 82 84 86 88 90 92 94 96 9 8 Year Figure 14 Indices of all species abundance (catch number per trip including zero catches) modeled under different error distribution assumptions, 1981-98. but usually best characterized by the Poisson or nega- tive binomial distribution, depending on the manner in which the catch rate is configured. This finding suggests that standardization methods for MRFSS catch-rate data where Poisson (in the case of per hour rates) or negative binomial (for per trip rates) error structures are assumed would usually be more appropriate than methods where normal or lognormal error structures are assumed. The modeling of both the simulated and empirical MRFSS catch rates indicates that one may draw errone- ous conclusions about stock trends by assuming the wrong error distribution in procedures used to developed stan- dardized indices of abundance. The results demonstrate the importance of considering not only the overall model fit and significance of classification effects, but also the pos- sible effects of model misspecification, when determining the most appropriate model construction. In particular, the simulation exercise indicates that assuming a lognormal model in the calculation of indices of abundance from recreational fishery catch-per-trip data with a true under- lying negative binomial distribution will provide indices that will strongly underemphasize the true trends in the indices, and therefore in stock abundance. This underesti- mation applies equally to populations that may be declin- ing or increasing faster than the lognormally standardized indices might indicate. The MRFSS catch-per-trip indices standardized with the negative binomial model, which the descriptive statistics and goodness-of-fit results suggest should be the appropri- ate model, differ relatively little from the unstandardized indices, indicating that the model effects accounted for a low percentage of the variation in mean catch rate. The classification categories recorded in the general MRFSS sampling are broad, and even measures of angling avidity such as "angler-reported days of saltwater fishing during the previous 12 months" may not be adequate proxies for the real factors (besides stock abundance) that account for variation in recreational fishery mean catch rates. To make standardization analysis of MRFSS catch rate data poten- tially more useful, by accounting for a significantly larger part of the unexplained variance and thus providing more accurate indices of abundance, more information on the characteristics of individual fishing trips may be needed. Such information might include details on the type of equipment used, the skills, experience, avidity, and identity of the individual fishermen, and detailed temporal and spa- tial information about fishing trips. In the future, collection of detailed trip data for general recreational fisheries may be best accomplished by the identification and sampling of "test fleets" of known, individual fishermen. Acknowledgments I thank Vic Crecco of the Connecticut Department of Envi- ronmental Protection, for raising questions about the best way to calculate indices of abundance from recreational fishery catch rate data during debates over the bluefish assessments; Paul Rago of the Northeast Fisheries Science Center, for numerous discussions about statistical distribu- tions and tests; and two anonymous Fishery Bulletin ref- erees, whose comments helped improve the quality of the analyses and therefore the usefulness of the results. Literature cited Banneret, S. P., and C. B. Austin. 1983. Using frequency distributions of catch per unit effort to measure fish-stock abundance. Trans. Am. Fish. Soc. 112:608-617. Berry, D. A. 1987. Logarithmic transformations in ANOVA. Biometrics 43:439-456. Bliss, C. I., and R. A. Fisher. 1953 . Fitting the negative binomial distribution to biological data. Biometrics 9:177-200. Bradu, D., and Y. Mundlak. 1970. Estimation in lognormal linear models. J. Am. Stat. Assoc. 65:198-211. Brown, C. A. 1999. Standardized catch rates for yellowfin tuna (Thun- nus albacares) and bigeye tuna (Thunnus obesus) in the Virginia-Massachusetts (US) rod and reel fishery. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 49(3):357-369. 2001. Standardized catch rates for yellowfin tuna (Thun- nus albacares) in the Virginia-Massachusetts (U.S.) Rod and reel fishery during 1986-1999. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 52:190-201. 672 Fishery Bulletin 101(3) Brown, C. A., and J. A. Browder. 1994. Standardized catch rates of small bluefin tuna in the Virginia-Rhode Island (U.S.) rod and reel fishery. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 32(2):248-254. Brown, C. A., and C. E. Porch. 1997. A numerical evaluation of lognormal, delta-lognormal and Poisson models for standardizing indices of abundance from west Atlantic bluefin tuna catch per unit effort data (pre- hrainary results ). ICCAT ( International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 46(2):233-236. Brown, C. A., and S. C. Turner. 2001. Updated standardized catch rates of bluefin tuna, Thunnus thynnus, from the rod and reel/handline fishery off the northeast United States during 1980-1999. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 52:984-1006. Finney, D. J. 1951. On the distribution of a variate whose logarithm is normally distributed. Suppl. J. Stat. Soc. 7:155-161. Gavaris, S. 1980. Use of a multiplicative model to estimate catch rate and effort from commercial data. Can. J. Fish. Aquat. Sci. 37:2272-2275. Gulland, J. A. 1956. On the fishing eifort in English demersal fisheries. Fish. Investig. Ser. II Mar. Fish. G. B. Minist. Agric. Fish. Food 20(5), 41 p. Hilborn, R. 1985. Fleet dynamics and individual variation: why some people catch more fish than others. Can. J. Fish. Aquat. Sci. 42:2-13. Jones, C. M., D. S. Robson, H. D. Lakkis, and J. Kressel. 1995. Properties of catch rates used in analysis of angler surveys. Trans. Am. Fish. Soc. 124:911-928. Kimura, D. K. 1981. Standardized measures of relative abundance based on modeling log (c.p.u.e.), and their application to Pacific ocean perch {Sebastes alutus). J. Cons. Int. Explon Mer 39:211-218. Lawless, J. F. 1987. Negative binomial and mixed Poisson regression. Can. J. Stat. 15(3):209-225. Lo, N. C, L. D. Jacobson, and J. L. Squire. 1992. Indices of relative abundance from fish spotter data based on delta-lognormal models. Can. J. Fish. Aquat. Sci. 49:2515-2526. Manton, K. G., Woodbury, M. A., and E. Stallard. 1981. A variance components approach to categorical data models with heterogenous cell populations: analysis of spatial gradients in lung cancer mortality rates in North Carolina counties. Biometrics 37:259-269. McCullagh, P., and J. A. Nelder. 1989. Generalized linear models, 511 p. Chapman and Hall, London. NMFS (National Marine Fisheries Service). 1995. Status of the fishery resources off the northeastern United States for 1994. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NE-108, 140 p. 1996. Our living oceans. Report on the status of U.S. living marine resources, 1995. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/SPO-19, 160 p. O'Brien, L. S,, and R. K. Mayo. 1988. Sources of variation in catch per unit effort of yellow- tail flounder, Limanda ferruginea (Storer), harvested off the coast of New England. Fish. Bull. 86(1):91-108. Ortiz, M., S. C. Turner, and C. A. Brown. 1999. Standardized catch rates for bluefin tuna, Thunnus thynnus, from the rod and reel fishery off the northeast United States from 1980-1997. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. Vol. 49(2 ):254-269. Pennington, M. 1983. Efficient estimators of abundance, for fish and plank- ton surveys. Biometrics 39:281-286. Power, J. H., and E. B. Moser. 1999. Linear model analysis of net catch data using the negative binomial distribution. Can. J. Fish. Aquat. Sci. 56:191-200. Robson, D. S. 1966. Estimation of the relative fishing power of individual ships. Comm. N.W. Atl. Fish. Res. Bull. 3:5-14. SAS Institute. 2000. SAS OnlineDoc, version 8. SAS Institute Inc., Cary NC. http://www.sas.com/ts. Searle, S. R. 1987. Linear models for unbalanced data, 536 p. John Wiley and Sons, Inc., New York, NY. Smith, B. D. 1999. A probabilistic analysis of decision-making about trip duration by Strait of Georgia sport anglers. Can. J. Fish. Aquat. Sci. 56:960-972. Smith, S. J. 1990. Use of statistical models for the estimation of abun- dance from groundfish trawl survey data. Can. J. Fish. Aquat. Sci. 47:894-903. 1996. Analysis of data from bottom trawl surveys. NAFO Scientific Council Studies 28:25-53. Snedecor, G W, and W G Cochran. 1967. Statistical methods, 593 p. Iowa State Univ. Press, Ames, lA. Sokal, R. R., and F J. Rohlf 1981. Biometry, 859 p. WH. Freeman and Co., New York, NY. Taylor, C. C. 1953. Nature of the variability in trawl catches. Fish. Bull. 54:145-166. Turner, S. C, C. A. Brown, and H. Huang. 1997. Standardized catch rates of small bluefin tuna, Thun- nus thynnus, from the U.S. rod and reel fishery off Vir- ginia-Rhode Island in 1980-1995. ICCAT (International Commission for the Conservation of Atlantic Tunas) Col. Vol. Sci. Pap. Vol. 46(2):295-310. USDOC (U.S. Department of Commerce). 1992. Marine recreational fishery statistics survey, At- lantic and Gulf coasts, 1990-1991, 275 p. Current Fisher- ies Statistics 9204. 2001. Marine recreational fishery statistics survey. U.S. Dep. Commer, Washington, DC. http://www.st.nmfs.gov/ stl/recreational/index. (Accessed 30 January 2001.1 Williams, D. A. 1976. Improved likelihood ratio tests for complete contin- gency tables. Biometrika. 63:33-37. 673 Abstract — Spatial variation in demo- graphic parameters of the red throat emperor (Lethrinus miniatus) was examined among 12 coral reefs in three geographic regions (Townsville, Mackay, and Storm Cay) spanning over 3° of latitude of the Great Bar- rier Reef, Australia. Estimates of demographic parameters were based on age estimates from counts of annuli in whole otoliths because there was no significant difference in age estimates between whole and sectioned otoliths. There were significant regional differ- ences in age structures, rates of somatic and otolith growth, and total mortality. The Townsville region was character- ized by the greatest proportion of older fish, the smallest maximum size, and the lowest rates of otolith growth and total mortality. In contrast the Mackay region was characterized by the highest proportion of younger fish, the largest maximum size, and the highest rates of otolith growth and total mortality. Demographic parameters for the Storm Cay region were intermediate between the other two regions. Historic differ- ences in fishing pressure and regional differences in productivity are two alternative hypotheses given to explain the regional patterns in demographic parameters. All demographic param- eters were similar among the four reefs within each region. Thus, subpopula- tions with relatively homogeneous demographic parameters occurred on scales of reef clusters. Previous stud- ies, by contrast, have found substantial between-reef variation in demographic parameters within regions. Thus spa- tial variation in demographic param- eters for L. miniatus may differ from what is assumed typical for a coral-reef fish metapopulation. Scales of spatial variation in demography of a large coral-reef fish— an exception to the typical model? Ashley J. Williams School of Marine Biology and Aquaculture and CRC Reef Research Centre James Cook University Townsville, Queensland. 4811, Australia E-mail address, astiley williams(a'|cu edu au Campbell R. Davies Bruce D. Mapstone CRC Reef Research Centre James Cook University Townsville, Queensland 4811, Australia Garry R. Russ School of Marine Biology and Aquaculture James Cook University Townsville, Queensland, 481 1, Australia Manuscript approved for publication 17 December 2002 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:673-683 (2003). Estimates of demographic parameters, such as growth and mortahty rates, are fundamental to the understanding of a species population dynamics and for predicting responses of populations to exploitation. Processes affecting popu- lation dynamics operate at a number of spatial and temporal scales (Levin, 1992) and can result in subpopulations with distinct demographics. Differences in demography between populations may suggest geographic or reproduc- tive isolation (or both) and as such have been used in stock identification for fisheries assessment and manage- ment purposes (e.g. Begg et al., 1999). Identifying the "unit stock" has been the primary focus of studies of spatial structure of harvested populations in most fisheries. Knowledge of spatial structure within a unit stock is impor- tant for both fisheries management, because potential yields may vary spatially within a population (Caddy, 1975), and for conservation, in order to maintain intraspecific diversity (Niel- sen, 1998). Hence, it is important to estimate demographic parameters over a range of temporal and spatial scales to determine the scale(s) at which the parameters vary significantly (Caley et al., 1996) and, therefore, to infer which scales are of greatest importance for assessment and management purposes (Sale, 1998). Most coral-reef fish exist as metapop- ulations of sedentary adult populations linked by pelagic larval dispersal (Sale, 1998). Consequently, adult populations of reef fish are commonly spatially seg- regated and may be exposed to different environmental, biological, and ecologi- cal processes, resulting in spatial dif- ferences in demographic parameters at a range of spatial scales. Relatively few studies, however, have focussed on spatial variation in demographic pa- rameters of harvested species of coral- reef fish. Those that have, have gener- ally focussed on spatial scales within individual reefs or among reefs within a single region (e.g. Ferreira and Russ, 1995; Hart and Russ, 1996; Newman et al., 1996). Comprehensive multiscale approaches are rare (but see Adams et al., 2000; Meekan et al., 2001). The spatial structure of coral-reef populations has generated considerable interest in terms of the use of spatial closures, or marine protected areas (MPAs), as an effective tool for their management (Roberts and Polunin, 674 Fishery Bulletin 101(3) 1991). However, the lack of information about the stock structure of, and connectivity among, adult populations has hindered MPA design (Walters and Bonfil, 1999). Conser- vation management of the Great Barrier Reef (GBR) has included the use of spatial closures of areas to activities, including fishing, for more than 15 years. The majority of spatial closures to line fishing are of individual coral reefs or groups of reefs. This spatial management strategy is underpinned by the assumption of the metapopulation model of coral-reef fish described above. That is, closing individual reefs to fishing will protect the adult popula- tions on those reefs, and potentially provide a source of larvae to areas open to fishing. Management of line fishing on the GBR currently includes bag limits for recreational fishermen and minimum-size restrictions that are uniform for all fishermen and across the entire area of the fishery. Such management regulations are based on the assump- tion that the demography of target species does not vary substantially over the species range and on the assump- tion that that populations on the GBR represent a single, homogeneous stock. The red throat emperor (Lethrinus miniatus) (also known as the trumpet emperor) is a relatively long-lived (>20 years) (Loubens, 1980: Brown and Sumpton, 1998) member of the Lethrinidae and has a restricted distribu- tion in the western Pacific and eastern Indian Oceans (Car- penter and Allen, 1989). On the GBR it is the second most important demersal species in a multispecies line fishery, contributing up to 1000 metric tons annually to the com- bined commercial and recreational catch (Mapstone et al.^; Higgs-). As with many tropical lethrinids, information on the biology and ecology of L. miniatus is scarce. The limited data available indicate that L. miniatus is usually associ- ated with coral reefs, but that it is also commonly caught in deeper water, in sand, and rubble areas between reefs (Carpenter and Allen, 1989; Newman and Williams. 1996; Williams and Russ'^). The habitat of juvenile L. miniatus is unknown, but Williams and Russ^ have suggested that juveniles may occupy the deeper rubble areas adjacent to reefs. Like some other coral-reef fish, L. miniatus is thought to form large aggregations associated with spawning ' Mapstone, B. D., J. P. McKinlay. and C. R. Davies. 1996. A description of commercial reef line fishery logbook data held by the Queensland Fisheries Management Authority. Report to the Queensland Fisheries Management Authority from the Cooperative Research Centre for the Ecologically Sustainable Development of the Creat Barrier Reef, and the Department of Tropical Environmental Studies and Geography. James Cook University, Queensland. Australia, 480 p. lAvailable from the Queensland Fisheries Service, G.P.O. Box 46, Brisbane. Queensland, Australia 4001.1 - Higgs, J. 2001. Experimental recreational catch estimates for (Queensland residents. Results from tbi' 1999 diary round. KFISH technical report no. .'!. Queensland Fisheries Sei-vice, Aus- tralia, 62 p. lAvailable from the Queensland Fisheries Service, G.RO. Box 46, Bri.sbane, Queensland, Australia 4001.1 ■' Williams, D. McB., and G. R. Russ. 1994. Review of data on fishes of commercial and recreational fishing interest on the Great Barrier Reef Report to the Great Barrier Reef Marine Park Authority, 103 p. [Available from the Great Barrier Reef Marine Park Authority, P.O. Box 1379, Townsville, Queensland, Australia, 4810.1 (RusselH). These available data suggest that L. miniatus adults have the capacity to move among individual reefs on the GBR. This movement pattern contrasts with informa- tion on movement patterns of other large coral-reef spe- cies such as the coral trout (Plectropomus leopardus) (also known as the leopard coral grouper, Heemstra and Randall, 1993) where adults show limited movement within a single reef and very restricted movements between reefs (Davies, 1995). It also contrasts with movement patterns of the ma- jority of coral-reef fish, where adults are known to have very restricted home ranges and display little, if any, move- ment between reefs (Lewis 1997; Sale, 1998). Therefore the relevant spatial scale affecting demographic parameters of L. miniatus may be larger than an individual reef and thus is different from that for most "typical" coral-reef fish. The central objective of this study was to determine how the spatial patterns in demogi'aphy of large, more mobile reef fish differ from smaller site-attached reef-fish species. To achieve this we used validated age estimates to examine spatial variation in demographic parameters of populations of L. miniatus across two spatial scales most relevant to as- sessing and managing the species on the GBR: 1 ) among individual reefs within regions and, 2) among geographic regions. Specifically, we estimated age structures, growth, mortality, and otolith growth rates for among four reefs (all closed to fishing) within each of three geographic regions spanning over 500 km (over 3° of latitude) of the GBR. Materials and methods Collection methods Samples of L. miniatus were collected from three geo- graphic regions of the GBR as part of a large-scale manip- ulative experiment to examine the effects of line fishing on the GBR (Davies et al.^; Mapstone et al.'^). The three regions cover most of the distribution of L. miniatus on the GBR (Fig. 1), which is restricted to the southern 50% of the GBR. Within each region L. miniatus were collected from six individual reefs. Four of these reefs were zoned "Marine National Park B" and were closed to all forms of fishing (referred to as "closed reefs" in this article) whereas the other two reefs were zoned "General Use B" and were '' Russell, M. 2001. Spawningaggregationsof reef fishes on the Great Barrier Reef: implications for management. Report from the Great Barrier Reef Marine Park Authority. 37 p. [Available from the Great Barrier Reef Marine Park Authority, P.O. Box 1379. Townsville, Queensland, Australia. 4810|. = Davies, C. R., B. D. Mapstone, A. Ayling, D. C. Lou, A. Punt, G. R. Russ, M. A. Samoilys, A. D. M. Smith. D. J. Welch, and D. McB. Williams. 1998. Effects of line fishing experiment 1995-1997: project structure and operations. -Supplementary to progress report. CRC Reef Research Centre, Townsville, Australia, 28 p. lAvailable tioni the CRC Reef Research Centre. PO. Box 772, Townsville, Queensland, Australia 4810). 6 Mapstone, B. D.. C. R. Davies. I). C. Lou, A. E. Punt, G. R. Russ, D. A. .1. Ryan. A. D. M. Smith, and D. McB. Williams. 1998. Effects ofline fishing experiment 199.5-1997: progress report. CRC Reef Research Centre, 86 p [Available from the CRC Reef Research Centre, P.O. Box 772, Townsville, Queensland, Austraha 4810). Williams et al.: Scales of spatial variation in demography of a large coral-reef fishi 675 140°00E fv 150°OOE io°oos- \ " imsl imsvilleV,^ 20°oos- Mackay\ Gladstones. _L| Op Reef Faraday Reef QowReef Yankee Reef / c Tov 20-136^^ BaxReef 20-137 20-142^ 21-130 / 21-131 21-133 \ 21-132 AUSTRALIA Figure 1 Location of reefs sampled for L. miniatus from October 1995 to January 1996 in the 1) Townsville, 2 ) Mackay, and 3 ) Storm Cay regions of the Great Barrier Reef, AustraUa. Reefs maps indicate the relative position of the four reefs closed to fishing that were sampled in regions 1, 2, and 3. open to line and spear fishing (referred to as "open reefs"). Fishing had been prohibited from the closed reefs for at least seven years prior to sampling. Each reef was sampled for two days by the same four commercial line fishermen using gear and sampling designs standardized across all reefs (Davies et al.''). Fork length (FL) was measured to the nearest millimetre immediately upon capture. Sagittal otoliths were removed from frozen frames in the laboratory, cleaned of any residual material, dried, and weighed to the nearest 0.1 mg. A total of 1015 L. miniatus were collected from the four closed reefs in each region between October 1995 and Janu- ary 1996. Sample sizes from the open reefs were small and mortality and growth estimates from these reefs were unreliable. Therefore, these samples were used only to in- crease the sample size of older fish for a comparison of the two methods for reading otoliths (whole and sectioned). Comparison of otolith reading methods The annual periodicity of opaque increment formation in L. miniatus otoliths has been validated (Brown and Sumpton 1998). A subsample of 355 L. miniatus otoliths from both open and closed reefs was used to assess whether readings of whole otoliths provided age estimates similar to those from sectioned otoliths, but at substantially lower cost (in time). Otolith weight was used to select a broad range of age classes for this assessment on the assumption that otolith weight was a coarse indicator of age, thus avoiding the need to preread otoliths to obtain a sample covering all age classes. Each otolith in the subsample was read, both whole and sectioned, on three separate occasions in random order with no prior knowledge of collection date, location, or fish size. For consistency, the right otolith was chosen to estimate the age of all fish unless it was missing or damaged, in which case the left one was used. Otoliths to be read whole were placed in a small black dish of immer- sion oil and examined under reflected light with a stereo dissecting microscope. Counts of opaque increments were made from the nucleus to the dorsoposterior edge on the convex face of the otolith. For otoliths from older fish it was necessary to rotate the otolith approximately 45° to clearly observe increments on the otolith margin. Otoliths to be sectioned were embedded in epoxy resin and cut transversely, adjacent to the anterior side of the nucleus with a Buehler Isomet low-speed saw. The poste- rior portion of the otolith was retained and mounted on a glass microscope slide with Crystalbond adhesive. A second transverse cut adjacent to the posterior side of the nucleus resulted in a thin section, incorporating the otolith nucleus, remaining on the slide. Otolith sections were then ground on 800- and 1200-grade sandpaper to remove saw marks and a single drop of immersion oil was placed on sections to fill surface irregularities. Otolith sections were examined under a stereo dissecting microscope with reflected light and a black background. Counts of opaque increments were 676 Fishery Bulletin 101(3) made from the nucleus to the proximal surface, along the dorsal margin of the sulcus acousticus. The precision of age estimates from whole and sectioned otoliths was calculated by using the index of average per- cent error (Beamish and Fournier 1981). The estimates of age from whole and sectioned otoliths were compared by a paired ?-test. Difference in bias between the two reading methods was observed by plotting the difference between the two readings (sectioned age minus whole age) against sectioned age, based on the assumption that sectioned age provided the best estimate of true age (Beamish 1979). The results from this comparison indicated no significant dif- ference between whole and sectioned otolith readings and there was no discernible difference in bias in the plot. As a result, all remaining otoliths were read whole for greater efficiency. Age estimates from whole otoliths were accepted and used in subsequent analyses when counts from the first two readings agreed. If the counts differed, otoliths were read a third time. The otolith was excluded from sub- sequent analyses if no two counts agreed, but included if any two counts agreed. Comparison of demographic parameters The central objective of this study was to estimate the vari- ation in demographic parameters of L. miniatus, specifi- cally otolith and somatic growth rates, age structure, and mortality, at different spatial scales. In the first instance, parameters were compared among the four reefs within each region to estimate the magnitude of variation at the inter-reef scale. Data were then pooled from individual reefs within each region to generate regional parameter estimates, which were used to estimate the magnitude of variation at the regional spatial scale. The relationship between otolith weight and age (rep- resenting otolith growth) was examined for each reef by least-squares regression analysis, with otolith weight as the dependent variable. The relationship was compared among reefs within each region and among regions by us- ing analysis of covariance (ANCOVA). Reef-specific age-frequency distributions were construct- ed for all reefs. Multidimensional contingency tables were used to compare age frequencies among reefs within re- gions and among regions. Age classes 4 years and younger and age classes 10 years and older were pooled into 4 and 10* age classes, respectively, because of low frequencies in the tails of the age distributions. As a result, the analyses included a total of seven age classes. Age-based catch curves (Ricker, 1975) were used to esti- mate the instantaneous rate of total mortality (Z) at each reef expres.sed on an annual basis. The number offish in each age class was regressed against the corresponding age, and the descending slope provided an estimate of Z. Regressions were fitted from the first age class that was fully selected by the sampling gear through to the oldest age class that was preceded by no more than two consecu- tive zero frequencies. As a result, the age range used to estimate mortality varied slightly among reefs. Mortal- ity rates were compared among reefs within regions and among regions by using ANCOVA. The von Bertalanffy growth function (VBGF) provided the best fit to length-at-age data for most reefs according to the parameter estimates of the Schnute (1981) growth function. For consistency, and to enable spatial compari- sons of growth, the VBGF was used to estimate growth parameters for each reef and region: L, = L^[1-^ -KH-lr,)] where L, = the fork length at age t; L^ = the mean asymptotic fork length; K = the rate at which L^ is approached; and ^Q = the age at which fish have a theoretical length of zero. It was difficult to obtain a reliable estimate of initial growth because the youngest fish collected was 2 years old. There are also no published size-at-age data for larval or juve- nile L. miniatus, or any other lethrinid. We constrained the VBGF parameter t^ to zero to provide a better description of the likely early growth of L. miniatus. This procedure also allowed growth curves to be compared among reefs within regions and among regions by using 959'f confidence regions of the VBGF parameters L^ and K described by Kimura(1980). Results Comparison of otolith reading methods Age estimates from whole and sectioned otoliths did not vary significantly over the range of ages between 2 and 21 years (^00.5 2 3.54-'^-'^^' ^=0.73). That is, for each age class estimated from sectioned otoliths, the average difference between whole and sectioned otolith readings did not differ significantly from zero (Fig. 2). The index of average percent error was very low for whole ( 1.6^* ) and sectioned ( 1.4%) otolith readings, indicating that otolith readings for both methods were highly repeatable. This low index was reflected in the agreement of at least two age estimates for all whole otoliths, and hence no otoliths were excluded from analyses. Otolith growth There was a significant positive linear relationship between otolith weight and age for all reefs, with regression coef- ficients ranging from 0.64 to 0.90. ANCOVA revealed that the slope of this relationship was not significantly different among reefs within each region (Townsville: F., .j^g=1.91, P=0.13; Mackay: F.^.^^^=1.02, P 0.38; Storm Cay: Fl~i^=l.?>h. P=0.20). Thus, otolith weight and age data were pooled for each region to compare the region-specific relationships between otolith weight and age (Fig. 3). The slopes of the region-specific relationships differed significantly among all regions (F.^y4g=28.9, P<0.001). The average growth in otolith weight was greater in the Mackay region (26.91 mg/yr) than in the Storm Cay region (24.48 mg/yr), and was least in the Townsville region (19.50 mg/yr). Williams et al : Scales of spatial variation in demography of a large coral-reef fisfi 677 10 ^ 0.5 o 5 I 6 years) in the Towns- ville region than in the Mackay and Storm Cay regions (Fig. 4). However the oldest fish were from the Storm Cay region, where a small number offish persisted in the older age-classes up to 19 years of age. The relative abundances of age classes 4 and 5 were greater in the Mackay region than in the Townsville and Storm Cay regions (Fig. 4). Mortality Estimates of annual total mortality rates (Z) for individ- ual reefs were generally similar among reefs within each 600 500 400 300 200 100 A Townsville « * : :.r' m 500 fc ^ 4(10 U) % 300 r ■M) n L) 100 0 B Mackay 600 C storm Cay « 500 400 • It^ ♦ 300 ,llir 200 .#• 100 n r* 0 5 10 15 20 Age (years) Figure 3 Least-squares linear regression of otolith weight on age for L. miniatus from three regions of the Great Barrier Reef: (A) Townsville, (B) Mackay, and (C) Storm Cay. region, with the exception of the Storm Cay region where the estimated Z appeared much lower for reef 21-131 than for other reefs in that region (Table 1). ANCOVA indicated no significant difference in mortality among reefs in any region (Townsville: F3 3o=0.80, P=0.50; Mackay: F3 2o=0.08, P=0.97; Storm Cay: P.j ,5=1.14, P=0.37). Therefore mortal- ity rates were estimated for each region from the pooled age structures for all reefs within each region (Fig. 5). A comparison among regions of the regression slopes from the pooled age structures indicated significant differences among regions {F2^g=T-ll, P=0.005). Tukey's multiple comparison tests revealed that the estimated Z for the Townsville (Z=0.42) and Mackay (Z=0.71) regions differed significantly, whereas the estimate from the Storm Cay region (Z=0.60l did not differ significantly from either Townsville or Mackay. 678 Fishery Bulletin 101(3) A Townsvllle n = 347 B Mackay 10 12 14 16 18 20 n = 370 0) 40 2 4 6 8 10 12 14 16 18 20 C Storm Cay n = 298 2 4 b b 10 12 14 16 18 20 Age (years) Figure 4 Age structures for L. miniatus from three regions of the Great Barrier Reef: (A) Townsvllle, (B) Mackay, and (C) Storm Cay. Age structures were pooled from four closed reefs within each region. Somatic growth Estimates of VBGF parameters varied considerably among reefs within the Townsvllle and Mackay regions but in the Storm Cay region, estimates of L^ and particularly K' were very similar (Table 1). Examination of BS^f confidence regions for VBCJF parameters for individual reefs (Fig. 6) indicated considerable uncertainty in the estimates of both K and L.^ and no clear differentiation among reefs within regions. The similarity in VBGF parameters for individual reefs within the Storm Cay region was particularly evident from the 95% confidence regions. In both the Townsvllle and Mackay regions, three reefs showed overlap in 95% J= 2 TownsvNIe Mackay 0 2 4 6 8 10 12 14 16 18 Age (years) Figure 5 Age-based catch curves fori, miniatus from three regions of the Great Barrier Reef • = Townsvllle; ■, Mackay; ▲ = Storm Cay. Catch curves were derived from pooled age structures from four closed reefs within each region. Open data points were not used in the regressions. confidence regions, whereas only a single reef in each region appeared to have significantly different VBGF parameters from the others (Fig. 6). Given the lack of differentiation in growth among reefs, the data from individual reefs were pooled for each region to examine regional patterns in growth. VBGF parameters varied significantly among regions (Table 1) with no over- lap in the 95% confidence regions (Fig. 7). It appeared that L. miniatus in the Mackay region attained a larger average asymptotic size (L^=472.21 mm FL) than in the Storm Cay region (L^=462.83 mm FL), where in turn these fish grew larger than fish in the Townsvllle region (L^=453.36 mm FL). It should be noted that the constrained fitting of the VBGF itg set to zero) provided a conservative estimate of regional variation in growth, and regional differences were considerably larger when the VBGF parameter £ O 600 U. 500 400 300 200 100 0 C Storm Cay ^ «♦ 500 400 300 200 100 0 490 480 470 460 450 B Mackay D 95% confidence regions Mackay 5 10 15 20 Age (years) 0 34 0 39 0 44 0 49 0 54 K (per year) Figure 7 Von Bertalanffy growth function { VBGF ) fitted to size-at-age data for L. miniatus from three regions of the Great Barrier Reef (A) Townsville, (B) Mackay, and (C) Storm Cay. The VBGF parameter Jq is constrained to zero in all cases. Dashed lines represent predicted growth between 0 and 2 years of age. (D) 95% confidence regions for the VBGF parameters K and L^ for fit to pooled data from four reefs within each of the three regions. accuracy in relation to those estimated from sectioned oto- liths. In contrast, Brown and Sumpton (1998) concluded that whole otoliths from larger and presumably older L. miniatus underestimated age by up to 40% with respect to sectioned otoliths. The discrepancy between studies may be due to differences in the techniques used to count incre- ments in whole otoliths. It was noted early in the present study that otoliths from older fish needed to be rotated to reveal a number of increments close to the otolith margin. By not using this technique Brown and Sumpton (1998) may have underestimated ages from whole otoliths of older fish. Readings from whole otoliths have been shown to consistently underestimate the age of a number of reef fish species (e.g. Ferreira and Russ, 1994; Newman et al., 2000) resulting in biased estimates of mortality and sub- sequent yield estimates (Newman et al., 2000). The results from this study suggest that whole otoliths are adequate for estimating the age of L. miniatus and that estimates of demographic parameters presented in the present study were not biased by underestimates of age. The spatial patterns in the demography of L. miniatus de- scribed in the present study are based on data collected from a single survey in one year, thus leaving the temporal stabil- ity of the patterns open to question. Continued monitoring of the populations will be required to determine the stability of the patterns, and focussed stock structure studies are required to determine the most likely causal mechanism(s) of the patterns. Notwithstanding the need for this work, the significant regional differences in demographic parameters found in the present study suggest different levels of produc- tivity of L. miniatus populations in each region. Consequent- ly, there is the potential for less productive populations to be overfished, even where the fishing effort for the stock as a whole is managed at sustainable levels (Caddy, 1975; Sheperd and Brown, 1993). This argues for assessments and management of L. miniatus stocks to explicitly consider the regional structure in demography in order to meet both sustainable use and conservation objectives for the Great Barrier Reef World Heritage Area overall and on a regional basis. Furthermore, this study highlights a more general need for the use of multiscale sampling and analyses offish populations to understand the relative importance of the processes affecting demographic parameters, and the scales at which these processes operate. Acknowledgments We acknowledge financial support from the Cooperative Research Centre for the Great Barrier Reef World Heritage Area, the Fisheries Research and Development Corpora- tion, the Great Barrier Reef Marine Park Authority, and the CRC Reef Research Augmentative Grant Scheme. The VBGF ConfRegion program developed by J. Kritzer, CRC Reef Research Centre, was used to estimate the VBGF 95% confidence regions. We would like to thank Robin Stewart, Mary Petersen, and the ELF field team for assistance in the collection and processing of the fish from which this 682 Fishery Bulletin 101(3) work arose. We would also like to thank three anonymous reviewers for comments that improved the quality of this manuscript. This manuscript is a contribution from the CRC Reef Effects of Line Fishing Project, CRC Reef Research Centre, Townsville, Australia. Literature cited Adams, S., B. D. Mapstone, G. R. Russ, and C. R. Davies. 2000. Geographic variation in the sex ratio, sex specific size, and age structure of Plectropomus leopardus (Serranidae) between reefs open and closed to fishing on the Great Bar- rier Reef. Can. J. Fish. Aquat. Sci. 57:1448-1458. Beamish, R. J. 1979. Differences in age of Pacific hake (Merluccius produc- tus) using whole otoliths and sections of otoliths. J. Fish. Res. Board Can. 36:141-151. Beamish. R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Begg, G. A., J. A. Hare, and D. D. Sheehan. 1999. The role of life history parameters as indicators of stock structure. Fish. Res. (Amst.) 43:141-163. Brown, I. W., and W. D. Sumpton. 1998. Age, growth and mortality of redthroat emperor Leth- rinus miniatus (Pisces: Lethrinidae) from the southern Great Barrier Reef Queensland, Australia. Bull. Mar. Sci. 62:905-917. Caddy, J. F 1975. Spatial model for an exploited shellfish population, and its application to the Georges Bank scallop fishery. J. Fish. Res. Board Can. 32:1305-1328. Caley, M. J., M. H. Carr, M. A. Hixon, T P Hughes. G. P Jones, and B. A. Menge. 1996. Recruitment and the local dynamics of open marine population. Annu. Rev. Ecol. Syst. 27:477-500. Carpenter, K. E., and G. R. Allen. 1989. FAO species catalogue. Emperor fishes and large-eyed breamsof the world (family Lethrinidae). An annotated and illustrated catalogue of lethrinid species known to date. FAOFish. Synop. 125, 118p. FAO, Rome. Conover, D. O. 1992. Seasonality and the scheduling of life history at differ- ent latitudes. J. Fish Biol. 41(suppl. B):161-178. Davies, C. R. 1995. Patterns of movement of three species of coral reef fish on the Great Barrier Reef PhD diss., 203 p. Dep. Marine Biology, James Cook Univ., Townsville, Queensland, Australia. DeMartini, E. E. 1993. Modelling the potential of fishery reserves for manag- ing Pacific coral reef fishes. Fish. Bull. 91:414-427. Doherty P. J. 1983. Tropical territorial damselfishes: is density limited by aggregation or recruitment? Ecology 64:176-190. Doherty, P. J., and A. Fowler 1994. Demographic consequences of variable recruitment to coral reef fish populations: a congeneric comparison of two damselfishes. Bull. Mar Sci. 54:297-313. Doherty P. J., S. Planes, and P. Mather. 1995. Gene flow and larval duration in seven species offish from the Great Barrier Reef Ecology 76:2373-2391. Dudgeon, C. L., N. Gust, and D. Blair 2000. No apparent genetic basis to demographic differences in scarid fishes across continental shelf of the Great Barrier Reef Mar Biol. 137:1059-1066. Ferreira, B. P., and G. R. Russ. 1994. Age validation and estimation of growth rate of the coral trout, Plectropomus leopardus. (Lacepede, 1802) fi'om Lizard Island, Northern Great Barrier Reef Fish. Bull. 92:46-57. 1995. Population structure of the coral trout, Plectropomus leopardus (Lacepede 1802), on fished and unfished reefs off Townsville, Central Great Barrier Reef Aust. Fish. Bull. 93:629-642. Hart, A. M., and G. R. Russ. 1996. Response of herbivorous fishes to crown-of-thorns starfish Acanthaster planci outbreaks. III. Age, growth, mortality and maturity indices of Acanthurus nigrofuscus. Mar Ecol. Prog. Sen 136:25-35. Heemstra, P. C, and J. E. Randall. 1993. FAO species catalogue, vol. 16. Groupers of the world (family Serranidae, subfamily Epinephelinae). An anno- tated and illustrated catalogue of the grouper, rockcod, hind, coral grouper, and lyretail species known to date. FAO Fish Synop. 125, 292 p. FAO, Rome. Jones, G. P. 1987. Competitive interactions among adults and juveniles in a coral reef fish. Ecology 68:1534-1547. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967. Lewis, A. R. 1997. Recruitment and post-recruit immigration affect the local population size of coral reef fishes. Coral Reefs 16: 139-149. Loubens, G. 1980. Biologie de quelques especes de poissons du lagon Neo- Caledonien. 111. Croissance Cah. I'lndo-Pac. 2:101-153. Lough, J.M. 1994. Climate variation and El Nino-Southern Oscillation events on the Great Barrier Reef 1958 to 1987. Coral Reefs 13:181-195. Meekan, M. G., J. L. Ackemian, and G. M. Wellington. 2001. Demography and age structures of coral reef dam- selfishes in the tropical eastern Pacific Ocean. Mar. Ecol. Prog. Sen 212:223-232. Newman, S. J., M. Cappo, and D. McB. Williams. 2000. Age, growth, mortality rates and corresponding yield estimates using otoliths of the tropical red snappers, Lut- janus ery'thropterus, L. malabancus and L. sebae, from the central Great Barrier Reef Fish. Res. 48:1-14. Newman, S. J., D. McB. Williams. 1996. Variation in reef associated assemblages of the Lut- janidae and Lethrinidae at different distances offshore in the central Great Barrier Reef Environ. Biol. Fishes 46: 123-138. Newman, S. J., D. McB. Williams, and G. R. Russ. 1996. Variability in the population structure of Luljanus adetii (Castelnau, 1873) and L. quinquelineatus (Bloch, 1790) among reefs in the central Great Barrier Reef Aust. Fish. Bull. 94:313-329. Nielson, J. L. 1998. Population genetics and the conservation and man- agement of Atlantic salmon (.Salmo salar). Can. J. Fish. Aquat. Sci. 55(suppl. 1):145-152. Ricker.W. E. 1975. Computation and interpretation of biological statis- Williams et al.: Scales of spatial variation in demography of a large coral-reef fish 683 tics offish populations. Bull. Fish. Res. Board Can. 191, 382 p. Roberts, C. M., and N. V. C. Polunin. 1991. Are marine reserves effective in management of reef fisheries? Rev. Fish Biol. Fish. 1:65-91. Russ, G. R., A. C. Aleala, and A. S. Cabanban. 1992. Marine reserves and fisheries management on coral reefs with preliminary modeling of the effects of yield per reci-uit. Proc. of the 7"^ int. coral reef svmp. 2, p. 988-995. Univ. Guam, Marine Lab., Mangilao, Guam. Sale, P. F. 1998. Appropriate spatial scales for studies of reef-fish ecology. Aust. J. Ecol. 23:202-208. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. Sheperd, S. A., and L. D. Brown. 1993. What is an abalone stock: implications for the role of refii- gia in conservation. Can. J. Fish. Aquat. Sci. 50:2001-2009. Shulman, M. J., and E. Bermingham. 1995. Early life histories, ocean currents, and the population genetics of Caribbean reef fishes. Evolution 49:897-910. van Herwerden, L., J. Benzie, and C. R. Davies. In press. Microsatellite variation and population genetic structure of red throat emperor {Lethrinus miniatus) in the Great Barrier Reef Australia. J. Fish Biol. Walters, C. J., and R. Bonfil. 1999. Multispecies spatial assessment models for the Brit- ish Colombia groundfish trawl fishery. Can. J. Fish. Aquat. Sci. 56:601-628. 684 The occurrence of yellowfin tuna iThunnus albacares) at Espiritu Santo Seamount in the Gulf of California A. Peter Klimley Salvador J. Jorgensen Bodega Marine Laboratory University of California, Davis Westside Road Bodega Bay, California 94923 Present address (for A. P. Klimley): Department of Wildlife, Fisfi, and Conservation Biology University of California Davis Davis, California 95616 E-mail address (for A. P. Klimley): apklimley@ucdavis.edu, Arturo Muhlia-Melo Centre de Investigaciones Biologicas del Baja Norte Apartado Postal 128 La Paz, Mexico later identifying them from these tags. This method results in the removal of individuals from the population and yields a percentage of individuals that have either left the area or have been captured (Holland et al., 1999). Detecting coded ultrasonic tags by an automated monitor provides additional information because marked individu- als can be detected repeatedly over a period of time. However, fewer tags can be deployed because of their greater cost. We used this method to reveal synchronicity among visits of yellowfin tuna, time of visits, and duration of visits at the Espiritu Santo Seamount in the Gulf of California. Methods Sallie C. Beavers Bodega Marine Laboratory University of California, Davis Westside Road Bodega Bay, California 94923 Pelagic fishes are not evenly dispersed in the oceans, but aggregate at dis- tinct locations in this vast and open environment. Nomadic species such as mackerels, tunas, and sharks form assemblages at seamounts (Klimley and Butler, 1988; Fontenau, 1991). Fishermen have recognized this behavior and have placed moorings with surface buoys in deep waters to provide artificial landmarks, around which fish concentrate and are more easily captured. These fish aggregating devices (termed FADs) are common in the tropical oceans (see review, Hol- land, 1996). In a sense, it may only be the larger size that separates a sea- mount from a man-made FAD. Fish may aggregate at seamounts for very different reasons. The opportunity to feed is greater because biomass at all trophic levels, from primary producer to apex consumer, is greater than in the open ocean (Boehlert and (Jenin, 1987). The disturbance of flow by the seamount creates eddies downstream that retain nutrients critical to the growth of phytoplankton, and this enrichment supports a greater abun- dance of consumers from zooplankton to apex predators. The dipole nature of seamount magnetic fields and the out- ward radiating valleys and ridges of magnetic minimums and maximums might provide landmarks in oceanic landscape that fish use as a reference to guide migration (see discussion of mag- netic "topotaxis" in Klimley, 1993). Yel- lowfin (Thunnus albacares) and bigeye (Thiinniis obesus) tunas do not reside long at the Cross Seamount near Ha- waii, an observation inconsistent with the theory that tunas feed on prey that remain aggregated at the site; rather their rapid passage suggests that the site is a landmark used to guide migra- tions (Holland et al., 1999). Adult yel- lowfin tuna also stay briefly (<5 min) at FADs off Kaena Point, Oahu (Klimley and Holloway 1999). Describing the degree of residency of pelagic fishes at different geographic locations helps ascertain whether the affinity to seamounts and FADs is com- mon throughout the oceans. Holland et al. (1999) determined the rates of dispersion of tuna by attaching unique tags to individuals, releasing them, and We tagged 23 yellowfin tunas with coded ultrasonic beacons during a five-month period between 11 April and 12 September 1998. They were tagged <150 m from two monitoring stations: Espiritu Santo North (ESN) and South (ESS), separated by 500 m at the Espiritu Santo Seamount (24°42'N; 110°18'W) in the south- ern Gulf of California (Fig. 1). The seamount rose to within 18 m of the surface and extended 700 m along a northwesterly-southwesterly axis. Monitoring station ESN was situated at the northwest end of the seamount ridge at a depth of 47 m; station ESS was at 37 m on the southwest end. The monitors were deployed for 30 months, during which they recorded when the tagged tuna swam within the 150-m range of reception of the monitors. Using SCUBA, we removed the monitors from the moorings at four-month intervals, downloaded the records of tuna presence near the seamounts to a laptop computer, and replaced the monitors during the same day We located a station by the rosette of buoys, which floated at a depth of <10 m and which was visible from the surface, by towing a diver at the surface near the GPS coordinates for the mooring. Manuscript approved for publication 30 January 2003 by Scientific Editor. Manuscript received 4 ApriL 2003 at NMFS Sciontifu- Publications OfTicc. Fish. Bull. 101:684-692 (2003). NOTE Klimley et al.: Occurrence of Thunnus albacares in the Gulf of California 685 115° 110° Espiritu Santo Seamount (ES) Figure 1 Bathymetric contour map of seamount Espiritu Santo (ES). The circles with cross-hatching indicate the range of the tag-detecting monitor from the sea- mount. The insert shows the geographic location of the seamount (ES) in the Gulf of Cahfomia. We determined the maximum range of signal-detection of one monitor by lowering a transmitter to 10 m under a small boat and lowering the monitor to a similar depth under a larger support vessel. We recorded the separation distance between the two boats using radar because the small boat and transmitter drifted away from the support vessel that was anchored in place at the highest point on the seamount. The VROl monitor (Vemco Ltd., Shad Bay, Nova Scotia, Canada) detected tags at a distance of 150 m in seas with waves <0.5 m high (see circles. Fig. 1). Later models (Vemco Ltd., VR02) used in the study have a pub- lished reception range of aSOO m in calm seas (see http:// www. vemco.com). The range of tag detection by the moni- tors decreases with rising sea state because of the increase in wave-generated ambient noise. The tuna were caught by rod and reel and lifted aboard 1- 30 minutes after being hooked. Smaller individuals (^15 kg) were weighed with a scale with a hook that fit into the oper- culum; intermediate sized fish (>15 and s25 kg) were weighed in the net and the net's mass subtracted from the cumulative value; and the masses of largest tuna (>25 kg) were estimated on the basis of their length by using the re- gression equation, >'=0.216ac + 2.981 given in Moore (1951). The tags were inserted into the peritoneum of the tuna while salt water was flushed over their gills by using the technique described in Klimley and Holloway (1999). The tuna were retained on board for tag implantation less than a minute. The transmitters (Vemco Ltd., V16-6L) were cylindrical and had a diameter of 16 mm, length of 106 mm, and net mass in water of 16 g. They emitted individually coded tone bursts of 70 kHz separated by 60-90 s intervals. The amplitude of the pulses was 147 dB (re: 1 ^P) at a distance of 1 m. The theoretical operating life of a transmitter was 476 days. Each tag was distinguished on the basis of a unique pulse burst by an automated tag-detecting moni- tor attached to the ESS and ESN detection stations. Water 686 Fishery Bulletin 101(3) Table 1 Length and mass of the 23 yellowfin tuna {Thunnus albacares) tagged in the present study and the date and time of tagging. "N" indicates tagging near northern monitor; "S" denotes tagging near southern monitor. An asterisk in front of a measurement indi- cates that the value is derived from the mathematical relationship between mass and length given in Moore (1951); "TL" denotes total length. Tuna no. Date Time (h) TL (cm) Mass (kg) Site (N/S) 1 11 Apr 1998 13:04 80.0 7.3 S 2 11 Apr 1998 13:21 96.0 10.8 S 3 12 Apr 1998 08:46 91.0 10.3 8 4 12 Apr 1998 08:51 106.0 13.8 S 5 12 Apr 1998 09:54 104.0 15.5 S 6 17Jun 1998 09:54 91.5 17.0 S 7 24Jun 1998 10:38 86.5 11.3 S 8 26 Aug 1998 10:05 138.0 *51.7 N 9 26 Aug 1998 10:45 58.0 4.5 N 10 26 Aug 1998 11:43 66.0 5.5 N 11 26 Aug 1998 12:16 76.0 7.0 N 12 26 Aug 1998 10:14 155.0 *73.1 N 13 28 Aug 1998 10:50 71.0 7.2 N 14 28 Aug 1998 11:25 155.0 *73.1 N 15 10 Sep 1998 17:44 149.9 *66.2 S 16 10 Sep 1998 18:32 91.5 18.50 S 17 10 Sep 1998 18:44 *75.0 8.50 s 18 10 Sep 1998 19:07 111.8 *27.6 s 19 11 Sep 1998 17:25 114.5 20.5 N 20 11 Sep 1998 17:51 71.0 7.00 N 21 11 Sep 1998 18:25 106.5 20.5 N 22 12 Sep 1998 6:41 104.5 23.0 N 23 12 Sep 1998 7:30 141.0 *55.1 N temperature was recorded every half hour at the seamount by a Stoaway Tidbit temperature logger ( Onset Computers Corp., Pocassett, MA) attached to the mooring line adjacent to the tag-detecting monitor. We calculated a daily tem- perature by averaging the half-hourly temperatures. We used log-survivorship analysis (Fager and Young, 1978) to ascertain whether the tunas returned to the moni- toring stations after favored time periods. A frequency his- togram of the time intervals between randomly occurring point events in a Poisson process is described by a negative exponential distribution. A log-survivor plot of these inter- vals generates a straight line with a slope proportional to the probability of an event occurring at a given time after the preceding event. This analysis is used to identify inter- vals between events that occur more frequently than ex- pected by chance because inflections in the resulting curve are more easily distinguished from a .straight line than the shape of the distribution on a frequency histogram with a negative exponential distribution. An inflection in the log- survivor curve indicates a change in the probability of an event occurring at a given time after the last event — in our case the time between successive arrivals of tunas within the ranges of the two monitors. Results Twenty-three yellowfin tunas were tagged from 11 April 1998 to 12 September 1998 (Table 1). Individuals were tagged during daylight hours from 6:41 to 19:07 hours. The tunas ranged in length from 71.0 to 155.0 cm TL. They ranged in mass from 7.25 to 73.1 kg. There appeared to be two discrete size classes, small individuals varying from 7.25 to 23.0 kg and large ones from 51.7 to 73.1 kg. The masses of the larger individuals were determined from their lengths by using a regression equation (Moore, 1951). The yellowfin tunas stayed at seamount Espiritu Santo over varying time periods (Fig. 2). Nine of the 23 tunas left the seamount on the same day that they were tagged (Fig. 2A). Two of the nine returned to the seamount once for a single day. one within a week of tagging and the other after two and one-half months. Six tunas stayed intermediate periods of time after tagging, ranging from two to six weeks. One of these tunas (no. 9) was eventually caught at the seamount. Another tuna (no.lO) visited for a single day after an absence of five weeks and returned again after a similar period to stay for 15 months. Four NOTE Klimley et a\. Occurrence of Thunnus albacares in the Gulf of California 687 Apr May Aug Sep Oct Nov Dec 'I ' Jan Feb Apr 28- c 2000 ^ v^--^-'^~Jt..v^;^ 24- / v'^ ; 20 J 16- ■ ' ' '1 ^-, r i V — ...,.., \ 23- 21- 19 IT- IB- IS- 11 9 7 5 3 1 — Beacon not yet recovered. T Tuna tagged with beacon. C Tuna captured with beacon. F Beacon drops to surface of seamount. Apr (itey Aug Sep Oct Jun Jul IVIonth of year Figure 2 Chronology of daily visits by 23 tagged yellowfin tuna to the seamount and temperature record over a 30- month period beginning April 1998 and ending October 2000. Each visit, indicated by a solid diamond, is based on the detection of a tag during a 24-h period by either the north (ESN) or south (ESS) monitoring stations. The lines in the graphs show that the ultrasonic tag had yet to be recovered from a yellowfin tuna. T = day of tagging, C = day of capture, and F = date of shedding of tag. individuals (nos. 5, 19, 21, and 23) stayed for longer peri- ods of time, ranging from six to 18 months. One of these tunas (no. 51 was also caught by a fisherman. It is likely that some tunas are nomadic and stay only a single day, whereas others are resident, remaining at the seamount throughout the year. 688 Fishery Bulletin 101(3) It is unlikely that the tags on the two tunas (nos. 10 and 23), which stayed at the seamount longest, were shed and lay on the bottom. The reasons supporting their being attached to living tunas are as follows. First, the two tags were not recorded with equal frequency during all times of the day as might be expected of a tag lying at one location within the range of the monitors. The tags were usually de- tected for a few hours and then absent for a similar period. This pattern of detection is consistent with the tunas mov- ing within the range of the monitor and later outside its range. Second, the two tags were jointly detected after long periods of absence or ceased being detected simultaneously after long periods of presence. This reception pattern is consistent with the two tunas moving in and out of the de- tection range of the monitors within the same school. Third, one tuna (no. 23) was detected by the monitor on the south side of the seamount, but not on the north side during one day; the same tuna was detected by the northern monitor, but not the southern monitor on the next day. This pattern of detection was consistent with the tuna swimming over the northern region of the seamount on the first day and over the southern region on the second day. The yellowfin tunas were present at the seamount at all seasons of the year Five of the tunas tagged during August and September 1998 (nos. 7, 8, 9, 16, and 17) emi- grated during early fall as the water temperature began to decrease (Fig. 2A). However, three individuals (nos. 10, 21, and 23) remained at the seamount from January 1999 to April 1999 when the temperature dropped to 18°C. Two (nos. 10 and 23) remained present when the subsurface water temperature descended to 16°C during the following winter of 2000 (Fig. 2B). The yellowfin tunas remained at the seamount at all times of the day. This is evident from a 24-h record of the arrivals of 10 tunas during a 15-d period from 16 to 30 September 1998 (Fig. 3). The tunas were present more often during daytime, from 06:00 to 18:00 hours, during the first 12 days. Notice the clustering of the different symbols in Figure 3, each indicating a specific tuna, in separate columns during the period from 06:00 to 18:00 hours. However, the amount of time spent at the sea- mount became more evenly distributed between daytime and nighttime by 28 September Note the even dispersion of the symbols over the 24-h period during the last three days of the 15-day period. There was little variation evi- dent in the frequency of arrivals at different times of the day when the arrivals were summed over the entire study (Fig. 4). The percentage of arrivals during each hour of the day (see crosshatched polygon) differed little from an even distribution of arrivals (4.2''// /h) throughout the day (see dashed circle). We determined the frequency of various lengths of stays at the north (Fig. 5A) and south sites (Fig. 5B) at the Espiritu Santo Seamount. A stay for a particular tuna was defined as the period of detections with no separation intervals greater than 15 min. Let us say that tuna 1 was detected at 08:00, 08:14, 08:28, and 09:00 hours. The dura- tion of the stay of tuna 1 would be 28 min. The second detec- tion followed the first by 14 min (<15 min), and the third followed the second also by 14 min (also < 15 min). However, the fourth detection followed the third by 32 min (>15 min) and was therefore not pooled into a single duration. This stay would then be placed in the 15:00-29:59 min time class in Figure 5. Twenty-seven percent of the detections at ESN and 33% of those at ESS were separated by greater than 15 min and were thus considered single detections and included in the 00:00-h class. Fifty-three percent of the visits to ESN and 37% of the visits to ESS were between 00:01 and 14:59 min. Twenty-one percent of the visits to ESN and 20% of the visits to ESS were between 15:00 and 59:59 min. The majority of visits were less than 1 hour in duration and only a few exceeded an hour. The intervals spent away from the seamount were simi- larly short. Sixty percent of all absences at ESN were less than 1 h (Fig. 6) as were 65% of the absences from ESS. Ninety percent of the absences from both sites were less than 5 hours. Only 0.1% of the visits exceeded 23 hours. There appeared to be no favored period of absence as indicated by the smooth slope of both curves in the log- survivor plot. Only 72 periods of absence at ESN and 114 periods at ESS exceeded a day. Of these longer periods, 42% of the absences from ESN (Fig. 7A) and 46% of the absences from ESS (Fig. 7B) were for two days. Only 7 % of the absences from ESN and 4% of the absences from ESS were between 10 and 19 days. Only 2% of periods of absence from ESN exceeding a day were greater than 100 days (Fig. 7A). Discussion We found that yellowfin tuna remained at the seamount for periods ranging from a few days to greater than a year. Fifty percent of 458 yellowfin tuna tagged with dart tags at the Cross Seamount off Hawaii were recaptured at that seamount within 15 days of tag application (Holland et al., 1999). This "half-life" of tuna residence was short, sug- gesting that the seamount served as a landmark to guide migration and not as a destination for feeding. Thirty-eight yellowfin tuna were tagged with ultrasonic beacons at two buoys off the western coast of Oahu and monitored over a 13-month period by automated "listen- ing" monitors (Klimley and Holloway, 1999). These moni- tors (VR20) possessed a more sensitive receiver than our monitors (VROl and VR02). The former had a maximum range of 1.1 km. The maximum published range of our monitors was 0.5 km. Twenty-seven of the tuna returned to the buoys a mean of 4.2 visits per tuna. The mean dura- tion of each visit was only 40.1 min and the mean period of absence was 17.2 days. Seventy-three percent of the tuna tagged on the same day returned together The tunas often arrived at the same time of the day and returned only to the buoy at which they were tagged. This allegiance of tunas to one school, their predilection for returning to the site of tagging, and the precise timing of their visits are consistent with the theory that the species has migratory pathways consisting of way-points that are visited with regularity. That the tuna spent little time at the FAD suggests that the buoys are not feeding destinations, but rather land- marks used in migration. NOTE Klimley et al.: Occurrence of Thunnus albacares in the Gulf of California 689 Tuna arrivals at Espiritu Santo North 24 00 22 00 02 00 - 02 00 -r? 00:00 T — f M' ' 1 **■ I ■ " r— r-» r**- r"'^* h oooo :£ 15-Sep-98 16-Sep-98 17-Sef)-98 18-Sep-98 19-Sep-98 20-Sep-98 21-Sep-98 22-Sep-98 23-Sep-98 24-Sep-98 V) E r (Order of ten symbols from left to right, indicate tunas witti tags 5, 8, 9, 16, 17, 19, 20,21,22 and 23) 23-Sep-98 24-Sep-98 25-Sep-98 26-Sep-98 27-Sep-98 28-Sep-98 29-Sep-98 30-Sep-98 1-Oct-98 Figure 3 Twenty-four hour chronology of visits by 10 tagged tuna to the monitoring station ESN during 15 days from 16 to 30 September 1998. A unique symbol indicates the presence of a particular individual within the range of the monitor during a 15-min position. Note the predominance of daytime visits during the first nine 24-h periods and then a progressive shift to an equal number of visits during daytime and nighttime (see 28-30 Sept. 1998). Tuna repeatedly moved in and out of the monitor range over many days or left for the duration of the study. Sixty percent of all absences at ESN and 65 % of the absences from ESS were for less than 1 hour. If these tunas were to swim at a sustained rate of 0.5 m/s (see Magnuson, 1978), they would not move more than 900 m out the reception range of the monitors (60 min x 60 s x 0.5 m/s/2).This close attachment to the seamount contrasts with the behavior of tuna at FADs offshore of Hawaii. Tunas visited the FADs there rarely and spent little time within the range of the monitor before departing for a period of several weeks (Klimley and Holloway, 1999). The present study suggests that the Espiritu Santo Seamount is a substantial feeding ground that can support a year-round resident population of yellowfin tunas. However, other tunas may stay only briefly at the seamount, using it as a landmark, before continuing on their nomadic migrations. Seamoimts have dipole magnetic fields associated with them because of the antiparallel polarity of magnetite within volcanic magma extruded during periods when the earth's polarity was reversed (Parker et al., 1987). Furthermore, maxima (ridges) and minima (valleys) in the magnetic field often lead outward from seamounts due to the extrusion of magma. Klimley (1993) proposed that hammerhead sharks use these for guidance during their nocturnal migrations into the surrounding water to forage. This physical property of the sea floor, originating fair below where the fishes swim, could provide a fixed reference (or waypoint) for yellowfin during their migrations. This species of tuna has been shown to sense distinct patterns in a magnetic field (Walker, 1984). 690 Fishery Bulletin 101(3) Espiritu Santo North (n=7,654) Espiritu Santo South 2100 (n=1 3,754) 2159 Time Percent ^^'^^ Arrivals qIOO- 0159 B 0900- 0959 Figure 4 Percentage of visits by 23 yellowfin tuna at dif- ferent times of day at monitoring stations ESN (A) and ESS (B). Circle with dashed line indi- cates an equal percentage of returns occurring every hour of the day. 60, A 50- 40- Espiritu Santo North 30 n=7,654 20- 1 _ 10- o I 1. 0) 1 1 1 1 1 1 1 1 i I 1 T — 1 1 r — r -T 1 1 Perce n B 40- Espiritu Santo South n=1 3,754 30- 20- 10- 1 ll. R^CMTrin^csiTj-io^csi-^io^CNTj-in^P '-'00000000000000000° ■^oooooooooooooooo O000O00000C3O0OOO0 oinoin^u^omomoioomoino o^co-^o^cots-o^co-^o^co-^o OOOOO^^r-fNCNCSICNCncOCOCO'^ OOOOOOOOOOOOOOOOO Period of presence (hh:mm;ss) Figure 5 Percentage of visits of increasing duration recorded for 23 yellowfin tuna at monitoring stations ESN (A) and ESS (B). A single detection of a tagged fish would be placed in the 00:00-min time category. Conclusions Twenty-three yellowfin tuna were tagged with coded ultrasonic beacons during a five-month period between 11 April and 12 September 1998. These tunas were cap- tured, tagged, and released <150 m from two monitoring stations: Espiritu Santo North (ESN) and Espiritu Santo South (ESS), which were separated by 500 m, at the Espiritu Santo Seamount in the southern Gulf of Califor- nia (24°42'N: 110°18'W). The monitors were deployed for a period of .'SO months, ranging from April 1998 to October 2000, during which they recorded tagged tunas swimming within their 150 m range of reception. The tunas ranged in length between 71.0 and 155.0 cm TL and in mass from 7.2 to 73.1 kg. The tunas stayed at the Espiritu Santo Seamount for varying time periods. Nine of the 23 tunas left the seamount on the same day that they were tagged. Two of the nine returned to the seamount twice for a single day, one within a week of tagging and another after 2.5 months. Five additional tunas stayed at the seamount for intermediate periods, ranging from two to six weeks. Four individuals stayed for longer periods of time, ranging from 6 tol8 months. Tunas were present at the seamount at all times of the day. They moved in and out of the range of the monitors, most often staying for periods <14:59 min. Fifty- three percent of the visits to ESN and 37*7? of the visits to ESS were of this duration. Smaller percentages of the visits, 2K^ and 20'/r^ , lasted 15:00 to 59:59 min, respectively. The majority of visits were <1 hour in duration and only a few exceeded an hour. The intervals spent away from the seamount were also brief Sixty percent of all absences at ESN and 65';; of the absences from ESS were <1 hour. Ninety percent of the visits to both sites were <5 hours. Only 0.17c of the visits exceeded 23 hours. Tuna individuals NOTE Kllmley et al : Occurrence of Thunnus albacares in the Gulf of California 691 100.o^ I A Iv^ -•- Espiritu Santo North (W=7841) > 100- B c ^W,^ -•- Espiritu Santo Soulti(W=13. 854) cn ra ^^•^■^ \ 0 2 4 6 8 10 12 14 16 18 20 22 24 Interval t between visits (hrs) Figure 6 Log-survivor plots of percentages of intervals between successive tuna arrivals greater than time ( over 24-h periods at monitoring stations ESN (A) and ESS (B). may use the site either as a landmark during their migra- tory transit or as a feeding destination as suggested by the short and long periods of time spent at the seamount. Acknowledgments We would like to thank those on the staffof Centro de Inves- tigaciones Biologicas del Baja Norte of La Paz, Mexico, who helped us tag yellowfin tuna at seamount Espiritu Santo. This work was funded by the Biological Oceanography Program of the National Science Foundation (grant: OCE- 9802058) and CONACYT of Mexico (grant: PN-9509-1995 and PN-1297-1998). SOt A 45^ 40- 35- Espiritu Santo North 30- (t > 1 day, N=72) 25 20- 15- 10- m 5- Mil .1 1 1 1 I i"i i™i"i"i 1 1 i"; i™i i"i o T-cvjco-^Lncoi^cpcpo ■^ oooooooocS'T o °- 50. B 45- 40^ 35- 30 Espiritu Santo South (t> 1 day, /V=114) 25- 20 15- 10- 5 IbI_bbI| _■- O^ 1 1 1 1 1 1 1 1 1 I I 1 1 1 1 i 1 : '-cvjco'^tncor^cocno OOOOOOOOOT '-fytofi/itor^ooo) '^ Interval between visits (days) Figure 7 Percentages of visits of greater than one day made by tuna to two monitoring stations with single day intervals ranging from 2-9 days and 10-day inter- vals ranging from 10-19 to 90-99 days. Literature cited Boehlert, G. W., and A. Genin. 1987. A review of the effects of seamounts on biologi- cal processes. In Seamounts, islands, and atolls (B. H. Keating, P. Fryer, R. Batiza, and G. W. Boehlert, eds.), p. 319-334. Geophys. Monogr. Sen 43. Pagan, R. M., and D. Y. Young. 1978. Temporal patterns of behavior: durations, intervals, latencies, and sequences. In Quantitative ethology (P. W. Colgan, ed), p.78-114. John Wiley & Sons, New York, NY. Fonteneau, A. 1991. Seamounts and tuna in the tropical Atlantic. Aquat. Living Resour. 4:13-25. Holland, K. N. 1996. Biological aspects of the association of tunas with FADs. SPC Fish Aggregating Device Information Bull. 2:2-7. Holland, K. N., P. Kleiber, S. M. Kajiura. 1999. Different residence times of yellowfin tuna, Thun- nus albacares, and bigeye tuna, T. obsesus, found in mixed aggregations over a seamount. Fish. Bull., 97:392-395. Klimley A. P. 1993. Highly directional swimming by scalloped hammer- head sharks, Sphyrna lewini, and subsurface irradiance, temperature, bathymetry, and geomagnetic field. Mar. Biol. 117:1-22. 1985. Schooling in the large predator, Sphyrna lewini, a species with low risk of predation: a non-egalitarian state. Ethology, 70:297-319. Klimley A. P, and S. B. Butler 1988. Immigration and emigration of a pelagic fish assem- blage to seamounts in the Gulf of California related to water mass movements using satellite imagery. Mar. Ecol. Progr. Ser 49:11-20. Klimley, A. P., and C. Holloway. 1999. Homing synchronicity and schooling fidelity by yel- lowfin tuna. Mar Biol. 133: 307-317. Magnuson, J. J. 1978. Locomotion by scombrid fishes: hydrodynamics, mor- phology, and behavior Fish Physiol. 239-313. Parker, R. L., L. Shure, and J. A. Hildebrand. 1987. The application of inverse theory to seamount magnetism. Rev. Geophys. 25:1-65. 692 Fishery Bulletin 101(3) Moore, H. L. Walker, M. M. 1951. Estimation of age and growth of yellowfin tuna (A^eo- 1984. Learned magnetic field discrimination in yellow- thunnus macropterus) in Hawaiian waters by size fre- fin tuna, Thunnus albacares. J. Comp. Physiol 155:673- quencies. Fish. Bull. 52:131-149. 679. 693 Larvae of Dactylopsaron dimorphicum (Perciformes: Percophidae) from oceanic islands in the southeast Pacific Mauricio F. Landaeta Laboratorio de Oceanografia Pesquera y Ecologia Larval Departamento de Oceanografia Universidad de Concepcion Casilla 160-C Concepcion, Chile Francisco J. Neira Faculty of Fisfienes and Marine Environment Australian Mantime College PO Box 21 Beaconsfield, Tasmania 7270, Australia Leonardo R. Castro Laboratorio de Oceanografia Pesquera y Ecologia Larval Departamento de Oceanografia, Universidad de Concepcion Casilla 160-C Concepcion, Cfiile E-mail addres (for L R Castro, contact author): lecastro@udec.cl Percophids are a family of small marine benthic fishes common over soft bot- toms from inshore to the outer slopes in tropical to teinperate regions of the Atlantic and in the Indo-West and southeast Pacific (Reader and Neira, 1998; Okiyama, 2000). Five species belonging to four genera have been recorded around the Salas y Gomez Ridge in the southeast Pacific, all of which are endemic to the area except for Chrionema chryseres, a species which also occurs off the Hawaiian Islands and Japan (Parin, 1985, 1990; Parin et al., 1997). Of these five species, larval stages have been described only for Osopsaron karlik and Chrionema pallidum (Belyanina 1989, 1990). Dactylopsaron dimorphicum (Parin and Belyanina, 1990) is a dwarf per- cophid (29 mm maximum body length) previously recorded only at the Cupole (26°S; 86°W) and Baral (25°S; 96°W) seamounts located to the west of the Salas y Gomez Ridge and at the junction of this and the Nazca Ridge, respectively, at depths of 240-345 m (Parin, 1990; Parin et al., 1997 ). Aduhs of this monotypic genus differ from other percophids in that the first dorsal fin is positioned at the back of the head and is in line with the mid-operculum, 8-10 digitiform processes are present on the posterior upper opercular mar- gin, and expanded lobes are present at the distal end of the medial branchio- stegal rays (Parin, 1990). This species is sexually dimorphic, males have a thicker and much longer first dorsal-fin spine than females (Parin, 1990). There is no information on their reproductive biology and eggs are unknown (Watson etal., 1984). We describe the postflexion larvae of D. dimorphicum using material col- lected around Salas and Gomez and Easter Islands in the southeast Pacific. We also provide information on the spa- tial distribution of this species around both islands, and on how to distinguish the larvae from those of teleosts with similar larvae in the area. This note constitutes the first record of D. dimor- phicum off Easter Island, as well as the first record of the larvae in nearshore waters of both Pacific islands. Methods Field work Larvae were obtained during an ocean- ographic expedition (CIMAR-5) to Easter Island (27°10'S; 109°20'W) and Salas y Gomez Island (26°30'S; 105°20'W), approximately 3750 km west of Chile, in November 1999. Sam- ples were collected onboard the Chil- ean navy research vessel AGOR Vidal Gormaz by using a bongo sampler equipped with two conical nets (0.6-m diameter mouth openings, 3 m long, 350-nm mesh size). The mouth of each net was fitted with an OSK flowmeter to estimate volume of water filtered. Tows were carried out for 10 min obliquely to the surface from either the maximum permissible depth in shallow (<200 m) stations or from 400 m in deeper sta- tions. Samples around Easter Island were obtained at 10 stations located approximately one nautical mile (nmi) from the coast both during day and night, and along four transects (NW-SE and NE-SW) each containing four stations located at 3, 7, 12, and 20 nmi offshore (Fig. 1). Samples around Salas y Gomez Island were obtained along four transects (N-S and E-W), each containing four stations at 1, 3, 6, and 10 nmi offshore (Fig. 2). Addi- tionally eight deep stations (>1500 m) were also sampled between Easter and Salas y Gomez islands. All samples were fixed in 5% formalin and later pre- served in 70% ethanol. Water volume sampled per tow ranged between 112.6 and 517.7 m-'. Larval abundances were standardized to 1000 m-^ and mapped by using SURFER® (Golden Software, Golden, CO). Statistical analyses were performed using STATISTICA (Star- Soft, Inc., Tulsa, OK). Larval identification and processing Postflexion larvae were identified as those of Dactylopsaron dimorphicum by a combination of dorsal and anal-fin meristics (D. IV [III-V] + 22 [20-22] Manuscript approved for publication 15 January 2003 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:693-697 (2003). 694 Fishery Bulletin 101(3) and A. 24 [23-25]; Table 1), and by the presence of the unique digitiform opercular processes (Parin, 1990; Oki- yama, 2000). Identification was verified by using fin meris- tics from cleared and stained specimens (Potthoff, 1984). 26.70 26 90 27.10 5 zr.x> 27 50 Dactylopsaron dimorphicum A total of 55 postflexion larvae of D. dimorphicum (8.2-15.3 mm standard length) were examined to describe morphometries, meristics, and pigmentation. Three larvae (9.1, 13.1 and 13.4 mm SL) were cleared and stained fol- lowing the method of Potthoff ( 1984). Terminology and morphometric measurements follow Neira et al. (1998). Measurements were made to the near- est 0.01 mm by using a dissecting microscope fitted with an eyepiece micrometer Body length (BL, Neira et al., 1998) in postflexion larvae cor- responds to standard length (SL), i.e. tip of snout to posterior margin of hypurals. Measurements of body depth (BD), head length (HL), and preanal length (PAL) were converted to a percentage (%) of SL (Table 2). Eye diameter (ED) and snout length (SnL) were converted to a percentage (%} of HL. Pigment described refers solely to melanin. Illus- trations were made with the aid of a camera lucida. Larvae per 1000 m o 0 1 -5 5-10 10-50 50-100 >100 109 80 109 60 109 40 Longitude (W) 109 20 109 00 Figure 1 Spatial distribution of postflexion larvae of Dactylopsaron dimorphi- cum (numbers/1000 m^) around Easter Island in November 1999. ID ■a 3 itl Dactylopsaron dimorphicum 26 35 rtJ o o 26.45 Salasy Gbmez Island 0 o 0 • 0 • Laivae per 1000 m^ 26 55- o 0 • 1-5 o # 5-10 % > 10 26.66 106.56 105.45 105.35 Longitude (W) 105.25 Figure 2 Spatial distribution of postflexion larvae of Dactylopsaron dimorphi- cum (numbers/1000 m^) around Salas y G6mez Island in November 1999. Results Description of larvae Postflexion larvae of Dactylopsaron dimorphicum are elongate (BD 13.1-18.3%; Table 2), and have a moderate to large head (HL 28.6-36.7%) and an elongate snout (Fig. 3). Eyes are round and pigmented by 8.2 mm SL. The mouth is large, pro- trusible, and has a long ascending premaxillary process giving a characteristic "duckbill" appear- ance. Small villiform teeth are present along the premaxilla and dentary. There are no head spines. The digitiform processes on the upper opercular margin are present in larvae >13.2 mm SL; the lower, rearward-directed process that reaches the end of the pectoral-fin base in adults was still forming in the largest larva examined (15.3 mm SL). The branchiostegal membranes are free from the ithsmus. The short first dorsal fin is located at the nape and lies in line with the mid-operculum; pterygiophores of the five first-dorsal fin spines in two of the cleared and stained larvae ( 13.1 and 13.4 mm SL) were located between the neural spines of the second and third trunk vertebrae. The 9. 1-mm- SL cleared and stained larva possessed only 15 of the 17-20 pectoral-fin rays, and first dorsal-fin spines were developing. The elongate pelvic fins are thoracic, i.e. inserted in front of the pectoral- fin bases. Lateral line scales begin to form at >13 mm SL. Larvae are unpigmented, although a few had a small melanophore at the base of the 17 or 18* dorsal-fin ray There are 31-35 myomeres. The number of vertebrae in the cleared and stained larvae is 34-35 (11-12 -h 22-24). Larval distribution Postflexion D. dimorphicum larvae were collected within 6 nmi off both Easter and Salas and Gomez NOTE Landaeta et a\ Larvae of Dactylopsaron dimorphicum in the southeast Pacific 695 Figure 3 Postfiexion larvae of Dactylopsaron dimorphicum collected around Easter Island in November 1999. (A) 8.5 mm SL; (B) 14.8 mm SL. Illustrated by F. J. Neira. Table 1 Meristic counts of percophid species recorded in submarine ridges of Salas y Gomez Island in the southeastern Pacific (from Parin 11985, 19901 and Okiyama [2000]). Dorsal Anal Pectoral Pelvic Caudal (branched) Vertebrae Chrionema chryseres Chrionema pallidum Dactylopsaron dimorphicum Enigmapercis acutirostris Osopsaron karlik VI +16 26 23 1,5 15(11) 27-28 VI +14-15 18 20-22 1,5 15(11) 27-28 IIII-V] + 22 [20-22] 24 (23-251 18 [17-20] 1,5 14(8) 34-35 11+21 25 21 1,5 15 (8-9) — V- VI +19-20 22-23 19-20 1,5 14 32 This study. Table 2 Standard length range (mm), and mean values (±1 SD) of selected body proportions (given as a percentage of body length) of postfiexion larvae o( Dactylopsaron dimorphi- cum from Easter and Salas y Gomez islands in the south- eastern Pacific Ocean («=55). Standard length (mm) 8.2-15.3 Head length (%SL) 28.6-36.7 (32.3 ±3.3) Eye diameter (%HL) 18.2-24.7 (22.0 ±3.1) Snout length (%HL) 25.6-36.2 (31.0 ±5.51 Body depth (7fSL) 13.1-18.3 (15.3 ±2.2) Preanal length C/fSL) 45.4-56.9 (49.5 ±4.0) Islands (Figs 1 and 2). Around Easter Island, larvae were caught only in nearshore stations ( <2 nmi ) over the narrow shelf and were more abundant along the southern edge. 8 9 10 11 12 13 14 Body length (mm) Figure 4 Combined body length (SL, mm) frequency distribution of postfiexion larvae of Dactylopsaron dimorphicum around Easter Island and Salas y Gomez Island in November 1999. 696 Fishery Bulletin 101(3) The highest larval concentrations (>100 larvae/1000 m'') occurred at the southeastern tip of the island and averaged 27 ±46 larvae/1000 m'' (Fig. 1). No significant differences were found between day and night larval concentrations (Kruskal-Wallis test=0.047; P>0.05). Around Salas and Gomez Island, larvae were caught only at two stations 6 nmi west and south of the island, and in mean concentra- tions <10 larvae lOOO/m-^ (Fig. 2). No larvae were caught in any of the eight stations sampled between the two islands. Body lengths of larvae caught in both islands ranged from 8 to 16 mm SL, and over 30% of the larvae were around 12 mm SL (Fig. 4). Discussion Postflexion larvae of D. dimorphicum are likely to be confused with those of four other co-occurring percophid species (see Table 1), and those of the creediid Crystallo- dytes pauciradiatus that occur in the same region (Castro and Landaeta, 2002) and have similar bodies with little or no pigment. In the case of the percophids, the digitiform opercular processes exclusive to D. dimorphicum , together with dorsal and anal-fin meristics, should be sufficient to distinguish between postflexion larvae of all species. Larval C. pauciradiatus can be identified by using myo- mere counts (56-58 vs. 31-35 in D. dimorphicum ) and their small, early forming posterior preopercular spines (Reader et al., 2000). Our collection of D. dimorphicum larvae at Easter Island, some 453 km to the southwest of Salas y Gomez Island where it was first described (Parin, 1990), constitutes the first record for Easter Island, thereby extending the known range of this species over the South Pacific plate. Despite numerous past fish surveys around Easter Island (i.e. Ran- dall and Cea-Egaiia, 1984; Mujica, 1993), adults of this dwarf percophid had not been reported there, a fact that could be attributed to factors such as collection methods, depth of surveys, and the very small size of these larvae. However, the presence and abundance of larval D. dimorphicum reported in this study, and the fact that they were among the five most abundant larval taxa caught around Easter Island (Castro and Landaeta, 2002), implies the existence of a well-estab- lished breeding population. Biogeographically, this finding also suggests that larval drift could play an important role in the expansion of this and other fish species that have pelagic larvae in this region of the southeast Pacific. In this context, it is perhaps relevant that expansions offish ranges are not uncommon in this region, even though both Easter and Sa- las y Gomez Islands lie in different biogeographic provinces (Parin et al., 1997). A good example is the pentacerotid Pen- taceros decacanthus, which was regarded as endemic of the Nazca and Salas y Gomez Ridges until it was recorded in Easter Island (Parin and Kotlyar, 1988). Acknowledgments We would like to thank Paula Rosenborg and all the crew from the AGOR Vi.dal Gormaz for their help with sampling. We also thank Muneo Okiyama for his comments on iden- tification of percophid larvae. This research was funded by the Comite Oceanografico Nacional (CONA), Chile, and forms part of a study on distribution patterns and larval accumulation around oceanic islands headed by Leonardo Castro. Literature cited Belyanina, T. P. 1989. Ichthyoplankton in the regions of the Nazca and Salas y Gomez submarine ridges. J. Ichthyol. 29(5):84— 90. 1990. Larvae and fingerlings of little-known benthic and benthopelagic fishes from the Nazca and Salas y Gomez ridges. J. Ichthyol. 30(6):1-11. Castro. L. R., and M. F. Landaeta. 2002. Patrones de distribucion y acumulacibn larval en tomo a islas oceanicas: Isla de Pascua y Salas y Gomez. Cienc. Tecnol. Mar. CONA 25(1):131-145. Mujica, A. 1993. Zooplancton de las aguas circundantes a la Isla de Pascua (27°08'S-109°26W). Cienc. Tecnol, Mar. CONA 16:55-61. Neira, F. J., A. G. Miskiewicz, and T. Tmski. 1998. Larvae of temperate Australian fishes. Laboratory guide for larval fish identification, 474 p. Univ. Western Australia Press, Nedlands, Australia. Okiyama, M. 2000. Percophidae (sandfishes, duckbills). //; The larvae of Indo-Pacific coastal fishes: an identification guide to marine fish larvae (J. M. Leis and B. M. Carson-Ewart, eds.), p. 554-560. Brill, Leiden, The Netheriands. Parin, N. V. 1985. A new hemerocoetine fish, Osopsaron karlik ( Percophi- dae, Trachinoidei) from the Nazca submarine ridge. Jpn. J. Ichthyol. 3 1(4 1:358-361. 1990. Percophid fishes (Percophidae) from the Salasy Gomez ridge (Southeast Pacific). J. Ichthyol. 30(l):68-79. Parin, N. V., and A. N. Kotlyar 1988. A new boarfish, Pentaceros quinquespinis (Pentacero- tidae), from the Southeast Pacific. Vopr Ikhtyol. 28(3): 355-360. Parin, N. V., A. N. Mironov, and K. N. Nesis. 1997. Biology of the Nazca and Sala y Gomez submarine ridges, an outpost of the Indo-West Pacific fauna in the Eastern Pacific Ocean: composition and distribution of the fauna, its communities and history. Adv. Mar Biol. 32: 147-242. PotthofT, T 1984. Clearing and staining techniques. In Ontogeny and systematics of fishes (H. G. Moser, W. J. Richards. D. M. Cohen, M. P. Fahay A. W. Kendall, and S. L. Richardson, eds.). p. 35-37. Am. Soc. Ichthyol. Herpetol. Special Pub- lication 1. Randall, J. E., and A. Cea-Egana. 1984. Native names of Easter Island fi.shes, with comments on the origin of the Rapanui people. Occas. Pap. Bernice P Bi.shop Mus. 25(12):1-16. Reader, S. E., and F J. Neira. 1998. Percophidae: sandfishes, duckbills. In Larvae of temperate Australian fishes. Laboratory guide for larval fish identification IF J. Neira, A. G. Miskiewicz, and T. Tmski. eds.). p. 358-361. Univ. Western Australia Press; Nedlands, Australia. NOTE Landaeta et a\ Larvae of Dactylopsaron dimorphicum in the southeast Pacific 697 Reader, S. E., J. M. Leis, and D. S. Rennis. Watson, W., A. C. Matarese, and E. G. Stevens. 2000. Creediidae (tommyfishes). In The larvae of Indo- 1984. Trachinoidea: development and relationships. In Pacific coastal fishes. An identification guide to marine fish Ontogeny and systematics of fishes (H. G. Moser, W. J. Rich- larvae (J. M. Leis and M. Carson-Ewart leds.). p. 575-578. ards, D. M. Cohen, M. P. Fahay, A. W. Kendall Jr, and S. L. Brill, Leiden, The Netherlands. Richardson, eds.), p. 554-561. Am. Soc. Ichthyol. Herpetol. Special Publication 1. 698 Assessment of sampling methods to estimate horseshoe crab (JLimu/us polyphemus L.) egg density in Delaware Bay Penelope S. Pooler David R. Smith us. Geological Survey Leetown Science Center 11700 Leetown Road Kearneysville, West Virginia 25430 E-mail address (for D. R Smith, contact auttior): david_r_smitti@usgs gov Robert E. Loveland Department of Ecology and Evolution Cook College Rutgers University New Brunswick, New Jersey 08901 Understanding the reliability of egg density estimates at multiple scales will help develop effective monitoring programs. We addressed all three questions with respect to eggs found in both shallow (0-5 cm) and deep (0-20 cm) sediments. Horseshoe crabs are gener- ally thought to lay most of their eggs at a depth of 15-20 cm (Brockmann, 1990; Botton et al., 1994). Through processes of bioturbation and wave-generated sediment activation, horseshoe crab eggs are brought onto the beach and made available to foraging shorebirds (Botton et al., 1994; Kraeuter and Feg- ley, 1994; Jackson et al., 2002). Materials and methods Mark L. Botton Fordtiam University 113 West 60"' Street New York, New York 10023 Stewart F. Michels Delaware Division of Fisfi and Wildlife RO. Box 330 Little Creek, Delaware 19961 Each spring horseshoe crabs {Limtilus polyphemus L.) emerge from Delaware Bay to spawn and deposit their eggs on the foreshore of sandy beaches (Shuster and Botton, 1985; Smith et al., 2002a). From mid-May to early June, migra- tory shorebirds stopover in Delaware Bay and forage heavily on horseshoe crab eggs that have been transported up onto the beach (Botton et al., 1994; Burger et al., 1997; Tsipoura and Burger, 1999). Thus, estimating the quantity of horseshoe crab eggs in Delaware Bay beaches can be useful for monitoring spawning activity and assessing the amount of forage avail- able to migratory shorebirds. We evaluated procedures to estimate horseshoe crab egg density by asking three questions that address sampling at a different spatial scale. 1) How many samples of sediment are needed for precise estimation of egg density within a segment of beach? 2) Does egg density within a segment of beach ad- equately represent egg density across a larger stretch of beach? 3) How many beach segments should be sampled to monitor bay-wide egg density? We chose these three questions because the objective of egg studies might focus on any of these scales. We ask the first question to determine the sampling effort necessary to detect changes in egg density over time within a specific beach segment. The second question allows us to examine the reliability of using egg density in a beach segment to infer egg density over a larger stretch of beach. The third question deals with the level of precision in estimates of bay-wide egg density and how many beaches must be sampled to detect bay-wide declines in density over time. During May and June 1999, we col- lected sediment on 16 beaches in Delaware Bay (Fig. 1), eight along the eastern shore (New Jersey) and eight along the western shore (Delaware), to estimate egg density. Methods used to collect sediment and extract horseshoe crab eggs are summarized in the pres- ent study, but are presented in detail in Smith et al. (2002b). Beach sediment was collected in cores (5 cm diameter) within a 3-m wide strip along a 100-m segment of beach. Each 3-m wide strip was centered on the mid-beach eleva- tion where a majority of horseshoe crab nests occur (Botton et al., 1988). The mid-beach elevation is halfway between the spring high water level and the beach break at the low tide ter- race. Within each egg-sampling strip, 40 locations were selected randomly for sediment collection. At each location, a pair of core samples was taken: one to a depth of 5 cm and the other to a depth of 20 cm. We sampled eggs on 25-26 May and 14-15 June 1999, which fol- lowed the heaviest spawning activity in Delaware Bay that year (Smith et al., 2002a). We mixed the entire core contents thoroughly and then removed 80-mL aliquots. We ran the aliquots Manuscript approved for publication 12 February 2003 by Scientific Editor. Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:698-703 (2003). NOTE Pooler et al.: Assessment of sampling methods to estimate egg density of Limulus polyphemus 699 Woodland\,^ "^Sea Breeze NEW JERSEY \ Raybins ,Fortescue ~\ Kitts Hummockr Kimbles /Reeds North Bowers* South Cape . Shore Lab / /Highs \< Big Stone\^ North ( ) Slaughter^ DELAWARE powieK Prime Hool^^ Cape May / ^ BroadkilK^ «i = 10 0 10 20 Kitometers ~1 s Figure 1 Delaware Bay beaches ( • ) where horseshoe crab eggs were sampled in May £ind June 1999. through a 1-mm sieve to separate eggs and larvae from ambient sediments and then counted eggs (dead or live) and larvae in each aliquot. Depth of aerobic sand varied; thus we measured core volume prior to extrapolating egg counts to totals per core and then estimated the total den- sity of eggs and larvae. The larvae comprised a small frac- tion of total eggs and larvae, and for the purposes of this paper we evaluated the sampling of eggs only. Question 1 : How many sediment cores should be sampled per beach segment? We addressed this question in two steps. First, we deter- mined the precision of egg-density estimates as a function of egg density and sample size. Second, we translated the precision of the estimates into statistical power to detect change in egg density over time. For simplicity, variance of the egg-density estimate was calculated from a random sample from an infinite population. Coefficient of variation (CV) was calculated as CV = Vvar(.v)/H/y, where var(jy) = variance of eggs among cores; and y = egg density. We modeled the relationship between egg density and variance among cores (i.e. var[y]=/'[ y] to predict coefficient of variation (CV) for different sample sizes and across the observed range of egg densities (i.e. CV = ■]f\y\l n jy). Using predicted CVs, we estimated the probability of de- tecting a change in egg density over time. The probability of detecting decline (i.e. statistical power) was calculated by using a one-tailed ^test with a type-I error rate of 0.2 and a constant rate of annual change for CVs = {0.1, 0.2, 0.3, 0.4) with the software program TRENDS (Gterrodette, 1993). Question 2: Is egg density within a beach segment representative of egg densities along a larger stretch of beach? Smith et al. (2002b) modeled the relationship between counts of spawning females and egg densities within beach segments. Spawning females are counted annually as part of a bay-wide survey of spawning activity (Smith et al., 2002a), and in 1999, egg sampling was conducted on some of the same beaches as the spawning survey (Smith et al., 2002b). For eggs that were sampled in May 1999 on six New Jersey beaches, the relationship was fairly strong, linear, and pre- dictive (/•2=0.62; Smith et al., 2002b). Although we sampled for eggs on only one 100-m segment of beach, we used the above relationship to predict egg densities for all 100-m segments along the beach where spawning females were counted. We limited the predictions to the six New Jersey beaches where we felt the relationship between spawning females and egg densities was sufficiently strong (Smith et al., 2002b). We compared egg density in the observed 100-m segment to the distribution of densities predicted in all 100-m segments on the beach. If the observed density was within the interquartile range of the distribution of predicted densities, we concluded that the 100-m segment was representative of the larger stretch of beach. Question 3: How many beaches should be sampled? Using the observed variation in egg density among the 16 beaches sampled in 1999, we predicted the CV for bay-wide egg density estimates as a function of the number of beaches sampled and under a stratified sampling design where the two strata were New Jersey smd Delaware. We could not evaluate CV across a range of bay-wide densities because the 1999 results provided only one datum point, and we expected variation among beaches to be a function of egg 700 Fishery Bulletin 101(3) Table 1 Mean eggs per core and standard errors (SE) for horseshoe crabs [Limul IS polyphemus) at 16 beaches sampled in Delaware Bay that were sampled in May and June 1999. Cores were 5 cm in diameter. At 40 random locations on each beach, a pair of sediment cores were sampled: one core at 5 cm depth (shallow sediment) and another at 20 cm depth (deep sediment). No. of eggs per core on 25-26 May 1999 No. of eggs per core on 14-15 June 1999 Shallow Deep Shallow Deep State Beach sediment SE sediment SE sediment SE sediment SE Delaware Broadkill 0.0 0.00 1.5 1.47 1.2 0.36 101.7 60.42 Prime Hook 0.2 0.08 81.9 76.17 7.4 2.06 223.9 112.14 Fowler 0.1 0.05 1.8 0.65 2.7 1.22 211.3 116.56 Slaughter 11.7 2.69 814.8 186.04 41.7 5.35 664.5 97.81 Big Stone 0.1 0.05 11.3 5.09 0.7 0.53 24.2 14.27 North Bowers 23.0 6.49 950.3 234.18 105.1 23.54 400.4 70.81 Kitts Hummock 26.4 8.23 325.1 78.63 15.2 5.49 124.8 43.94 Woodland 0.5 0.17 0.1 0.06 7.0 3.68 60.9 29.75 New Jersey North Cape May 0.3 0.25 0.0 0.00 0.5 0.33 0.7 0.37 South Cape Shore Lab. 25.5 0.86 1085.4 140.29 4.5 0.81 1399.2 144.03 Highs 2.1 0.71 1128.6 96.99 4.4 0.94 1456.8 173.80 Kimbles 9.7 4.80 1561.3 286.32 1.7 0.55 1008.0 105.63 Reeds 2.0 0.52 540.4 79.90 18.2 2.52 468.0 62.67 Raybins 3.5 1.91 65.8 43.88 0.1 0.06 6.7 4.57 Fortescue 2.0 0.43 645.9 108.71 20.6 3.85 465.7 193.64 Sea Breeze 27.5 7.95 347.3 94.70 0.2 0.09 3.1 2.01 density. However, we examined the probability of detecting a percentage change in bay-wide egg density over time as a function of the number of beaches sampled by using the 1999 bay-wide egg density as the initial value in the time series. Results When the objective is to monitor egg density within a seg- ment of beach, a sample size of 40 sediment cores is suf- ficient for detecting substantial changes in egg density in the top 20 cm of sediment, but >40 cores would be needed to monitor egg density in the top 5 cm of sediment. Distribu- tions of egg densities were skewed right with median densi- ties of 3 and 275 eggs per core for shallow and deep cores, respectively (Table 1 ). A sample size of 40 cores resulted in a CV of 0.26 for a median density of eggs 0-20 cm deep (Fig. 2B). In contrast, about 100 cores would need to be sampled to bring the CV down to 0.3 when .sampling shallow sedi- ment and when egg density was at the median (Fig. 2A). A CV of 0.3 corresponds to a 75% chance of detecting a 509f decline in egg density over 5 years (Fig. 3A) and an 80% chance of detecting a 40% decline over 10 years (Fig. SB). A sample size of 60 shallow cores would result in CV of 0.4 for median egg density (Fig. 2A), which would be sufficient for monitoring over 10 years, but not over 5 years. A CV of 0.4 would lead to a better than 85% chance of detecting a 50% decline in density over 10 years (Fig. 3B). Precision and power would improve when sampling higher densities of eggs (Fig. 2). At most beaches, observed egg densities within a 100-m segment of beach were not representative of egg densities throughout a larger beach. On only two of the six New Jer- sey beaches examined (South Cape Shore Lab and Reeds) did the observed egg density fall within the interquartile range of beach-wide densities (Fig. 4). On three beaches the observed egg density was greater than all predicted densities, and on one beach observed egg density was less than all predicted densities. With egg density at the 1999 level and sampling at 16 beaches (i.e. eight beaches per state) distributed throughout the bay, the CV for densities of eggs in 0-20 cm of sediment was 0.26 in May and 0.29 in June (Fig. 5). For densities of shallow eggs, the CV was 0.33 for egg densities in May and 0.43 in June. Variability in egg densities among beaches was greater for sampling in June 1999 than in May 1999. Discussion Eggs in shallow sediment (0-5 cm) consistently yielded lower densities and higher variability than eggs in deep sediment (0-20 cm). A sample size of 40 sediment cores was sufficient for estimating and monitoring density of eggs 0-20 cm deep within a 100-m beach segment. However, a larger sample size (260 sediment cores) would be needed for estimating and monitoring density of eggs 0-5 cm deep within a segment of beach. Because egg density in a 100-m segment of beach is not necessarily representative of the larger surrounding beach. NOTE Pooler et al : Assessment of sampling methods to estimate egg density of Limutus polyphemus 701 S 0.6 a 0.4 0) 0.2 0.0 500 1000 1500 Density of eggs per core Figure 2 The relationship between density and coefficient of variation (CV) for (A) shallow sediment ( 0-5 cm ) and ( B ) deep sediment ( 0-20 cm ). Curves in each figure depend on sample size: circle is n = 20, triangle is a; = 40, square is n = 60, diamond is n = 80, and x is n = 100. Vertical lines represent median egg densities that we observed in 1999. 09 = 0,6 0.4 0 3 0 4 0 5 Rate of decline over five years 0.3 04 0.5 Rate of decline over ten years Figure 3 Probability of detecting a decline (i.e. statistical power) for various magnitudes of decline and for annual surveys over five (A) and 10 IB) years. Statistical power was calculated for a one-tailed f-test with a type-I error rate of 0.2, and a constant annual rate of change. it is important to realize that if sampling is restricted to a short segment of beach, then the scope of inference is strictly limited to that segment. If a rehable estimate of egg density along a beach is required, then it will be nec- essary to take samples along the entire beach. Because of the logistics of sampling sediment it would be difficult to sample throughout a long stretch of beach in one stage of sampling. However, a two-stage sampling design could be considered in which beach segments are selected at the 1*' stage and sediment cores within segments are selected at the 2"'' stage. Consistent with our findings on sampling within a beach, bay-wide egg density can be more precisely estimated for eggs 0-20 cm deep than for eggs 0-5 cm deep. A stratified 702 Fishery Bulletin 101(3) 0 15000 30000 45000 60000 0 15000 30000 45000 60000 Predicted density of egg cores Figure 4 Density curves of predicted egg densities on 100-ni beach segments at six New Jersey beaches. An arrow represents the egg density in the observed 100-m segment. These density curves were generated by dividing the area sur- veyed for spawning females into 100-m segments and using the observed relationship between egg densities and spawning females to predict egg density for each seg- ment. The beaches shown are (A) Fortescue. (B) Highs, (C) Kimbles, (D) North Cape May, (E) Reeds, and (F) South Cape Shore Laboraton,'. random sample of eight beach segments per state would result in CV sO.3 for estimates of egg densities 0-20 cm deep. If this level of effort were maintained, it would be sufficient to detect biologically significant declines in egg density over a 5- or 10-year period. However, greater effort would be required to monitor change in egg densities 0-5 cm deep. According to results from the May samples, to estimate egg densities in shallow sediment with CV sO.3, a stratified random sample of 10 segments per state would be required. Sampling eggs is a costly process; therefore sampling efficiency and reducing sample size are important consider- ations. Although sediment can be collected quickly, the pro- cess of extracting and enumerating eggs from the sediment can be time consuming. Quantifying the eggs in surface sediments to assess shorebiixi forage biomass is likely to be the main objective of many egg sampling studies because horseshoe crab spawning activity can be assessed by other methods, such as through counts of spawning horseshoe crabs (Smith et al., 2002a). However, a primary finding in the present study is that estimating eggs in 0-5 cm 5 7 10 12 15 17 20 22 Number of beaches sampled per state Figure 5 Predicted coefficient of variation (CV) shown for the pos- sible range of number of beaches sampled per state. This figure is based on the observed levels of bay-wide density during the two sampling periods in 1999. Curves are based on egg densities found at different depths and time periods: triangle is shallow sediment in June, circle is shallow sedi- ment in May, diamond is deep sediment in June, and square is deep sediment in May. Shallow sediment is 0 to 5 cm deep, and deep sediment is 0 to 20 cm deep. of sediment will be more costly than estimating eggs in 0-20 cm of sediment. In the future, alternatives in survey design, such as stratification of the beach foreshore, should be considered to reduce the amount of sediment that needs to be collected for precise estimates of horseshoe crab egg density. Acknowledgments This work was funded through the USGS/State Partner- ship Project (no. 99HQAG0050). Additional funding was provided through the New Jersey Endangered & Nongame Species Program. Literature cited Botton, M. L., R. E. Loveland, and T. R. Jacobsen. 1988. Beach erosion and geochemical factors: influence on spawning success of horseshoe crabs {Limulus polyphemus) in Delaware Bay Mar Biol. 99:325-332. 1994. Site selection by migratory shorebirds in Delaware Bay, and its relationship to beach characteristics and abundance of horseshoe crab (Limnlus polyphcmus) eggs. Auk 111:605-616. Brockmann, H. J. 1990. Mating behavior of horseshoe crabs, Limulus polyphemus. Behaviour 114:206-220. Burger, J.. L. Niles, and K. E. Clark. 1997. Importance of beach, mudflats and marsh habitats to migrant shorebirds on Delaware Bay. Biol. Conserv. 79: 283-292. NOTE Pooler et a\: Assessment of sampling methods to estimate egg density of Limulus polyphemus 703 Gerrodette, T. 1993. TRENDS: software for a power analysis of linear regression. Wildl. Soc. Bull. 21:515-516. Jackson, N. L., K. F. Nordstrom, and D. R. Smith. 2002. Geomorphic-biotic interactions on beach foreshores in estuaries. J. Coast. Res. 36:414-^24. Kraeuter. J. N., and S. R. Fegley. 1994. Vertical disturbance of sediment by horseshoe crabs (Limulus polyphemus) during their spawning season. Es- tuaries 17:288-294. Shuster, C. N., Jr., and M. L. Botton. 1985. A contribution to the population biology of horseshoe crabs, Limulus polyphemus (L.), in Delaware Bay. Estu- aries 4:363-372. Smith, D. R., R S. Pooler, B. L. Swan, S. F. Michels, W. R. Hall, R J. Himchak, and M. J. Millard. 2002a. Spatial and temporal distribution of horseshoe crab {Limulus polyphemus) spawning in Delaware Bay: implica- tions for monitoring. Estuaries 25:115-125. Smith, D. R., P. S. Pooler, R. E. Loveland, M. L. Botton, S. F. Michels, R. G. Weber, and D. B. Carter. 2002b. Horseshoe crab (Limulus polyphemus) reproductive activity on Delaware Bay beaches: interactions with beach characteristics. J. Coast. Res. 18:730-740. Tsipoura, N., and J. Burger. 1999. Shorebird diet during spring migration stopover on Delaware Bay. Condor 101:635-€44. 704 Larval abundance, distribution, and spawning habits of spotted seatrout iCynoscion nebulosus) in Florida Bay, Everglades National Park, Florida Allyn B. Powell Center for Coastal Fisheries and Habitat Research National Ocean Service National Oceanic and Atmospheric Administration 101 Pivers Island Road Beaufort, North Carolina 28516 E-mail address; allyn. powelliaJnoaa.gov The spotted seatrout (Cynoscion nebu- losus ) is one of the most sought after recreational fish in Florida Bay. and it spends its entire life history within the bay (Rutherford et al. ,1989b). The biol- ogy of adult spotted seatrout in Florida Bay is well known (Rutherford et al., 1982, 1989b) as is the distribution and abundance of juveniles within the bay. The habitats and diets of juveniles are well documented (Hettler, 1989; Ches- ter and Thayer, 1990; Thayer et al., 1999; Florida Department of Environ- mental ProtectionM. Nevertheless, the spatial and temporal spawning habits of spotted seatrout and the distribu- tion of larvae have only been partially described (Powell et al., 1989: Ruther- ford et al., 1989a i. An excellent description of the eco- logical history of Florida Bay is given by Fourqurean and Robblee (1999). Briefly, Florida Bay is subtropical and is generally oligotrophic. The bay is a network of shallow basins, mud banks, and mangrove islands (keys). Tides are influenced by the Gulf of Mexico and Atlantic Ocean, but mud banks, which are connected to basins by channels, restrict circulation in the bay and at- tenuate tidal energy very quickly. As a result there is essentially no lunar tide over most of the central and northeast- ern portion of the bay. This impediment to circulation could have a negative effect on the recruit- ment of early-stage planktonic larvae into these portions of the bay. Within the next few decades, plans to restore the Everglades include increasing freshwa- ter flows to Florida Bay. Prerestoration information on larval distribution and spawning patterns of spotted seatrout is a high priority because increased freshwater flows can have potential positive and negative impacts. At low salinities, the planktonic eggs of spot- ted seatrout could sink to the bottom and would not be viable (Holt and Holt, 2002; Alshuth and Gilmore^). On the other hand, increased freshwater flows can alleviate hypersaline conditions that could result in an expansion of the distribution of the early life stages of spotted seatrout (Thayer et al., 1999; Florida Department of Environmental Protection'). The objective of the pres- ent study is to document the distribu- tion and abundance of spotted seatrout larvae to determine their early life his- tory habitats and spawning habits in Florida Bay. Methods and materials To describe the distribution and abundance of spotted seatrout larvae in Florida Bay, I devised a series of ichthyoplankton surveys between 1994 and 1999. The initial survey was conducted during nine nonconsecutive months between September 1994 and August \i)9r^. A total of 14 fixed sta- tions wore selected in basins of Florida Bay (Fig. 1). In accordance with rec- ommendations by the South Florida Ecosystem Restoration Prediction and Modeling (SFERPMi, Program Man- agement Committee (PMC), Florida Bay was divided into six zones for ease of reporting results (Table 1, Fig. 1). These zones are based on the benthic moUuscan and benthic plant commu- nities (Fourqurean and Robblee, 1999). Paired bongo nets, 60 cm wide, were fitted with 0.333-mm mesh and fished from the port side of a 5.4-m boat. Nets were towed during daylight, approxi- mately 1 m below the surface for 5 minutes and volume estimates were obtained from flowmeter readings. In 1996, sampling was conducted monthly from April through Septem- ber at stations where recently hatched spotted seatrout occurred during 1994-95 (stations 5, 6, 9-13). In 1996, 1 used a paired 60-cm bow-mounted push nets with 0.333-mm mesh similar to that described by Hettler and Chester ( 1990). Nets were fished approximately 1 m below the surface for 3 minutes. The volume of water sampled with the push net was slightly greater than that sampled with the bongo nets (60 m'^ vs. 50 m-'). To test the efficiency of the two gears, both were fished simultaneously at 23 stations during 1996. A Kruskal- Wallis nonparametric test was used to evaluate differences (Sokal and Rohlf 1981 ). No significant differences in den- sities offish larvae were found between gear types (P=0.50). During and after September 1997 sampling for spotted seatrout was limited to four stations in four zones (Table 1; stations 6. 15, 16, 17) where paired bow-mounted push nets were employed. Sampling occurred during July and September 1997; March, May, June, July, and September 1998; and May, July, and November 1999. ' Florida Department of Environmental Protection. 1995. Fisheries-indepen- dent monitoring program, annual report. Florida Department of Environmental Protection, Florida Marine Research In- stitute, 100 8'h Avenue SE. St. Petersburg, FL 33701. ^ Alshuth, S., and R. G. Gilmore Jr 1994. Salinity and temperature tolerance limits for lai'vai spotted seatrout. Cynoscion neb- (//<).s».s- C. (Pisces: Sciaenidac). Int. Coun. Explor Sea, Coun. Meet. Pap., ICES-CM- 1994/L: 17. 19 p. Manuscript accepted for publication 19 February 2003 by Scientific Editor Manuscript received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:704-711 (2003). NOTE Powell: Larval abundance, distnbution, and spawning habits of Cynosaon nebulosus 705 — 25°20' — 25°10' — 25°00' 24°50' 8roo' 80°45' 80°30' Figure 1 Location of stations in Florida Bay sampled in 1994-99. See Table 1 for station latitudes and longitudes. Table 1 Florida Bay sampling stations including zone locations as defined by the South Florida Ecosystem Restoration Prediction and Modeling Program, Program Management Committee. Stations 1-14 were sampled in 1994-95; stations 5 ,6, 9-13 in 1996; and stations 6, 15-17 in 1997-99. Station Latitude Longitude (degrees and minutes) (degrees and minutes) Florida Bay zones 24 59.42 80 34.06 Atlantic transition 25 04.42 80 31.24 eastern 25 10.54 80 29.12 eastern 25 009.24 80 37.12 eastern 25 08.30 80 43.19 central 25 04.57 80 46.32 central 25 03.54 80 40.12 central 24 52.46 80 47.31 Atlantic transition 24 55.60 80 55.40 Gulf transition 24 58.48 80 59.48 Gulf transition 25 06.49 8105.16 Gulf transition 25 07.22 80 55.62 Gulf transition 24 59.98 80 55.46 western 24 59.06 80 46.54 central 25 10.80 80 37.80 northern 25 06.00 80 52.50 western 25 07.67 80 57.32 Gulf transition Location 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Cowpens Cut Butternut Key Duck Key between Eagle Key and Madeira Point Big Key Whipray Basin between Calussa and Russel Keys between Old Dan and Peterson Key banks Sprigger Bank between Oxfoot and Sprigger Banks Cape Sable Dave Foy Bank between Blue and Ninemile Banks between Rabbit and Gopher Keys Little Madeira Bay entrance Palm Key Basin Bradley Key 706 Fishery Bulletin 101(3) At stations where replicate tows were taken, densities were averaged. Ichthyoplankton samples were preserved in 95% ethanol. Temperature was measured at all stations with a hand-held thermometer and salinity was measured with a refractometer. Size at age of larvae was estimated from the equation L,, standard length = -1.31 + 1.2162 (L,, age in daysl (Powell et al.^). Because of the high coefficient of variation associ- ated with ichthyoplankton samples (Cyr et al., 1992), my sampling design was probably inadequate for multiway statistical comparisons. Therefore, I used nonparametric Kruskal-Wallis tests with o = 0.10 (Sokal and Rohlf, 1981) and relied on patterns and trends to infer differences in densities of spotted seatrout between stations and time periods. In the period 1994-95 we tested densities among nine months and 14 stations to determine trends in spatial and temporal spawning habits. We also tested differences between the period 1994-95 and the period 1996 to de- termine interannual spatial and temporal differences or similarities. Only those months (April through August) that were sampled during the two periods were included. During the period from 1997 through 1999 it was only appropriate to determine spatial differences because we sampled irregularly during this period. A general description of the diverse habitats in relation to stations in the present study is described by Holmquist et al. ( 1989, decapod and stomatopod communities ); Thayer and Chester (1989, fish distribution, seagrass distribution and abundance, sediment depth, and organic content); Zie- man et al. ( 1989, macrophyte distribution); and Fourqurean and Robblee (1999, general description of the Florida Bay ecosystem). Results In 1994-95, salinities were lowest and most variable at sta- tions 1-5 and 7 in the eastern part of Florida Bay (Figs. 1 and 2 ). Hyperhaline conditions were never observed during this period. Salinities in 1996, which were recorded monthly from April through September at stations where recently hatched spotted seatrout were collected in 1994-95, were generally euhaline and, as in 1994-95, hyperhaline con- ditions were never observed. From July 1997 through November 1999 at four trout monitoring stations, mean salinities were similar at stations 6, 16, and 17 but were most variable at station 6. Station 15 had the lowest mean salinity and the greatest variation. At this station salini- ties ranged from 10.0 (March 1998) to 33.0 psu (May 1999). Highest salinities for all four stations were observed in May 1999. Hyperhaline conditions were observed only at station 6 in June 1998 and May 1999 and at station 16 in May 1999 (Fig. 2). In general, spawning in Florida Bay occurred between March and October and peaked in June, August, and Sep- 3 Powell, A. B., R. Cheshire, E. H. Laban, J. Colvocnresses, P. O'Donnell.and M. Davidian. In review. Growth, mortality and hatchdate distributions of larval and juvenile spotted seatrout, Cynoscion nehulosus, in Florida Bay, 26 p. 15 10 40 -1 35 S 30 Q. $• 25 c « on - 15 10 ~i — I — I — I — I — I — I — I — I — I — I — I — r 1 2 3 4 5 6 7 8 9 10 11 12 13 14 B i ■ i i « i ~i I I I r I I 5 6 9 10 11 12 13 40 35 30 - 25 - 20 15 10 \ I 1 1 1 1 6 15 16 17 Station Figure 2 Mean salinities SD for (A) 14 stations sampled from September 1994 through August 1995, (B) 7 stations sampled monthly from April through September 1996, and (C) 4 stations sampled from July 1997 through November 1999. For months that samples were taken, refer to Tables 2 and 3. tember ( Table 2 ). Densities of spotted seatrout were signifi- cantly different among months in 1994-95 (P<0.01) and 1996 (P=0.01). Spotted seatrout larvae were absent during December, February, and April in 1994—96 (Table 2), and in November 1997-99 (Table 3). Most spawning, based on larval collections, occurred between 26° and 34°C (Fig. 3). The coldest temperature at which larval spotted seatrout were collected was 20°C in March 1998 at station 6. Spotted seatrout larvae were collected mainly at salinities between 25 and 40 psu (Fig. 3), although larvae were collected in salinities as low as 12 psu at station 15. NOTE Powell: Larval abundance, distribution, and spawning habits of Cynoscion nebulosus 707 Spatially, there were significant differences in densities of spotted seatrout among 14 stations from 1994 through 1995 (P=0.01), and densities were highly variable (Table 2). In addition, a considerable number of zero catches oc- curred at stations where spotted seatrout were collected at least one time. This high variability indicated that the sampling design was inadequate to properly evaluate the spatial and temporal abundance of spotted seatrout larvae. However, some patterns could be discerned. Generally, lar- val spotted seatrout were absent or rarely collected in the eastern (stations 2, 3, and 4), Atlantic transition (stations 1 and 14), northern transition (station 15) zones and in a portion of the central zone (station 15) (Tables 2 and 3). They were consistently collected at station 6 in the central zone (Tables 2 and 3). There were no significant differences (P=0.14) in densi- ties of spotted seatrout among stations in 1996; stations where trout occurred on more than one occasion in 1994-95 (Table 2). As in 1994-95, high mean densities of spotted seatrout occurred at station 6 (Table 2). Significant differences in spotted seatrout densities at certain stations were observed between the periods 1994—95 and 1996. Differences were observed at stations 9 (P=0.02), station 10 (P=0. 02), and station 13 (P=0.02). At these three stations in 1996 spotted seatrout were collected only in one month (August). They were never collected at station 5 in 1996 (Table 2). Size of larvae was indicative of spawning locations. Recently hatched spotted seatrout larvae ( 1.0-1.9 mm no- tochord length; s5 d old) were collected mainly in central (stations 5 and 6), Gulf transition (stations 9, 10, 11, 12, 17) and western (stations 13 and 16) zones. Recently hatched larvae were rare at station 4 (eastern zone), station 14 (central zone) and station 15 (northern transition zone). They were absent at station 8 (Atlantic transition zone) (Figs. 4 and 5). Discussion Evidence from previous studies (Powell et al., 1989; Ruth- erford et al., 1989a) and the present study establishes the spatial and temporal spawning habits of spotted seatrout in Florida Bay. Length-frequency distributions of spotted seatrout larvae collected in 1984-85 (Powell et al., 1989) and data from the present study indicate that spotted seat- rout spawn mainly in the Gulf transition, central, and west- em zones of Florida Bay and that there is limited spawning in the northern transition and eastern zones (Figs. 4 and 5). Spawning also occurs in the far northeastern portion of Florida Bay in Little Blackwater and Blackwater Sounds (Rutherford et al., 1989a). However, there is no evidence for spawning in the Atlantic transition zone. The distribution of planktonic larvae is not necessarily a good indicator of postsettlement habitat requirements because abiotic fac- tors related to transport could influence their distribu- tion. However, in Florida Bay larvae are not distributed homogeneously throughout the bay, and mudbanks impede circulation (Fourqurean and Robblee, 1999). The adults are generally nonmigratory and inhabit shallow seagrass-rich environments (Chester and Thayer, 1990). Hence, the dis- cu u Q ■ S o -rr CO CO ,-H rA CD a ►J S M " 00 to •H Oi lO o cq ^~* M o t^ M o o O CD o CO 00 X, -6 M Z Z in Z a, OJ < a q -a 6 Tf c cs +1 a m 02 C CO 00 _Q CO C to 11 d +1 cfl C/2 o CO o o o o o CO CO CD S Z Z Z o a a. ■ -, CO T3 2; -^ bo o M M CO CO m CO CO CO CO c^ t g 3 < •z Z z Z Z (M 2 6 CM . 0) CO V, O 1-t i> o in °° -t ^ ^^ o o Tf o r-i CO o ,_; cc o ^ m aJ 3 ,-* r-» ,-» ro -a 1-5 i-4 3 a^ •S| in CD Oi "* 2 c 05 1— 1 c CO lO o iri CO CO CTJ o in o ■H o S^ 3 ,— 1 00 eg ■-5 Cfi CO ^5 q ,-H +1 Oi ,-H 00 o 00 ca cd >> a o o o o rH CX) § o o Tf o e4 o O o O o C3 g " Z o " -* CM ^ci OJ CO ~£ -■" I— 1 +1 ^ S Tf 135 o t~: o CO 00 "^ o. o CO Tf o d ,-i CO o CD Tf ,-4 S-2 2 a ?— 1 ?— 1 M Ol ,-H a |i o '-5 C CO rt OJ ^ -S -S ■^ in CO 00 Ol o O) CO ■* § Q CO M ,-» 708 Fishery Bulletin 101(3) Total fish larvae Spotted seatrout 10000 10000 1000 100 10 f 1 '•••*•• 1000 ••• • : ' '•••.' <•: . 10 • • •• 1 o 18 20 22 24 26 28 30 32 34 36 18 20 22 24 26 28 30 32 34 36 ^ Temperature ("C) E ID C J 10000 10000 S 1000 100 10 1 • :»«. ' •* 1000 • •••• ?ri^ 1 • • : .V' •••1 fcV 1 ) 10 15 20 25 30 35 40 45 10 15 20 25 30 35 40 45 Salinity (psu) Figure 3 Density of total fish larvae and spotted seatrout versus temperature and salinity. "Total fish larvae" indicates the complete range of temperature and salinity where all collections were made and is provided for visual comparison with spotted seatrout collections. Table 3 Densities (numbers/100 m-') of spotted seatrout collected at monitoring stations with a mesh. bow-mounted push net with 0.333-mm .Station 1997 1998 1999 Jul Sep Mar May Jun Jul Sep May Jul Nov Mean ±SD 6 15 16 17 Mean ±SD 0 6.2 3.2 20.1 7.1 1.1 6.6 1.0 4.2 ±3.3 7.1 ±9.0 1.2 40.2 0 0 1,0 1.2 30.8 0 8.2+15.0 10.3 ±10.4 57.0 0 75.6 0 33.1 ±39.0 16.0 93.1 0 0 0 5.2 0 0 4.0 ±8.0 24.6 ±45.7 56.8 0 0 0 14.2 ±28.4 1.1 0 0 0 0 0 6.9 0 2.0 ±3.3 0 30.2 ±33.2 2.6+6.7 9.1 ±23.5 5.0 ±10.1 tribution of larvae presented in the present study is most likely a good indication of adult spawning areas. As indicated by the larval collections in this study, spotted seatrout have a protracted spawning period from March through October, which is similar to that observed in Tampa Bay, Florida ( McMichael and Peters, 1989). To the contrary, Stewart (1961) reported that spotted seatrout in Florida Bay spawn throughout the year, and Rutherford et al. (1989a) indicated that some spawning occurred as early as February and continued into December. Powell et al.,'* studying hatchdate distributions of juveniles, reported peak spawning in early May, late June, and late August. Seagrass meadows appear to be critical habitats for ju- venile spotted seatrout (Chester and Thayer, 1990;Tolanet al., 1997; Rooker et al.. 1998; Thayer et al., 19991. Rooker et al. (1998) reported that juvenile spotted seatrout in a Texas estuary prefer H. wrightii over T. testudinum. In another Texas estuary, Tolan et al. (1997) reported that NOTE Powell: Laa/al abundance, distribution, and spawning habits of Cynoscion nebulosus 709 juvenile spotted seatrout prefer H. wrightii over Syringo- dium fiUforme. In Florida Bay juveniles are collected at highest densities in western Florida Bay basins near the Gulf of Mexico in habitats with deeper and more organic sediments and with greater density and biomass of S. fili- forme. In areas where spotted seatrout juveniles are rare or absent, which generally reflects the distribution of their larvae, organic matter and sediment depth are minimal, water depth is generally deeper, and seagrass standing crop, short shoot densities, and diversity are lower (Thayer and Chester, 1989; Chester and Thayer 1990), Spotted seatrout larvae were collected consistently at relatively high densities in Whipray Basin (station 6; central zone) and length-frequency distributions indicate spawning most probably occurs in this area. The major- ity of larvae collected in this area were 2.0-2.4 mm SL (5 to 6 d old), and it is possible that larvae could have been transported into this area from western Florida Bay. Nev- ertheless, Whipray Basin is a nursery area for juvenile spotted seatrout (Florida Department of Environmental Protection^. However, Whipray Basin has a relatively sparse standing crop of the seagrass Thalassia testudinum (12 g/m'-) compared to Palm Key (station 16; western zone) which has a higher standing crop of T. testudinum (28 g/m-) and Halodule wrightii (14 g/m^), and has been demonstrated to be an important nursery area for spotted seatrout juveniles (Florida Department of Environmental Protection'). Spotted seatrout eggs have been collected in other waters ranging from 15 to 50 psu (Holt and Holt, 2002). Presumably, larval spotted seatrout eggs sink to the bot- tom at salinities <15 psu and are not viable (Alshuth and Gilmore^). Therefore, it is surprising that recently hatched (<5-d old) larval spotted seatrout were collected, although infrequently, at Little Madeira Bay (station 15; northern transition zone) and only at 12 psu. Whether these low sa- linities, which occurred in July and September 1997, were a result of drastic changes in salinities caused by weather events that occurred after hatching is unknown. Still, it is highly unlikely that a significant number of viable eggs can be produced at those low salinities. The qualitative description of spotted seatrout spawn- ing habits provides necessary baseline data in relation to restoration activities, specifically freshwater inflow. Restoration activities could have both a negative and posi- tive effect because salinity can have significant effects on spotted seatrout reproduction and early life history stage processes ( Holt and Holt, 2002 ). For example, low salinities (as discussed above) could be detrimental to egg viability; whereas, alleviating hypersaline areas could expand the spawning area, particularly in the central zone where hy- persalinity conditions are persistent (Orlando et al., 1997; Thayer et al., 1999). At high and low salinities, growth and development rates of larval spotted seatrout have been reported to be reduced because these processes are con- strained by undeveloped osmoregulatory functions (Holt and Banks, 1989). On the other hand, there is evidence that spotted seatrout populations have adapted to reproduce in extreme-salinity environments where spawning salinities influence egg buoyancy and the salinity tolerance of early Sta 4 ^^ "1 — r- * Oi <• C7» * O) O — — c^ f>4 r^ t^ ^ I I I I I I ^ o tn o ■• o> ♦ a> o — — c«j ri -n ^ V I ) I I I I M O lA O )0 o «o — •-^ c« * rj m »o « I I I I I I ^ O U^ O U^ o in — — IM (N lO •^ Sla 1 1 10 ■ 0 - -n^ * Ot •♦ o> * Ol o — ^ rx M K> to * t I I I I I /< O T ai -^ cjj at 1— T— CNCNCM T— 1— CNJtNtN oinoin'^ oinoin'^ T-^c>jcs( T-^CNrsi Size class (mm) Figure 5 Length-frequency distributions of spotted seatrout from four stations where a bow- mounted push net was used to sample fish from July 1997 through November 1999. See Table 1 for station locations. larval stages. This adaptation would allow spotted seatrout to spawn over a wide range of salinities. Future monitoring of spotted seatrout larval abundances to evaluate restoration activities would probably require numerous samples per station because of the high degree of variability as shown by the present study. Therefore, it would seem prudent to continue monitoring spatial spawn- ing habits from larval collections, but to develop a juvenile abundance index to monitor the success of restoration in the Everglades, as well. Acknowledgments Sincere appreciation is extended to Al Crosby who was field party chief and processed numerous samples. I am also indebted to other Beaufort Laboratory staff members, notably Robin Cheshire, Peter Crumley, Mike Greene, Donald Hoss, Michael Johnson, and Michael Lacroix for their able assistance in the field; Robin Cheshire, Curtis Lewis, and Harvey Walsh for graphics; James Waters for computer programming; and Dean Ahrenholz, Patti Mar- raro Joe Smith and an anonymous reviewer for valuable comments in their review of the manuscript. The staff of the Polish Sorting and Identification Center processed many of the ichthyoplankton samples. This study was sup- ported through joint funding from the National Oceanic and Atmospheric Administration Coastal Ocean Program and National Marine Fisheries Service base funds to the Beaufort Laboratory. Literature cited Chester, A. J., and G. W. Thayer. 1990. Distribution of spotted seatrout {Cynoscion nebu- losus) and gray snapper {Lutjanus griseus) juveniles in seagrass habitats of western Florida Bay. Bull. Mar. Sci. 46:345-357. Cyr, H., J. A. Downing, S. Lalonde, S. B. Baines, and M. L. Price. 1992. Sampling lai-\'al fish: choice of sample number and size. Trans. Am. Fish. Soc. 121:356-368. Fourqurean, J. W., and M. B. Robblee. 1999. Florida Bay: a history of recent ecological changes. Estuaries 22:345-357. Hettler, W. F., Jr. 1989. Food habits of juveniles of spotted seatrout and gray snapper in western Florida Bay. Bull. Mar. Sci. 44: 152-165. NOTE Powell: Lan/al abundance, distribution, and spawning habits of Cynoscion nebulosus 711 Hettler, W. R, and A. J. Chester. 1990. Temporal distribution of ichthyoplankton near Beau- fort Inlet, North Carolina. Mar. Eeol. Prog. Sen 68:157- 168. Holmquist, J. G., G. V. N. Powell, and S. M. Sogard. 1989. Decapod and stomatopod communities of seagrass- covered mud banks in Florida Bay: inter-and intra- bank heterogeneity with special reference to isolated subenvironments. Bull. Mar. Sci. 44:251-262. Holt, G. J., and M. A. Banks. 1989. Salinity tolerance and development of osmoregulation in larval sciaenids. In The early life history offish (J. H. S. Blaxter, J. C. Gamble, and H. von Westernhagen, eds.); the third ICES symposium, Bergen, Norway 35 October 1988, p. 4-89. Rapp. P-V. Reun. 191. Holt, G. J., and S. A. Holt. 2002. Effects of variable salinity on reproduction and early life stages of spotted seatrout. It} Biology of the spotted seatrout (S. Bortone, ed.), p. 135-145. CRC Press. Wash- ington, DC. McMiehael. R. H., Jr., and K. M. Peters. 1989, Early life history of spotted seatrout, Cynoscion nebu- losus (Pisces: Sciaenidae), in Tampa Bay, Florida. Estuar- ies 12:98-110. Orlando, S. P, Jr,, M. B. Robblee, and C, J. Klein. 1997. Salinity characteristics of Florida Bay: a review of the archived data set (1955-95), 89 p. Office of Ocean Resources Conservation and Assessments, National Oceanic and Atmospheric Administration, Silver Spring, Maryland, Powell, A. B., D. E. Hoss, W. F. Hettler, D. S. Peters, and S. Wagner. 1989. Abundance and distribution of ichthyoplankton in Florida Bay and adjacent waters. Bull. Mar. Sci. 44:35- 48. Rooker, J. R., S. A. Holt, M. A. Soto, and G. J. Holt. 1998. Postsettlement patterns of habitat use by sciaenid fishes in subtropical seagrass meadows. Estuaries 21: 318-327. Rutherford, E. S., E. S. Thue, and D. G. Buker. 1982. Population characteristics, food habits and spawning activity of spotted seatrout, Cynoscion nebulosus, in Ever- glades National Park. Rep. T-668, 48 p. South Florida Research Center, U. S. National Park Service, Homestead, Florida. Rutherford, E. S., T. W Schmidt, and J. T. Tilmant. 1989a. Early life history of spotted seatrout (Cynoscion nebulosus) and gray snapper iLutJanus griseus) in Florida Bay, Everglades National Park, Florida. Bull. Mar. Sci. 44:49-64. Rutherford, E. S., J. T. Tilmant, E. B. Thue, and T. W. Schmidt. 1989b. Fishery harvest and population dynamics of spotted seatrout, Cynoscion nebulosus, in Florida Bay and adjacent waters. Bull. Mar. Sci. 44:108-125. Sokal, R. R., and F J. Rohlf 1981. Biometry, 2nd ed., 859 p. W. H. Freeman and Co., San Francisco, CA. Stewart, K. W. 1961. Contributions to the biology of the spotted seatrout (Cynoscion nebulosus) in the Everglades National Park, Flor- ida. M.S. thesis, 103 p. Univ. Miami, Coral Gables, FL. Tolan, J. M., S. A. Holt, and C. P. Onuf. 1997. Distributoin and community structure of ichthyo- plankton in Laguna Madre seagrass meadows: potential im- pact of seagrass species change. Estuaries 20:450^64. Thayer, G. W., and A. J. Chester. 1989. Distribution and abundance of fishes among basin and channel habitats in Florida Bay. Bull. Mar. Sci. 44: 200-219. Thayer, G. W., A. B. Powell, and D. E. Hoss. 1999. Composition of larval, juvenile and small adult fishes relative to changes in environmental conditions in Florida Bay Estuaries 22:518-533. Zieman, J. C, J. W Fourqurean, and R. L. Iverson. 1989. Distribution, abundance and productivity of sea- grass and macroalgae in Florida Bay. Bull. Mar. Sci. 44: 292-311. 712 Effect of analytical conditions in wavelength dispersive electron microprobe analysis on the measurement of strontium-to-calcium (Sr/Ca) ratios in otoliths of anadromous salmonids Christian E. Zimmerman Oregon State University Department of Fisheries and Wildlife Corvallis, Oregon 97331 Present address: US. Geological Survey Alaska Science Center 101 1 E Tudor Road Anchorage, Alaska 99503 E-mail address: czimmermanfSusgs gov Roger L. Nielsen College of Oceanic and Atmospheric Sciences 104 Ocean Administration Oregon State University Corvallis, Oregon 97331 The use of strontium-to-calcium (Sr/Ca) ratios in otoliths is becoming a standard method to describe life history type and the chronology of migrations between freshwater and seawater habitats in teleosts (e.g. Kalish, 1990; Radtke et al., 1990; Secor, 1992; Rieman et al., 1994; Radtke, 1995; Limburg, 1995; Tzeng et al. 1997; Volk et al., 2000; Zimmerman, 2000; Zimmerman and Reeves, 2000, 2002). This method provides critical information concerning the relation- ship and ecology of species exhibiting phenotypic variation in migratory behavior (Kalish, 1990; Secor, 1999). Methods and procedures, however, vary among laboratories because a standard method or protocol for measurement of Sr in otoliths does not exist. In this note, we examine the variations in analytical conditions in an effort to increase precision of Sr/Ca measure- ments. From these findings we argue that precision can be maximized with higher beam current (although there is specimen damage) than previously recommended by Gunn et al. ( 1992). Wavelength dispersive electron mi- croprobe analysis (WD-EM) has been used by most researchers, although other methods such as ()rot()n-induced x-ray emission (PlXExBabaluk et al.. 1997; Markowitz et al, 2000 ) have been used. WD-EM remains a common and relatively inexpensive method. The conceptual approach among research- ers using WD-EM is similar but the methodological approach or analytical (operating) conditions vary. In a com- parison of laboratories using common otoliths, Campana et al. (1997) found among-laboratory variation in mean Sr concentrations that could not be described by otolith variability. Al- though the laboratories were internally consistent in applying their methods, comparisons between laboratories dif- fered. Campana et al. suggested that the sensitivity of WD-EM to operating conditions might have led to this varia- tion between laboratories. Development of analytical techniques for measuring Sr/Ca ratios has been re- viewed to validate techniques in specific studies (Kalish, 1990; Secor, 1992; Toole and Nielsen, 1992; Limburg, 1995). Gunn et al. (1992) analyzed effects of counting times, beam current, accel- erating voltage, and beam diameter on measures of Sr and other elements and they warned that beam powers required for WD-EM were sufficient to cause specimen damage including pitting and chemical change. As a re- sult, Gunn et al. (1992) recommended limiting beam power densities to < 3pW/pm''. This recommendation has been followed in most studies using Sr/ Ca ratios to reconstruct the chronology of migrations between the freshwater and marine environments (Table 1). Toole and Nielsen ( 1992), however, con- cluded that Sr/Ca precision could be in- creased, with no loss of accuracy, by using analytic conditions that lead to a beam power density of just over 15pW/pm- (5-pm beam diameter; accelerating voltage=15 loA; beam current=25 kV). The inherent beam damage was not critical because of the similar behavior of Sr and Ca during progressive beam damage. In published studies using WD-EM to measure Sr/Ca ratios in otoliths, the operating conditions, including beam power densities, have varied greatly (Table 1). Establishing a microprobe protocol for measurement of Sr/Ca ratios in otoliths involves a balancing act of counting times, beam current, and beam diameter. The selection of optimum conditions is constrained by financial resources, allocation of time for use of instruments, and the required resolution of Sr/Ca ratios for any specific application. Each re- searcher must weigh the benefits and costs to best answer the question at hand. Generally, these parameters are manipulated to optimize precision and accuracy of analyses in relation to vari- ability within the otolith and implica- tions of the results. For Sr/Ca ratios to remain an ac- cepted and accurate means of describ- ing migration histories and other life history events, continued analytic and technical refinement and validation are required. We examined the effects of crystal choice, beam diameter, beam current, and beam power densities on Sr/Ca measurements (expressed as atomic ratios) in salmonid otoliths: 1) we measured Sr using both the TAP and PET crystals in regions with high Sr/Ca 00.003) and low Sr/Ca XO.OOl) Manuscript approved for publication 12 February 2003 by Scientific Editor Manu.script received 4 April 2003 at NMFS Scientific Publications Office. Fish Bull. 101:712-718 (2003). NOTE Zimmerman and Nielsen: Measurement of strontium-to-calcium ratios in otoliths of anadromous salmonids 713 Table 1 Analytic conditions reported by researchers using wavelength-dispersive (WD) electron spectroscopy to measure Sr/Ca ratios in otoliths. Beam power density was calculated for this study and minimum limit of detection is either directly reported from the work cited or from personal communications. Beam power Minimum limit Beam Accelerating Beam current density of detection Source diameter Ip m) voltage (kVl (nA) (pW/pm2) (Sr ppm) Brown and Severin (1999) 6 15 20 10.61 — Campana et al. (1997) WD-1 10 25 5 1.59 175 Campana et al. (1997) WD-2 9 15 4 4.25 480 Kafemannetal. (2000) 5x8 15 10 3.75 490 Kahsh(1990) 10x10 15 10 1.5 — Kawakami et al(1998) 1 15 50 954 — Limburg(1995) 20 20 25 1.59 290 Radtke(1995) 5 15 10 7,63 — Rieman et al. (1994) 5 15 50 38.14 — Secor(1992) 5x5 25 20 20.00 580 Thresher etal. (1994) 14 15 25 2.44 311 Toole etal. (1993) 5 15 20 15.27 — Volk etal. (2000) 10 15 15 2.86 237 Zimmerman and Reeves (2000) 7 15 50 19.50 43 levels; 2) we then compared the results of repeated Sr/Ca measurements collected at the same spots using various beam diameters, while holding accelerating voltage and beam current constant to determine the effect of beam damage on Sr/Ca measurements; and 3) we compared the results of repeated Sr/Ca measurements collected at the same spots using various beam currents, while holding ac- celerating voltage and beam diameter constant. We argue that increased precision of Sr measurements afforded by higher beam current (and hence, higher beam power densi- ties) is preferable for studies where only measurements of Sr/Ca ratios are required. Materials and methods Otolith preparation Sagittal otoliths from an adult sockeye salmon (Oncorhyn- chus nerka) collected in the Deschutes River, Oregon, and a juvenile chinook salmon (O. tshawytscha) collected in the Umatilla River, Oregon, were used to represent high (>0.003) Sr/Ca and low (<0.001) Sr/Ca ratios, respectively (Zimmerman, unpubl. data). High Sr/Ca ratios character- ized the saltwater growth region in the sockeye salmon otolith and low Sr/Ca ratios characterized the freshwater growth region of the chinook salmon otolith. Each oto- hth was mounted sulcus side down with thermo-setting plastic resin on a microscope cover slip attached at one end with super-glue to a standard microscope slide. The otolith was then ground with 1200-grit sandpaper in the sagittal plane to the level of the nucleus. The mounting medium was heated and the otolith turned sulcus side-up. The otolith was then ground with 1200-grit and 2000-grit sandpaper in the sagittal plane to the level of the primor- dia and polished with 0.05-pm alumina paste. The cover slip was then cut with a scribe and mounted with other prepared otoliths (those used in other studies) on a petro- graphic slide for microprobe analysis. The slide contain- ing several otoliths was rinsed with deionized water, air dried, and carbon coated (400 A). Elemental analysis was conducted with a Cameca SX-50 wavelength dispersive microprobe. Strontiantite (SrC03, USNM R10065) and calcite (CaCOs, USNM 136321) were used as standards for Sr and Ca, respectively. Standards were calibrated with a 30-pm-diameter beam and 10-s counts resulting in minimal effects of beam damage. Effect of spectrometer (crystal) choice To evaluate differences in diffracting crystals, we con- ducted a series of tests where Sr was measured by using both the PET and TAP crystals. A 15 kV, 50 nA beam was used for these comparisons. With a 7-pm-diameter beam, Sr was measured by using the TAP crystal (Sr La) and Ca was measured by using the PET crystal (Ca Ka). Two transects of 10 points each were sampled so that the points on adjacent transects covered the same temporal location on the otolith. Sr and Ca were analyzed simultaneously; counting times for the Sr and Ca peaks were 40 s, and back- ground counts were 40 s. A second set of transects covering the same temporal locations in the otolith was sampled, but Sr was measured with the PET crystal (Sr La). Because our microprobe has only one PET crystal, simultaneous measurement of elements was not possible. Transects were conducted on both high and low Sr/Ca regions. Sr/Ca ratios 714 Fishery Bulletin 101 (3) were calculated from normalized mole fractions of Sr and Ca. Limit of detection (3a: Potts, 1987) was calculated for all points in both high and low Sr/Ca regions. Effect of beam diameter and beam current We conducted five repeat measurements at each of five points within the high and low Sr/Ca regions using a 1-. 7-, 15-, 20-. and 25-pm-diameter beam at 15 kV and 50 nA. This resulted in beam power densi- ties of 961, 19.5, 4.2, 2.39, and 1.52pW/pm^ respec- tively. We conducted five repeated measurements at each of four locations within the high and low Sr/Ca regions using beam currents of 5, 10, 20, and 30 nA with a 10-m-diameter beam and accelerating voltage of 15 kV, with resulting beam power densities of 1.0, 1.9, 3.8, and 5.7pW/pm-, respectively. The coefficient of variation (CV) of Sr/Ca ratios was calculated as the ^^Sr/Ca ^ Mean Sr/Ca"' for each beam power density. If beam damage affects precision and accuracy of Sr/Ca ratios, subsequent measurements at the same spot should be increasingly divergent from the first and such divergence should be evident in high coeffi- cients of variation. The limit of detection was used as a measure of precision. Limit of detection (3a; Potts, 1987 ) for Sr was calculated for the first measurement taken at each beam power density in each region. Results Effect of spectrometer (crystal) choice Spectrometer (crystal) choice for the measurement of Sr had an apparent systematic effect on Sr/Ca ratios at high Sr/Ca levels but no effect at low Sr/Ca levels (Fig. lA). The mean Sr/Ca level in the high Sr/Ca region was significantly lower (-15%) when Sr was measured on the PET crystal «=7.189; P<0.001; df=38). Crystal choice had an effect on measurement of both Ca and Sr in the high Sr/Ca region. Ca did not differ significantly between the high and low Sr/ Ca regions (P>0.05) but was approximately 2% lower when Sr was measured with the PET crystal (Fig. IB). This dif- ference was attributable to beam damage, which occurred as Sr was measured. The mean Ca was 197.400 ppm when Sr was measured on the PET crystal and 202,100 ppm when Sr was measured on the TAP crystal. In the high Sr/Ca region, mean Sr was significantly lower when Sr was measured on the PET crystal (^=11. 58; P<0.001;df=38)(Fig. lBi.Themean(±SD)was7270±4ppm when Sr was measured with the PET crystal and 8870 ±3 ppm when Sr was measured with the TAP crystal. This difference is also reflected in the higher minimum limit of determination for Sr for PET in both the high Sr/Ca (695 ppm) and low Sr/Ca regions (126 ppm). Using the TAP crystal to measure Sr, we found that the minimum limit of determination for Sr was 103 ppm and 65 ppm in the high Sr/Ca and low Sr/Ca regions, respectively. To achieve simi- lar counting statistics for Sr with the PET crystal, count 0 005 - t-^-^-^^-^i^^^ 0.0O4 ■ r T V O 0 002 ■ ^"'^^^^^ 0 001 ■ 0 000 ■ 22 ■ 20 - 18 - ^^-^^-^^^^-^!^ ^=0=^=^ |- 0 20 ■ 0.15 B ^^=i^fe^#-4 t=:n9r==m — E o 1 " CO i; - O 10 - es=crptTS Sr atomic % o O 8 ■ 6 - ^—^ f=.=^=A^-A.-^ ^ vv ^-W\ - 0 00 )23456?89 10 Transect pent Figure 1 Transects of (A) Sr/Ca atomic ratio and (B) Ca (circles) and Sr (triangles) as atomic percentage when Sr is measured on the PET crystal (solid symbols) and on the TAP crystal (open symbols). Error bars represent 95% confidence intervals. times would need to be increased to 200 seconds on both the peak and background. Effect of beam diameter and beam current In repeated measurements with different beam diameters (same beam current) at the same locations, Sr/Ca ratios did not vary greatly with the exception of measurements made with the 1-pm beam (Fig. 2A). The CV of the Sr/Ca ratios for the 1-m beam was high and led to significant variation of Sr/Ca ratios in subsequent measurements at the same point (Table 2). The Sr/Ca ratio was least vari- able for the 7-pm beam in the high Sr/Ca region (Table 2). Limit of detection (3a) for Sr ranged from 80 ppm to 172 ppm in the low Sr/Ca region and from 299 ppm to 315 ppm in the high Sr/Ca region under the various beam diameters (Fig. 3A). Beam current had a significant effect on variation of Sr/Ca and limit of detection for Sr. The greatest variation in Sr/Ca ratios was observed at beam currents 5nA and 10 nA (Fig. 2B). The CV of Sr/Ca ratios was negatively related to beam current (r=-0.93; P>0.05) (Table 3). The CV of Sr was high in all treatments, ranging from 0.23 to 1.17 in the high Sr/Ca region and from 0.11 to 0.46 in the low Sr/Ca region. The CV of Ca was 0.04 for all beam configurations. The limit of detection of Sr as measured at the first sample NOTE Zimmerman and Nielsen: Measurement of strontium-to-calcium ratios in otoliths of anadromous salmonids 715 o 5 4 3 2 1 0 5 - 4 3 - 2 1 0 5 4 - 3 2 1 0 5 - 4 3 2 1 0 - 5 4 3 2 1 0 7 Lim •- 20 Mm B Sample sequence at same location Figure 2 Repeat measures of atomic Sr/Ca with (A) varying beam diameters, common beam current (50 nA), and (B) varying beam currents, common beam diameter (10 jim). Open circles represent low Sr/Ca regions and solid circles represent high Sr/Ca regions. for each beam-current configuration ranged from 99 ppm to 290 ppm in the low Sr/Ca region and from 312 ppm to 844 ppm in tiie high Sr/Ca region (Fig. 3B). Discussion In our experiment, for measuring Sr/Ca ratios in otoUths with WD-EM analysis, the TAP crystal was the best choice for the measurement of Sr because it provided the advan- tage of higher count rates and higher resolution of Sn Use of the PET crystal to measure Sr has not been reported in the literature. In fact, crystal choice is frequently not reported, making it difficult to know whether TAP or PET crystals have been used to measure Sr. Personal communication with several researchers confirmed that the TAP crystal is commonly used to measure Sr and has been cited as the crystal used (Kalish, 1990; Thresher et al., 1994). Given the different results possible with the use of different crystals to measure Sr, reporting the crystals used should be included in papers reporting Sr/Ca ratios measured on WD-EMs. Note that even though measurements of Sr with the TAP crystal appeared to be the best method for otoliths, the TAP crystal should not be used to measure Sr in materials con- taining Si because of analytical interference from Si. Variation of beam diameter has very little effect on the limit of detection of Sr, even at extremely high beam current densities. Rather, variation in beam current had 716 Fishery Bulletin 101(3) 1000 800 \ 1. 600 - a. \ W \.,^ 400 ^^--— ____^ 200 v^_____^ o o 1 7 15 20 25 0 10 20 30 40 50 Beam diameter (nm) Beam current (nA) Figure 3 Detection limit of Sr with (A) varying beam diameter and (B) varying beam current. Open circles represent low Sr/Ca regions and solid circles represent high Sr/Ca regions. Table 2 Coefficient of variation of Sr/Ca ratios and Sr (in parenthe- ses) in high and low Sr/Ca regions of otoliths. Beam diameter (|im) CV Sr/Ca and Sr in low region CV Sr/Ca and Sr in high region 1 0.076 (0.02) 0.416 (0.41) 7 0.082 (0.07) 0.009 (0.04) 15 0.218 (0.23) 0.041 (0.05) 20 0.107 (0.03) 0.028 (0.02) 25 0.133 (0.13) 0.030 (0.04) Table 3 Coefficient of variation of Sr/Ca ratios and Sr (in paren- theses) in high and low Sr/Ca regions of otoliths under varying beam current. Beam current (nA) CV Sr/Ca and Sr in low region CV Sr/Ca and Sr in high region 5 0.354 (0.94) 0.099 (0.11) 10 0.429 (1.11) 0.098 (0.41) 20 0.239 (1.12) 0.039 (0.45) 30 0.248 (0.23) 0.039 (0.46) 50 0.082 (1.17) 0.009 (0.11) significant efTects on the limit of detection of Sr. As a re- sult, higher beam currents (>20 nA) were appropriate for measuring Sr/Ca ratios in spite of beam damage observed at higher beam power densities. Beam diameters between 7 and 10m provide the best temporal resolution (i.e. cover- ing fewer daily increments). The lower CV of Sr/Ca ratios observed with the 7-m beam diameter was likely due to the lower temporal variation afforded to smaller beam diam- eters (compared to larger beam (liamctcrs) and lower error related to specimen damage (com[)ared to the 1-m beam). The lower CV of Sr/Ca ratios at the 7-pm beam diameter suggested that in spite of beam damage, the Sr/Ca ratio was not dramatically affected by beam damage. However, the increase in CV for the smaller diameters suggested that there are limits to usable beam densities. Greater precision of Sr/Ca measurements is critical to understanding life history of some species (Markowitz et al., 2000) or in situations where differences in environmental Sr/('a raios are less than those observed between ocean wa- ter and freshwatcrs (Rieman et al., 1994; Volk et al., 2000). Volk et al. (2000) found that timing of freshwater entry and length of freshwater residence by summer steelhead (O. mykiss) and spring chinook salmon had efTects on oto- lith core or primordia Sr/Ca levels. Summer steelhead and spring chinook enter freshwater and stay for up to several months before spawning. Volk et al. (2000) suggested that NOTE Zimmerman and Nielsen: Measurement of strontium-to-calcium ratios in otoliths of anadromous salmonids 717 significant egg development during this extended prespawn- ing freshwater residence led to a dilution of the Sr signature in these anadromous fish. Zimmerman and Reeves (2000, 2002) were able to distinguish between resident rainbow trout and summer steelhead in the Deschutes River, Or- egon, by comparing the Sr/Ca ratios in primordia and the first summer of juvenile growth (freshwater growth region). In essence the freshwater gi-owth region acts as a proxy for the freshwater environment and significantly higher Sr/Ca ratios in the primordia suggest an anadromous maternal origin. The greater precision of Sr measures afforded by higher beam currents may be important in distinguishing differences in seasonal ecotypes, such as summer steelhead and spring chinook salmon, or in distinuishing estuary habitats from freshwater and ocean environments. These results are applicable only to otolith calcium car- bonate in the mineral form of aragonite. Like Brown and Severin (1999), we have found that crystalline structure affects the distribution of Sr. Vateritic regions should be avoided when measuring Sr/Ca ratios in otoliths. In vater- itic portions of otoliths from chinook salmon and steelhead, Sr is often below our minimum detection limit of 43 ppm, yet the concentration of Ca does not differ from that found in aragonitic otolith regions (Zimmerman, unpubl. data). Studies offish migration between marine and freshwater environments are based on the general difference between Sr in marine and freshwater environments. Sr concentra- tions in seawater are generally an order of magnitude greater than in freshwaters (Bagenal et al., 1973; Kalish, 19901. Sr is substituted for Ca in the calcium carbonate matrix of the otolith at levels that correspond to those in the environment (Kalish, 1989; Farrell and Campana, 1996). Given this relationship, it has become a convention to report Sr as a fraction of Ca (Secor and Rooker, 2000). However, Secor and Rooker (2000) pointed out that Ca is relatively invariant in aragonitic otoliths and rarely var- ies more than 5% within an individual fish. At 8074 points sampled in the primordia, freshwater growth regions, and saltwater gi-owth regions of several species of salmonids the Sr/Ca ratio was entirely driven by differences in Sr (Zimmerman, unpubl. data). At these 8074 points, Sr was highly correlated with the Sr/Ca ratio {r^=99A5%) and Ca was not correlated with the Sr/Ca ratio (r-<0.01%). Given this relationship, increasing precision of Sr is desirable to increase precision of the Sr/Ca ratio. Our results suggest that tests of hypotheses related to Sr/Ca ratios can be conducted at higher beam power densi- ties than suggested by Gunn et al. ( 1992 ). High beam power densities resulting from higher beam current and beam diameter of 7 to 10-pm provide greater precision (spatial on the otolith and temporal in the life of the fish) of Sr. This is not true for studies of stock discrimination, such as those described by Thresher ( 1999), that rely on absolute values of multiple elements, including Sr. Acknowledgments Several people provided unpublished information concern- ing analjftic conditions and detection limits. Gordon Reeves, of the U.S. Forest Service Pacific Northwest Research Sta- tion, kindly provided office and laboratory space to CEZ. We thank Eric Volk, Ken Severin, and two anonymous review- ers for comments that improved this manuscript. Literature cited Babaluk, J. A., N. M. Halden, J. D. Reist, A. H. Kristofferson, J. L. Campbell, and W. J. Teesdale. 1997. Evidence for non-anadromous behaviour of Arctic charr iSalvelinus alpinus) from Lake Hazen, EUesmere Island, Northwest Territories, Canada, based on scan- ning proton microprobe analysis of otolith strontium dis- tribution. Arctic 50:224-233. Bagenal, T. B., F. J. H. MacKereth, and J. Heron. 1973. The distinction between brown trout and sea-trout by the strontium content of their scales. J. Fish Biol. 5: 555-557. Brown, R., and K. P. Severin. 1999. Elemental distribution within polymorphic inconnu iStenodus leucichthys) otoliths is affected by crystal structure. Can. J. Fish. Aquat. Sci. 56:1898-1903. Campana, S. E., S. R. Thorrold, C. M. Jones, D. Gunther, M. Tubrett., H Longerich, S. Jackson, N. M. Halden, J. M. Kalish, P. Piccoli, H. de Pontual, H. Troadec, J. Panfili, D. H. Secor, K. P Severin, S. H. Sie, R. Thresher, W. J. Teesdale, and J. L. Campbell. 1997. Comparison of accuracy, precision, and sensitivity in elemental assays of fish otoliths using the electron micro- probe, proton-induced x-ray emission, and laser ablation inductively coupled plasma mass spectrometry. Can. J. Fish. Aquat. Sci. 54:2068-2079. Farrell, J., and S. E. Campana. 1996. Regulation of calcium and strontium deposition on the otoliths of juvenile tilapia, Oreochromis niloticus. Comp. Biochem. Physiol. 115A:103-109. Gunn, J. S., I. R. Harrowfield, C. H. Proctor, and R. E. Thresher 1992. Electron probe microanalysis offish otoliths — evalua- tion of techniques for studying age and stock discrimination. J. Exp. Mar Biol. Ecol. 158:1-36. Kafemann, R., S. Alderstein, and R. Neukamm. 2000. Variation in otolith strontium and calcium ratios as an indicator of life-history strategies of freshwater fish species within a brackish water system. Fish. Res. 46:313-325. Kahsh, J. M. 1989. Otolith microchemistry: validation of the effects of physiology, age and environment on otolith composition. J. Exp. Mar Biol. Ecol. 132:151-178. 1990. Use of otolith microchemistry to distinguish the progeny of sympatric anadromous and non-anadromous salmonids. Fish. Bull. 88:657-666. Kawakami, Y, N. Mochioka, K. Morishita, T. Tajima, H. Nakagawa, H. Toh, and A. Nakazono. 1998. Factors influencing otolith strontium/calcium ratios in Anquilla japonica elvers. Env. Biol. Fishes 52:299-303. Limburg, K. E. 1995. Otolith strontium traces environmental history of subyearling American shad Alosa sapidissima . Mar Ecol. Prog. Ser 119:25-35. Markowitz, A., D. Grambole, F. Herrmann, W. J. Trompetter, T. Dioses, and R. W. Gauldie. 2000. Reliable micro-measurement of strontium is the key to cracking the life-history code in the fish otolith. Nucl. Instr and Meth. B. 168:109-116. 718 Fishery Bulletin 101(3) Potts, P. J. 1987. A handbook of silicate rock analysis. Chapman and Hall, New York, NY. Radtke, R. L. 1995. Otolith microchemistry of charr — use in life history studies. Nordic J. Freshwater Res. 71:392-395. Radtke, R. L., D. W. Townsend, S. D. Folsom, and M. A. Morrison. 1990. Strontiumxalcium concentration ratios in otoliths of herring larvae as indicators of environmental conditions. Env. Biol. Fish. 27:51-61. Rieman, B. E., Myers, D. L, and Nielsen, R. L. 1994. Use of otolith microchemistry to discriminate Oncor- hynchus nerka of resident and anadromous origin. Can. J. Fish. Aquat. Sci. 51:68-77. Secor, D. H. 1992. Application of otolith microchemistry analysis to investigate anadromy in Chesapeake Bay striped bass Morone saxatilis. Fish. Bull. 90:798-806. 1999. Specifying divergent migration patterns in the concept of stock: the contingent hypothesis. Fish. Res. 43:13-34. Secor, D. H., and J. R. Rooker 2000. Is otolith strontium a useful scalar of life cycles in estuarine fishes? Fish. Res. 46:359-371. Thresher, R. E. 1999. Elemental composition of otoliths as a stock delineator in fishes. Fish. Res. 43: 165-204. Thresher, R. E., C. H. Proctor, J. S. Gunn, and I. R. Harrowfield. 1994. An evaluation of electron-probe microanalysis of otoliths for stock delineation and identification of nursery areas in a southern temperate groundfish, Nemadactylus macropterus (Cheilodactylidae). Fish. Bull. 92:817-840. Toole, C. L., D. F Markle, and P H. Harris. 1993. Relationships between otolith microstructure, micro- chemistry, and early life history events in Dover sole, Microstomus pacificus. Fish. Bull. 91:732-753. Toole, C. L., and R. L. Nielsen. 1992. Effects of microprobe precision on hypotheses related to otolith Sr:Ca ratios. Fish. Bull. 90:421-427. Tzeng, W. N., K. P. Severin, and H. Wikstrom. 1997. Use of otolith microchemistry to investigate the environmental history of European eel, Anquilla anquilla. Mar. Ecol. Prog. Ser. 149:73-81. Volk, E. C, A. Blakley, S. L. Schroder, and S. M. Kuehner. 2000. Otolith microchemistry reflects migratory characteris- tics of Pacific salmonids: using otolith core chemistry to dis- tinguish maternal associations with sea and freshwaters. Fish. Res. 46:251-266. Zimmerman, C. E. 2000. Ecological relation of sympatric steelhead and resi- dent rainbow trout in the Deschutes River, Oregon. Ph.D. diss., 116 p. Oregon State Univ., Corvallis, OR. Zimmerman, C. E., and G. H. Reeves. 2000. Population structure of sympatric anadromous and non-anadromous Oncorhynchus mykiss: evidence from spawning surveys and otolith microchemistry. Can. J. Fish. Aquat. Sci. 57:2152-2162. 2002. Identification of steelhead and resident rainbow trout progeny in the Deschutes River, Oregon, revealed with oto- lith microchemistry. Trans. Am. Fish. Soc. 131:986-993. Fishery Bulletin 101(3) 719 Superintendent of Documents Publications Order Form *5178 I I YEd, please send me the following publications: Subscriptions to Fishery Bulletin for $55.00 per year ($68.75 foreign) The total cost of my order is $ . . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please tjrpe or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. IT'S EASY! Please Choose Method of Payment: I I Check Payable to the Superintendent of Documents I I GPO Deposit Account I I VISA or MasterCard Account (Credit card expiration date) -D 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 Guidelines for contributors Content of papers Articles Articles are reports of 10 to 30 pages (double spaced) that describe original research in one or a combination of the following fields of marine science: taxonomy, biology, genetics, mathematics (including modeling), statistics, engineering, eco- nomics, and ecology. Notes Notes are reports of 5 to 10 pages without an abstract that describe methods and results not supported by a large body of data. Although all contributions are subject to peer review, responsi- bility for the contents of articles and notes rests upon the authors and not upon the editor or the publisher It is therefore important that authors consider the contents of their manuscripts care- fully. Submission of an article is un-derstood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts must be written in English. Authors whose native language is not EngUsh are strongly advised to have their manuscripts checked for fluency by English-speaking colleagues prior to submission. Preparation of papers Text TMe page should include authors' full names and mailing addresses (street address required) and the senior author's telephone, fax number, e-mail address, as well as a list of key words to describe the contents of the manuscript. Abstract must be less than one typed page (double spaced) and must not contain any citations. It should state the main scope of the research but emphasize the author's con- clusions and relevEmt findings. Because abstracts are circulated by abstracting agencies, it is impor- tant that they represent the research clearly and concisely. General text must be typed in double- spaced format. A brief introduction should state the broad significance of the paper; the remainder of the paper should be divided into the following sec- tions: Materials and methods. Results, Discussion (or Conclusions), and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdi- vision (i.e. there can be no subdivision without at least two subheadings). The entire text should be intelUgible to interdisciplinary readers; therefore, all acronyms and abbreviations should be written out and all lesser-known technical terms should be defined 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 Sci- entific style and format: CBE manual for authors, editors, and publishers (6th ed.) for editorial style and the most current issue of the Aimerican Fish- eries Society's common and scientific names of fishes from the United States and Canada for fish nomenclature. Dates should be written as fol- lows: 11 November 1991. Measurements should be expressed in metric units, e.g. metric tons (t). The numeral one (1) should be typed as a one, not as a lower-case el (1). Footnotes Use footnotes to add editorial comments regarding claims made in the text and to document unpub- lished works or works with local circulation. Foot- notes should be numbered with Arabic numerals and inserted in 10-point font at the bottom of the first page on which they are cited. Footnotes should be formatted in the same manner as citations. If a manuscript is unpubhshed, in the process of review, or if the information provided in the footnote has been conveyed verbally, please state this information as "unpubl. data," "manuscript in review," and "personal commun.," respectively. Authors are advised wherever possible to avoid ref- erences to nonstandard literature (unpublished lit- erature that is difficult to obtain, such as internal reports, processed reports, administrative reports, ICES council minutes, IWC minutes or working papers, any "Research" or 'Svorking" documents, laboratory reports, contract reports, and manu- scripts in review). If these references are used, please indicate whether they are available fi'om NTIS (National Technical Information Service) or from some other public depository. Footnote format: author (last name, followed by first-name initials); year; title of report or manuscript; type of report and its administrative or serial number; name and address of agency or institution where the report is filed. Literature dted The Hterature cited section comprises works that have been published and those accepted for pub- lication (works in press) in peer-reviewed jour- nals and books. Follow the name and year system for citation format. In the text, write "Smith and Jones (1977) reported" but if the citation takes the form of parenthetical matter, write "(Smith and Jones, 1977)." In the literature cited section, list citations alphabetically by last name of senior author: For example, Alston, 1952; Mannly, 1988; Smith, 1932; Smith, 1947; Stalinsky and Jones, 1985. Abbreviations of journals should conform to the abbreviations given in the Serial sources for the BIOSIS previews database. Authors are responsible for the accuracy and completeness of all citations. Literature citation format: author (last name, followed by first-name initials); year; title of report or article; abbreviated title of the journal in which the article was published, volume number, page numbers. For books, please provide publisher, city, and state. Tables Tables should not be excessive in size and must be cited in numerical order in the text. Headings in tables should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted (use italic arable numerals for footnote markers). Use asterisks only to indicate probability in statistical data. Place table legends on the same page as the table data. We accept tables saved in most spread- sheet software programs (e.g. Microsoft Excel). Please note the following: • Use a comma in numbers of five digits or more (.e.g. 13,000 but 3000). • Use zeros before all decimal points for values less than one (e.g. 0.31). Rgures Figures include line illustrations, computer-gener- ated line graphs, and photographs (or slides). They must be cited in numerical order in the text. Line illustrations are best submitted as original draw- ings. Computer-generated line graphs should be printed on laser-quality paper Photographs should be submitted on glossy paper with good contrast. All figures are to be labeled with senior author's name and the number of the figure (e.g. Smith, Fig. 4). Use Helvetica or Arial font to label ana- tomical parts (line drawings) or variables (graphs) within figures; use Times Roman bold font to label the different sections of a figure (e.g. A, B, C). Figure legends should explain all symbols and abbreviations seen within the figure and should be typed in double-spaced format on a separate page at the end of the manuscript. We advise authors to peruse a recent issue of Fishery Bulletin for stan- dard formats. Please note the following: • Capitalize the first letter of the first word of axis labels. • Do not use overly large font sizes to label axes or parts within figures. • Do not use boldface fonts within figures. • Do not create outline rules around graphs. • Do not use horizontal lines through graphs. • Do not use large font sizes to label degrees of longitude and latitude on maps. • Indicate direction of degrees longitude and latitude on maps (e.g. 170°E). • Avoid placing labels on a vertical plane (except on y axis). •Avoid odd (nonstandard) patterns to mark sections of bar graphs and pie charts. Copyright law Fishery Bulletin, a U.S. government pubhcation, is not subject to copyright law. If an author wishes to reproduce any part of Fishery Bulletin in his or her work, he or she is obliged, however, to acknowledge the source of the extracted literature. Submission of papers Send four printed copies (one original plus three copies ) — clipped, not stapled — to the Scientific Edi- tor, at the address shown below. Send photocopies of figures with initial submission of manuscript. Original figures will be requested later when the manuscript has been accepted for publication. Do not send your manuscript on diskette until requested to do so. Dr. Norm£m Bartoo National Marine Fisheries Service, NCAA 8604 La Jolla Shores Drive LaJoUa, CA 92037 Once the manuscript has been accepted for pub- lication, you will be asked to submit a software copy of your manuscript. The software copy should be submitted in WordPerfect or Word format (in Word, save as Rich Text Format). Please note that we do not accept ASCII text files. Reprints Copies of published articles and notes are avail- able 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 101 Number 4 October 2003 Fishery Bulletin U.S. Department of Commerce Donald L Evans Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fistienes .^>^^'''°'%. ^^ATES 0» '^ 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 C 15700, Seattle, WA 981 15-0070. 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: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 domestic and .$35.00 foreign. See back for order form. ^ Scientific Editor Dr. Norman Bartoo Associate Editor Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 981 15-0070 Editorial Committee Dr. Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr Theodore W. Pietsch Dr. Joseph E. Powers Dr. Harald Rosenthal Dr. Fredric M. Serchuk Dr. George Watters University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service 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. Slate and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 101 Number 4 October 2003 Fishery Bulletih '/'O'lne Biological Laboratory/ Wooos Hole Ocoanngraphic Institution Library NOV 2 1 2003 WQSUft I iijiNi tiit\ tjiUi Contents The conclusions and opinions expressed in Ftshery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service INOAA) or any other agency or institution The National Marine Fisheries Service (NMFS) 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 721-731 Baum, Julia K., Jessica J. Meeuwig, and Amanda C. J. Vincent Bycatch of lined seahorses (Hippocampus erectus) in a Gulf of Mexico shrimp trawl fishery 732-736 Bjorndal, Karen A., Alan B. Bolten, and Helen R. Mattins Estimates of suiA/ival probabilities for oceanic-stage loggerhead sea turtles (.Caretta caretta) in the North Atlantic 737-744 Chen, Yong, Margaret Hunter, Robert Vadas, and Brian Beal Developing a growth-transition matnx for the stock assessment of the green sea urchin (Strongylocentrotus droebachiensis) off Maine 745-757 de Lestang, Simon, Norman G. Hall, and Ian C. Potter Reproductive biology of the blue swimmer crab (Portunus pelagicus, Decapoda: Portunidae) in five bodies of water on the west coast of Australia 758-768 Dew, Jodi R., Jim Berkson, Eric M. Hallerman, and Standish K. Allen Jr. A model for assessing the likelihood of self-sustaining populations resulting from commercial production of tnploid Suminoe oysters (Crassostrea ahakensis) in Chesapeake Bay 769-777 Diaz-Jaimes, Pindaro, and Manuel Uribe-Alcocer Allozyme and RAPD vanation in the eastern Pacific yellowfin tuna (Thunnus albacares) 778-789 Govoni, John Jeffrey, Elisabeth H. Laban, and Jonathan A. Hare The early life history of swordfish iXiphias gladius) in the western North Atlantic 790-799 Heales, Donald S., David T. Brewer, You-Gan Wang, and Peter N. Jones Does the size of subsamples take from multispecies trawl catches affect estimates of catch composition and abundance? Fishery Bulletin 101(4) 800—808 Laidig, Thomas E., Donald E. Pearson, and Lorraine L. Sinclair Age and growth of blue rockfish (Sebastes mystinus) from central and northern California 809—821 Marin E., Baumar J., Antonio Quintero, Dany Bussiere, and Julian J. Dodson Reproduction and recruitment of white mullet iMugil curema) to a tropical lagoon (fVlargarita Island, Venezuela) as revealed by otolith microstructure 822—834 McDonough, Christopher J., William A. Roumillat, and Charles A. Wenner Fecundity and spawning season of striped mullet (Mugil cephalus L.) in South Carolina estuaries 835-850 Nelson, Peter A. Manne fish assemblages associated with fish aggregating devices (FADs): effects of fish removal, FAD size, fouling communities, and prior recruits 851—859 Pajuelo, Jose G., Jose M. Lorenzo, and Muriel Gregoire Age and growth of the bastard grunt iPomadasys inasus) Haemulidae) inhabiting the Canarian archipelago. Northwest Africa 860-873 Punt, Andre E. Evaluating the efficacy of managing West Coast groundfish resources through simulations 874-888 Purves, Martin G., David J. Agnew, Guillermo Moreno, Tim Daw, Cynthia Yau, and Graham Pilling Distribution, demography, and discard mortality of crabs caught as bycatch in an experimental pot fishery for toothfish (Dissostichus deginoides) in the South Atlantic 889-899 Stabenau, Erich K., and Kimberly R. N. Vietti The physiological effects of multiple forced submergences in loggerhead sea turtles (Carena caretta) 900-909 Stephenson, Peter C, and Norm G, Hall Quantitative determination of the timing of otolith ring formation from marginal increments in four marine teleost species from northwestern Australia Notes 910-914 Chan, Ricky W, K., Patricia I. Dixon, Julian G. Pepperell, and Dennis D. Reid Application of DNA-based techniques for the identification of whaler sharks (Carcharhinus spp) caught in protective beach meshing and by recreational fishenes off the coast of New South Wales 915-922 Ebert, Thomas A., and John R. Southon Red sea urchins iStrongylocentrotus franascanus) can live over 100 years; confirmation with A-bomb '''carbon 923-932 Fulling, Gregory L., Keith D Mullin, and Carrie W. Hubard Abundance and distribution of cetaceans in outer continental shelf waters of the U.S. Gulf of Mexico 933-938 Hata, David, and Jim Berkson Abundance of horseshoe crabs iLimulus polyphemus) in the Delaware Bay area 939-948 Kerstetter, David W., Brian E. Luckhurst, Eric D. Prince, and John E. Graves Use of pop-up satellite archival tags to demonstrate sun/ival of blue marlin (Makaira nigricans) released from pelagic longline gear 949-950 2003 revieweis 951-960 2003 index 961 Subscription form 721 Abstract— Bycatch studies have largely ignored population level efTects on fish species of little commercial interest. Here we analyze bycatch of the lined seahorse ^Hippocampus erectus) in the bait-shrimp trawl fishery in Hernando Beach, Florida, providing the first fisheries data for this species. Based on catch per unit of effort (CPUE), size, sex, and reproductive status of trawled H. erectus, 1) approximately 72,000 seahorses were caught annu- ally by this fleet, from a population of unknown size, 2) trawling affected pop- ulation cohorts differentially because of temporal and spatial variation in CPUE and population size, and 3) a greater proportion of females than males was removed in trawling. Our findings sug- gest that trawling may affect seahorse populations through direct mortality, social disruption, and habitat damage. However, the lack of specific abundance or catchability estimates for H. erectus means that the precise impact of trawl- ing on this fish remains uncertain. This paper focuses attention on the need for research and monitoring of small fishes that are caught incidentally in nonselective gear. Bycatch of lined seahorses (Hippocampus erectus) in a Gulf of Mexico shrimp trawl fishery Julia K. Baum Jessica J. Meeuwig Amanda C. J. Vincent Proiect Seahorse Department of Biology McGill University, 1205 Dr Penfield Ave. Montreal, Quebec, H3A 1B1, Canada Present address (for J. K Baum); Department of Biology Dalhousie University 1355 Oxford St. Halifax, Nova Scotia, B3H 4J1, Canada E-mail address (for J K Baum) baum@mathstat.dal.ca Manuscript approved for publication 2 June 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull 101:721-731 (2003). The incidental capture of marine or- ganisms is now recognized as a seri- ous problem in fisheries management and marine conservation (Alverson et al., 1994; lUCN, 1996; Alverson, 1997; Jennings and Kaiser, 1998; FAO, 1999). Shrimp trawl fisheries are the single greatest source of bycatch, accounting for 35% of the world's total bycatch (Alverson et al., 1994). Bycatch research has focused on marine megafauna, sea- birds, and commercially important fish species (see as examples Polacheck, 1989; Graham, 1995; Weimerskirch et al., 1997; Julian and Beeson, 1998; Pikitch et al., 1998; Galloway and Cole, 1999; Diamond et al., 2000). The conser- vation impacts of bycatch for noncom- mercial fishes and invertebrate species remain largely unstudied (but see Chan and Liew, 1986; Pettovello, 1999; Milton, 2001). The few studies that have evaluated incidental capture of these species have focused on survival rates of individuals (Hill and Wassen- berg, 1990, 2000; Kaiser and Spencer, 1995; Probert et al., 1997; Mensink et al., 2000) without addressing popula- tion level effects of bycatch. However, even species that comprise only a small portion of the bycatch in a fishery may experience significant impacts of inci- dental harvest on their population size and structure. Seahorses are among those spe- cies inferred to be greatly affected by nonselective fishing gear, both because intense trawling often covers seahorse habitat and because their life history traits likely render these fishes vul- nerable to overexploitation (Vincent, 1996). Most studied seahorse species are strictly monogamous (i.e. sexually and socially), meaning that removal can disrupt pairs and may reduce reproduc- tive output (e.g. Vincent, 1995; Vincent and Sadler, 1995; Kvarnemo et al., 2000; Perante et al., 2002). Obligatory parental care by males, combined with relatively low fecundity, may reduce the potential for population recovery from overexploitation, although potentially high survival of young may also offset this apparent cost. In addition, sparse distributions and low mobility suggest that seahorses will be slow to recolonize depleted areas (Perante et al., 2002; Vincent et al.^). Seahorses derived from bycatch ap- pear to be contributing greatly to the large and growing international trade in these fishes (Vincent, 1996; Vincent and Perry2). Consumer demand for seahorses — both dried for traditional medicine and curiosity trades, and, less frequently, live for the aquarium trade — is very high (Vincent, 1996). Global demand for seahorses has sur- passed supply and therefore trade has increased and expanded geographically ' Vincent, A. C. J., K. L. Evans, and A. D. Marsden. 2003. Home range behaviour of the monogamous Australian seahorse. Hippocampus whitei. Manuscript in review. 722 Fishery Bulletin 101(4) (Vincent, 1996; Vincent and Perry'^), placing populations around the world under greater pressure. Much of this market demand is met from retention of incidental land- ings in shrimp trawls. Where no market has yet developed, incidentally caught seahorses are discarded and the sur- vival rate for these discarded seahorses is unknown. Trade records and anecdotal evidence from other coun- tries indicate that the United States has imported and exported considerable numbers of both live and dried seahorses in recent years (Vincent, 1996; Vincent and Perry^). In Florida, the primary source of live seahorses in the United States (Larkin and Degner, 2001), these fishes ranked as the seventh most economically important orna- mental fish group from 1990 to 1998, and seahorse landings rose 184% during this period, whereas landings of each of the more valuable fish groups declined (Adams et al., 2001). Many seahorses in the United States are probably obtained from bycatch of the shrimp trawl fisheries that operate in known seahorse habitats along the Atlantic coast of the United States. Our objective was to document bycatch of the lined sea- horse (Hippocampus erectus) in a live-bait shrimp trawl fishery. This fish is often retained for the aquarium trade. We quantify number, sex, size, and reproductive status of trawl-caught seahorses and examine how these parameters vary temporally and spatially. We also comment on poten- tial conservation concerns resulting from this fishery. Materials and methods We focused our study on the lined seahorse (Hippocam- pus erectus) because it was caught much more often than its sympatric congenerics in the Gulf of Mexico — the longsnout seahorse (H. reidi) and the dwarf seahorse (H. zosterae). Hippocampus erectus is a large, deep-bodied seahorse (adult height 5.5 to 18.5 cm: Lourie et al., 1999) and has a geographic range that extends from southern Canada to Argentina. Most species of seahorse have short lifespans and low fecundity: Hippocampus erectus lives for about four years (Lourie et al., 1999) and has broods of about 100-1500 young (Teixeira and Musick, 2000). Hip- pocam.pus erectus is found in shallow waters and offshore to depths of over 70 m, primarily in mangroves and seagrass beds (Vari, 1982). Like many other seahorse species, H. erectus is listed as Vulnerable (A2cd) by the International Union for Conservation of Nature and Natural Resources (lUCN, 2002), based on suspected declines resulting from habitat degradation and exploitation. However, as is the case for many small fishes, there is little information known about the biology of H. erectus, and no fishery data exist for it. Seahorse bycatch was assessed in the live-bait shrimp trawl fishery operating from Hernando Beach, Florida (Fig. 1). The fishery, using roller beam trawls, targets pink shrimp iPenaeus duorarum) at night in seagrass beds and relocates seasonally in Florida. Hernando Beach was cho- '^ Vincent, A. C. J., and A. Perry (eds). 2003. Global trade in seahorses. Manuscript in prep. sen as our study site because it is a moderate-size fishing port with 31 licensed trawlers and is active during the summer sampling period. Boats were equipped with trawls that had a slotted roller along the bottom of the frame, and stainless steel finger bars attached vertically, 5 cm apart along the length of the frame, to limit collection of benthic substratum and other debris (Berkeley et al., 1985). Trawls were towed from each side of the vessel in four configura- tions: 1) one trawl per side, each measuring either 3.66 m, 4.27 m, or 4.88 m in length, or 2) two trawls per side, each measuring 3.66 m in length. Net mesh sizes were 3.18-3.81 cm and the tail bag stretched mesh size was 2.54-3.18 cm. Trawls usually lasted between 30 and 60 minutes, and fish- ermen made multiple successive trawls in a night. Shrimp were culled from the catch and held live onboard in aerated holding tanks. Most bycatch was discarded overboard, al- though some fishermen retained certain species, including seahorses, for sale as aquarium fishes. Bycatch sampling and data set description Data on seahorse bycatch in the live-bait shrimp fishery were collected on 95 fishing nights, from June to August 1998 and from June to July 1999, using three methods: 1) we sampled bycatch onboard on 50 nights; 2) we recorded seahorse bycatch data onshore on 14 nights from fishers who retained seahorses to sell; and 3) we received data from fishermen on their seahorse catches (including loca- tion, time, number of trawls, and number of seahorses per trawl) on a total of 31 nights. Onboard sampling was semistratified in that we targeted our sampling to cover all lunar and tidal phases and a variety of areas. However, we were dependent on fishersmen's decisions about the time and location of tows. We also collected anecdotal informa- tion from 14 experienced fishermen about seahorse catches over time. During onboard sampling, we recorded the number of seahorses caught, start and end time of trawl, depth and lo- cation (Loran co-ordinates) of trawl, tidal and lunar phases, and presence or absence of all bycatch species, including biogenic habitat species. Hippocampus erectus found in the catch were placed in a container of surface water while measurements were made. They were then released, except when fishermen chose to retain them for sale. We measured, weighed, determined the sex, and recorded the life history stage, reproductive status, and any injury for each seahorse. Measurements were taken according to Lourie et al. (1999) and included standard length, defined as the length from the tip of the snout to the opercular ridge and from the opercular ridge to straightened tail tip. Seahorses that had lost tail rings were not included in the length analysis. Unless precluded by logistic constraints, wet weights were obtained onboard by using a 60-g Pesola spring scale and onshore by using a 200-g Ohaus electronic balance. Adult males were distinguished from females and juveniles by the presence of a brood pouch. The standard length of the smallest seahorse with a brood pouch (105.3 mm) was used as the division between adults andjuveniles with the assumption that males and females matured at the same size. Such an assumption may overestimate the Baum et al.: Bycatch of Hippocampus erectus in a Gulf of Mexico shirimp trawl fisfiery 723 United Stales of America GULF OF MEXICO ATLANTIC Florida OCEAN OI& Ofto C2003 - Km 82^50' 82^40' Figure 1 Map of the eastern Gulf of Mexico, showing the study site of Hernando Beach on the western coast of Florida. The shore is shaded; additional solid lines indicate depth contours in feet. Letter and number combinations (e.g. N3) represent fishing areas along the coast. number of adult females because males that matured after that size would have been included in our analysis as females. We defined males as reproductively active if they were pregnant or had recently released young (as indicated by a loose pouch). Female reproductive state was not included in this analysis because it is difficult to determine reliably. We defined mortality to include sea- horses already dead when the net was hauled and those that died onboard. We evaluated temporal and spatial patterns in catch per unit of effort (CPUE), standard length, population struc- ture, and reproductive status of the seahorse bycatch. Spe- cifically, we tested 1) temporal effects of year, lunar and tidal phase, and 2) spatial effects of area and depth. In our estimates of seahorse CPUE, we used standard length instead of biomass because female weight changes with egg hydration and male weight increases greatly when carrying embryos. Calculating CPUE per length (m) of roller beam trawl controlled for variation in gear size. The sampling unit was thus defined as the total or cumulative standard length of seahorses caught during each tow (per hour), per meter of trawl (per meter). Lunar phases were defined as continuous variables by converting lunar day to its angle, 6, based on a cycle of 29.5 days (=360°), with new moon defined as 6 = 0°. These angles were then converted to their cosine and sine functions for inclusion in linear regression (deBruyn and Meeuwig, 2001). Tides were semidiurnal in the Gulf of Mexico, ranging 1.3 m in tidal level in the study area. High and low tides were defined as each lasting two hours, with the remaining time classified as ebb or flood accordingly. Spatial effects were analyzed by dividing the total fishing ground into eight subareas ac- 724 Fishery Bulletin 101(4) cording to their position with respect to the depth contour and by identifying discrete geographical clusters of trawls (Fig. 1). We then compared varia- tion among fishing areas within years for those areas with at least five observ'ations (areas II, 12, Nl, N2, SI, S2 in 1998, and areas Nl, N3 and N4 in 1999). Interannual spatial comparisons were not possible because there was little overlap in sampled areas between the two years. Statistical analysis We based CPUE estimates on all trawls (?!=445). Statistical analyses evaluating temporal and spatial variation in seahorse bycatch included only those trawls with nonzero seahorse observa- tions (;i=205). The analysis required the data or their residuals to be normal; this was achieved by inverse hyperbolic sine transformations following exclusion of zeros (Zar, 1996). It should be noted that by excluding zeros we overestimated seahorse CPUE and we lost information about areas where seahorses were absent or rare. Our analyses should be interpreted as applying to locations where and times when seahorses were found in sufficient numbers as to be caught. We examined the data using ANOVA, AN- COVA, linear regression and chi-square analyses (Zar, 1996; SYSTAT, version 7.0, SPSS Inc., Chicago, ID. All two-way and three-way ANOVAs included tests for interactions. We used a general linear model because the data were unbalanced. Interactions were removed from the model if they were found to be nonsignificant. Models were then rerun, followed by pairwise Tukey tests to indicate where significant differences occurred (Zar, 1996; SYSTAT, version 7). We report results for the final ANOVA only. The Yates correction was applied to 2x2 chi-squares (Zar, 1996). All significance levels were set to reject Hq at P<0.05 and all means are reported with standard errors. Results Bait shrimp fishermen trawled from 11 to 24 km offshore, between 1.8 and 6.4 m deep water (mean=3.76 ±0.87 m). Fishermen typically left port between 17:00 and 19:00 and spent 5.8 ±0.23 h actively trawling per night (n=50 nights). Trawls lasted 40.2 ±11.4 min (n=445) on average, and fishermen usually set 8 to 9 trawls per night. Distance trawled could not be estimated because we were unable to track trawler trajectories continuously and because they changed direction during the tows. The benthic habitat was composed primarily of seagrass iThalassia testudinum)hut also included algae, coral, and sponge. Bycatch included at least 118 species of fishes, invertebrates, and marine flora. Catch per unit of effort Hippocomptis erectus was the only seahorse species com- monly caught in this fishery. Almost half of the trawls (46%) 6- 1998 • € O » • C 1999 O » • © O I 4 O O • 3 Q. U 4 i y^^WY^^ 150 160 170 180 190 200 210 220 230 Day of year Figure 2 Variation in mean CPUE estimates by lunar phase transformed to inverse hyperbolic sine. The curve represents the predicted values as a function of the cosine ( 1998) and sine ( 1999) of day. The lunar phases for both years are indicated. Filled circles and black line are for 1998; open circles and gray line are for 1999. caught H. erectus, and the number per trawl ranged from 0 to 16, whereas noH. reidi and only two of the much smaller species, H. zosterae, were caught. In total, 916 H. erectus were caught during the 95 documented fishing boat nights of the two fishing seasons, resulting in an overall mean of 9.64 seahorses per fishing boat night. Mean CPUE for H. erectus was 24.25 ±2.15 mm/{hxm), about one and a half seahorses per hour per boat. If only trawls with seahorses were included, CPUE was 52.52 ±3.80 mm/(hxm), or about three seahorses per hour per boat (;i=205). Very high CPUE was recorded on three nights: 16 July 1998 (mean CPUE=122.0 ±22.5 mmAhxm), n=12 trawls), 28 June 1999 (mean CPUE=118.1 ±24. mm/(hxm), « = 12 trawls) and 30 June 1999 (mean CPUE=154.9 ±36.6 mm/(hxm), n=8 trawls). Bycatch is characterized by a high number of low catches and infrequent large catches. Because the large catches more likely reflect the spatial-temporal distribu- tion characteristics of fish stocks rather than outliers of the data (Ortiz et al. 2000), we analyzed the entire dataset and then tested the robustness of our models by excluding these three nights in order to assess their influence on the CPUE patterns. CPUE of nonzero trawls varied between years and with lunar phase (Table 1), but not with tidal phase (P=0.15). Trawls captured significantly more seahorses in 1999 than in 1998 (P<0.0005). The effect of lunar phase varied be- tween years: CPUE was highest on the lunar third quarter in 1998, but only weakly significant and had slightly higher CPUE on the full moon in 1999 (Fig. 2). The temporal varia- tion in CPUE was largely driven by the three high CPUE nights, but the effect persisted when these were excluded (Table 1). Baum et al Bycatch of Hippocampus erectus in a Gulf of Mexico sfnrimp trawl fishery 725 Table 1 General linear model of effects of year, lunar phase, and ai'ea on Hernando Beach bait-shrimp trawl fishery. the CPUE (n- m/lh xm) for nonzero H. erectus bycatch trawls in the CPUE by traw 1 lexcluding CPUE by trawl CPUE of three highest nights) Source n F P n F P Year 205 15.2 <0.0005 173 3.9 0.049 Lunar phase 205 173 cost 6) 19.1 <0.0005 4.1 0.045 sinte) 8.9 0.003 1998 areas III, 12, N1,N2, SI, S2) 116 7.4 <0.0005 105 2.3 0.037 1999 areas INI, N2,N3.N4) 87 8.3 <0.0005 68 0.3 0.85 In both years, there were significant differ- ences in nonzero CPUE trawls among sites. In 1998, CPUE was significantly higher in S2 than in 12, Nl, N2 and SI iFig. 3A, Table 1). In 1999. CPUE was significantly higher in area Nl than in areas N3 and N4 IFig. 3B, Table 1). However, both of these spatial patterns, like the lunar patterns, were driven primar- ily by the high CPUE nights. Removing the three outliers left a significant difference in CPUE by area in 1998 only iTable 1). CPUE did not vary with depth of the fishing ground IP=0.67). Size of seahorses Mean standard length of adults (standard length >105.3 mm) was 139.5 ±21.7 mm (n=465, range 105.3-202 mm) and mean weight was 11.6 ±5.5 g (n=232, range 3-31 g). Hippocam- pus erectus was sexually dimorphic: males had a brood pouch and were significantly longer than females (n=465, P<0.0005, Table 2), and had a greater weight to standard length ratio, although this latter difference was relatively weak (P=0.04). Juveniles (standard length <105.3 mm) had a mean standard length of 83.3 ±16.7 mm ln=65, range 41.4-105 mm) and a mean weight of 2.4 ±1.0 g (?!=38, range 0.9-4.0 g). In adults, standard length varied by year {n=425, F=l.\, P=0.008) and by lunar phase (7!=425, cos(e):P=5.4, P=0.02, sin(0):P=7.7, P=0.006). Mean standard length was greater in 1998 than in 1999 and highest on the new moon. In 1998 significantly larger adult sea- horses were caught in areas 12 and S2 than in II and SI (n=229, P=13.7, P<0.0005; Fig. 4). There was no effect of area in 1999 (n=2\2, F=1.9, P=0.14). Standard length was not re- lated to depth or tide. 1 1() - A 1 IIHI - go - II B .so - 70 - hO ■ -Ml - <• .\Q Catcfi per unit of effort (mm/(hym)) 1 11 B 1 12 Nl N2 SI S2 1 B J\0 - (ill - 40 - ^ >0.10) or tidal phase (P>0.90). However, significantly more juveniles were caught during the first quarter than during other lunar phases (Table 3). Fishing area had a significant effect on size class, with the highest numbers of juveniles caught in N2 and N3 (Table 3). The sex ratio (males as fraction of total) of 0.42 differed significantly from a 1:1 ratio (;t:;-^= 19.56, df=l. P<0.001). The sex ratio did not vary temporally or spatially (Table 3), but it did vary as a function of size class. There were proportionally more males in the larger size class (>150 mm, 0.61) than in the smaller size class (<150 mm, 0.30) (X,^=14.95, df=l,P<0.005). About 25% of the male seahorses captured in 1998 were considered to be reproductively active, whereas fewer than 1% were reproductively active in 1999. Indeed, male reproductive activity was higher in 1998 than 1999 even after controUing for smaller male size in 1999 (Table 3). The proportion of reproductively active males did not vary with lunar or tidal phase but did vary significantly with area Baum et al : Bycatch of Hippocampus erectus in a Gulf of Mexico sfirimp trawl fisfiery 727 Table 2 Descriptive statistics (sample size [n], mean, standard deviation [SD], minimum [mini and maximum [max]) for standard length and weight of female (F) and male (M) seahorses. P values and sample sizes (n) indicate results off-tests evaluating sexual dimor- phism in SL and weight. Sex n mean SD min max Standard length (mm) P<0.0005,n=465 Weight (g) P<0.002, n=232 M F M F 201 264 105 140 146.1 134.4 12.5 11 0 21.9 20.1 5.0 5.8 105.3 105.3 6.0 30 200.5 202.0 27.0 31.0 Table 3 Contingency tables on effects of year, lunar phase, and area on population structure of seahorses in the bycatch of the Hernando Beach bait shrimp trawl fishery. Yates corrections were applied to 2x2 contingency tables. Source Juveniles: Adults Sex ratio df df X~ Reproductive state df P Year Lunar phase Area 2.46 11.00 8.84 >0.10 <0.025 <0.05 0.15 3.69 13.3 >0.50 >0.25 >0.05 27.89 7.80 21.64 <0.001 >0.05 <0.001 M (Table 3); almost half of the reproductively ac- tive males were found in one area (S2),and 83% of them were caught on one of the three nights with very high CPUE (16 July 1998). Mortality and injury Fewer than 1% of seahorses died during tows or sorting, but 4,7% (;i=28 of 588 seahorses) lost tail rings. The mean loss was 22 of the usual 36 tail rings (Lourie et al., 1999), or 61% of the tail (range=8-31 rings lost). Approximately 14% of the losses (n=4 of 588 seahorses) appeared to be the result of new wounds, probably caused by the focal trawl. There was no effect of year (P=0.25), sex (P=0.75), or reproductive status (P=0.75) on incidence of seahorse injury. Postre- lease mortality is unknown. Discussion Direct effects of the trawl fishery on seahorse mortality We estimate that this fleet catches almost 72,000 sea- horses incidentally per annum, based on the mean CPUE and given that 31 boats fished approximately 240 nights per year. Most seahorses were returned to the wild in the Hernando Beach fishery, but this may not be the case in other live-bait shrimp trawl fleets in Florida (Vincent, pers. _a jiMi i wkk 41) Ml Sll M)ll i:0 1411 IM) IXd :iKi Standard length (mm) Figure 5 Length-frequency histogram for seahorse bycatch in 1998 and 1999. obs.). We could not determine the potential fishing-induced mortality for the Hernandez Beach H. erectus population, even when all trawled seahorses were retained, because seahorse catchability in roller beam trawl gear is unknown and no studies have estimated H. erectus density in the wild. Focal behavioral studies on congenerics similar in 728 Fishery Bulletin 101(4) size to H. erectus have found varied densities: H. comes, an exploited species associated with coral reefs in the tropics, had localized densities of 0.019/m- in a marine protected area, and much less elsewhere (Perante et al., 2002), and an unexploited species, H. whitei, had localized densities of 0.088-0. 215/m^ in a study area, and no seahorses were found over large adjacent areas (Vincent et al.'). Our study also suggests very patchy distributions of//, erectus (549f of the trawls had no seahorses at all and the number of seahorses per trawl set ranged from 0 to 16). Although variation in CPUE may reflect differential catchability by habitat, we suggest that in areas where seahorses were caught, temporal rather than spatial ef- fects drove CPUE. It is difficult to make conclusions about variation in CPUE because data were unbalanced, in that the areas trawled differed between years and among lu- nar phases. However, analysis of variance on a subset of data for three sites (Nl, N2, N3) on three lunar phases (1^' quarter, full moon, 3'''' quarter) for which we had data in both years («=149 trawls), indicated that there was a strong effect of year, a weaker effect of lunar phase, and no effect of site. These results suggest that CPUE was mainly affected by temporal variation. Lunar patterns in CPUE as a result of fish behavior and ecology are common (e.g. Parrish, 1999). This would be consistent with observations for other species of seahorses; H. comes in the Philippines (Vincent et al.-*) and H. spinosissimus and H. trimaculatus in Vietnam (Meeuwig et al.**) exhibited patterns in CPUE with respect to lunar phase, although these species were also distributed in patches in space. Data from this study suggest that the H. erectus popu- lation was spatially structured. In 1999, the mean size of incidentally caught adult seahorses decreased, reflecting the absence of the largest size class of males and an in- crease in smaller females that year (Fig. 5). We attribute this difference to spatial structuring: the shallower areas (12, S2) where the largest male and female seahorses were caught in 1998 were not fished during the 1999 sampling season. Most of the seahorse bycatch were adult H. erectus; the dearth of juvenile H. erectus (and dwarf seahorses, H. zosterae) in the trawls reflects low catchability or retention due to mesh size. Similar proportions of juvenile seahorses were caught over the two sampling seasons. The ratio of juveniles to adults appears to be temporally influenced I proportionally more juveniles were caught on new moons), but this variation probably also reflects spatial structuring because these trawls occurred primarily in deeper offshore areas (N2, N3) that were fished almost exclusively during this lunar phase. Perhaps//, erectus undergoes ontogenetic movement, between juvenile and adult life history stages, and adults maintain site fidelity. Spatial size structuring probably also occurs in other seahorse species, for the en- tire population and for adults alone (//. comes, Meeuwig'''; H. guttulatus, Curtis''). A better understanding of the spatial structuring of populations could allow for spatial control of fishing effort to minimize bycatch. We found a consistent, female-biased sex ratio in the catch across the two years of our study, with only 42% males. This bias may reflect the sex ratio of the H. erectus population: a similar sex ratio (40% males) was found in a population of//, erectus in Chesapeake Bay, Virginia (Teix- eira and Musick, 2000). Female-biased sex ratios have also been found in H. zosterae (33% males) when sampled by pushnet (Strawn, 1958), and in H. abdominalis studied underwater in Australia (Martin-Smith^). Many other wild populations of seahorses studied underwater, however, have documented equal numbers of males and females (//. breviceps: H. comes: Moreau and Vincent*; Perante et al., 1998; H. reidi: Dauwe, 1993; H. whitei: Vincent and Sadler, 1995). Sexual dimorphism in H. erectus was too slight to explain different catchability of the two sexes and would, in any case, have favored the capture of males. The dispro- portionate catch of females could have arisen from spatial segregation by sex; the greater catches of reproductively active males in shallower areas suggests that males may spend most of their time inshore of the trawled area. We also cannot discount the possibility that some seahorses classified as females may have been immature males, and the sex ratio in the population could in fact be 1:1. The proportion of reproductively active seahorses in the bycatch was lower than expected, particularly in 1999. Our study occurred during summer, within the breeding season for the congeneric and sympatric //. zosterae in Florida (Feb- ruary to October; Strawn, 1958), and for H. erectus in Ches- apeake Bay (May to October; Teixeira and Musick, 2000; Vincent, personal obs.). Males of all studied seahorse species were reproductively active almost continuously throughout the breeding season (Dauwe, 1993; Nijhoff, 1993; Vincent and Sadler, 1995; Perante et al., 2002), often remating the same day that they release their young (Vincent and Sadler, 1995). In our study, trawling may have occurred outside the primary breeding areas for male H. erectus, but catches of re- productively nonactive adult males during the breeding sea- son also suggest that repeated trawling may have disrupted breeding in the population. A further indication of possible spatial structuring in the population (by reproductive sta- tus and size) is that almost half of the reproductively active males caught in 1998 were found in S2, the shallowest area; this area was not sampled in 1999 when few reproductively active males were found. Such spatial structuring offers the possibility of trawling outside the breeding area. ^ Vincent, A. C. J., J. J, Meeuwig, M. G. Pajaro, and N. C. I'eranto. Seahorse catches in the central Philippines: characteristics and conservation implications. Manuscript in prep. " Meeuwig, J. J., D. H. Hoang, T S. Ky, S.D. Job, and A. C. J. Vincent. Bycatch landings of seahorses in central Vietnam. Manuscript in prep. •'' Meeuwig, J. J. Life history parameters of the exploited seahorse Hippocnmpus comes: a length based analysis. Manuscript in prep. ^ Curtis, J. 2002. Unpubl. data. Project Seahorse. Fisheries Center, The University of British Columbia, 2204 Main Mall, Vancouver, BC, V6T 1Z4, Canada. " Martin-Smith, K. 2002. L'npubl. data. Project Seahorse, Fisheries Center. The University of British Columbia, 2204 Main Mall, Vancouver, BC, V6T 1Z4, Canada. 8 Moreau, M-A., and A. C. J. Vincent. 2000. Unpubl. data. Project Seahorse, Fisheries Center, The University of British Columbia, 2204 Main Mall, Vancouver, BC,V6T 1Z4, Canada. Baum et a\ : Bycatch of Hippocampus erectus in a Gulf of Mexico sfirimp trawl fisfiery 729 Indirect effects of the trawl fishery on mortality Direct immediate mortality from trawling and culling was rather low, probably in part because trawl sets were of very short duration in this live-bait fishery More importantly, most H. erectus caught in the Hernando Beach fishery were returned to sea, rather than retained as in some other Flor- ida trawl fisheries (Vincent, pers. obs.). Indirect impacts of the fishery may, however, be considerable. Seahorses caught in trawls may experience high postre- lease mortality. A study in the live-bait shrimp trawl fishery in Tampa Bay Florida (Meyer et al, 1999), found that only one of four of the congeneric dwarf seahorse (H. zosterae) caught as bycatch were alive in the holding tank of seawater 36 hours after collection (Meyer^). Such tows lasted only 5 minutes (Meyer et al., 1999); therefore trawl-induced mor- tality could be greater in the Hernando Beach fishery (with trawls of 30-50 minutes), although H. erectus are larger and perhaps more robust than H. zosterae. Like other discarded bycatch, seahorses may also be subject to intense predation upon release. Predation on discarded fish has been observed on prawn trawlers in Australia (Hill and Wassenberg, 1990) and within the bait shrimp fishery of Tampa Bay ( Meyer et al., 1999). Captains of bait-shrimp boats concurred that this is commonplace in the Hernando Beach fishery, and we fre- quently observed bottlenosed dolphins (Tursiops truncatus) and schools of ladyfish (Elops saurus) swimming alongside the boats, feeding on discarded bycatch. Trawling may significantly disrupt seahorse populations, particularly if they are spatially structured, as the present study suggests. The disproportionate removal of females could reduce mating opportunities, especially if//, erectus is monogamous, as are most studied seahorse species (e.g. Vincent, 1995; Vincent and Sadler, 1995; Kvarnemo et al., 2000; Perante et al., 2002). Trawling, on account of re- peated intrusion onto breeding grounds, could also disrupt courtship and negatively affect reproduction, In heavily exploited areas of the fishery where fishermen repeatedly trawl productive areas, seahorses may face cumulative stress. For example, tail injuries are likely a serious wound for seahorses, given that their tails are essential to grasp holdfasts and may play a key role in mating competition, as they do with Hippocampus fuscus (Vincent, 1994). Benthic habitat degradation is another potential indirect effect of live-bait shrimp trawling on seahorses. Bottom-fish- ing gear can reduce habitat complexity by removing emer- gent epifauna, smoothing sedimentary bedforms and by re- moving structure-forming species such as corals and sponges (Hutchings, 1990; Auster et al., 1996; Auster and Langton, 1999; Thrush and Dayton, 2002), Roller beam trawls also affect habitat complexity by redistributing macroalgae and seagrass (Meyer et al., 1999). We estimated that seagrasses comprised between 50% and 80% of the volume of the catch for each trawling operation. Although roller beam trawls are assumed to have low impact on seagrass habitat (Tabb and Kenny, 1969; Meyer et al., 1999), the effects of long-term re- petitive trawling have not been tested, and it is possible that species composition and abundance, including that ofH. erec- tus, have been adversely affected (Watling and Norse, 1998). Summary Despite the relatively low direct mortality of seahorse per boat, the live-bait trawl fishery has the potential to affect seahorse populations, both directly and indirectly. The key question is whether the level of exploitation, and subse- quent impacts, represents a conservation concern. Our evidence is inconclusive. Perhaps only the skewed sex ratio and low proportion of reproductively active males suggest a potential problem. However, fishermen have consistently reported that seahorse catch per boat has declined greatly over the past two decades in this area. Effects of trawling are also almost certainly greater in food shrimp trawl fish- eries, which trawl with much larger gear for longer peri- ods, and obtain substantially more bycatch, with higher mortality. Our analysis should thus be seen as a first step in identifying areas for which more information is needed, specifically estimating abundance and fishing mortality, and understanding spatial structuring in H. erectus. This paper focuses attention on the need for research on and monitoring of small fishes that may be affected by non- selective fishing gear Management responses to minimize bycatch have focused primarily on seabirds, sea turtles, and commercially important finfishes, but trawl fisheries may also have significant impacts on the many small ma- rine organisms obtained as bycatch, even if they comprise only a small proportion of the bycatch. Bycatch excluder devices are unlikely to be effective in reducing incidental catches of these species. Temporal variation in CPUE and spatial population structuring, as observed in our present study for H. erectus, suggest that time-area closures may be a pragmatic solution for reducing incidental catch. Acknowledgments This paper is a contribution from Project Seahorse. We thank Jana Schulz for her assistance with fieldwork, Daniel and Patricia Mohr for their support during fieldwork, James Boxall for preparing the map, A. DeBruyn, L. Crowder M. Kaiser, and an anonymous reviewer for providing helpful comments on an earlier draft of this manuscript. This study would not have been possible without the cooperation and support of many of the shrimp boat captains and crew in Hernando Beach. This research was funded through an NSERC summer undergraduate award to JKB, support from the Community Fund (UK) and Guylian Chocolates Belgium for JJM, and an NSERC operating grant to ACJV. Literature cited Meyer, D. 1999. Personal commun. NOAA Center for Coastal Fisheries and Habitat Research, Beaufort Laboratory, 101 Fivers Island Road, Beaufort, NC 28516. Adams, C, S. Larkin, and D. Lee. 2001. Volume and value of marine ornamentals collected in Florida, 1990-98. Aquar Sci. Conserv. 3(l-3):25-36. 730 Fishery Bulletin 101(4) Alverson, D. L. 1997. Global assessment of fisheries bycatch and discards: a summary overview. In Global trends: fisheries manage- ment {E. K. Pikitch, D. D. Huppert, and M. P. Sissenwine, eds.), 115-125 p. Am. Fish. Soc. Symp., Vol. 20. Alverson, D. L., M. H. Freeberg, S. A. Murawski, and J. G. Pope. 1994. A global assessment of fisheries bycatch and discards, 233 p. FAOFish. Biol. Tech. Pap. 339. FAO, Rome. Auster, P. J., and R. W. Langton. 1999. The effects of fishing on fish habitat. In Fish habitat: essential fish habitat (EFH) and rehabilitation (L. Benaka, ed. ), p. 150-187. Am. Fish. Soc. Rep. 22. Auster, P. J., R. J. Malatesta, R. W. Langton, L. Watling, P. C. Valentine, C. L. S. Donaldson, E. W. Langton, A. N. Shepard, and I. G. Babb. 1996. The impacts of mobile fishing gear on seafloor habi- tats in the Gulf of Maine (Northwest Atlantic): implica- tions for conservation of fish populations. Rev. Fish. Sci. 4:185-202. Berkeley, S. A., D. W. Pybas, and W. L. Campos. 1985. Bait shrimp fishery of Biscayne Bay, 16 p. Fla. Sea Grant Ext. Prog. Tech. Pap. 40. Florida Sea Grant, Gaines- ville, FL. Chan, E. H., and H. C. Liew. 1986. Characteristics of an exploited tropical shallow-water demersal fish community in Malaysia. In Proceedings of the first Asian fisheries forum (J. L. Maclean, L. B. Dizon, and L. V. Hosillows, eds.), p. 349-352. Asian Fisheries Society, Manila, Philippines. Dauwe, B. 1993. Ecology of the seahorse Hippocampus reidi on the Bonaire coral reef (N.A.): habitat, reproduction and com- munity interaction. M.S. thesis, 65 p. Rijksuniversiteit, Groningen, Netherlands. deBruyn, A. M. H., and J. J. Meeuwig. 2001. Detecting lunar cycles in marine ecology: periodic regression versus categorical ANOVA. Mar. Ecol. Prog. Ser. 214:307-310. Diamond, S. L., L. G. Cowell, and L. B. Crowder 2000. Population effects of shrimp trawl bycatch on Atlantic croaker Can. J. Fish. Aquat. Sci. 57(10):2010-2021. FAO (Food and Agriculture Organization of the United Nations). 1999. The state of world fisheries and aquaculture 1998, 112 p. FAO, Rome. Galloway, B. J., and J. G. Cole. 1999. Reduction of juvenile red snapper bycatch in the Gulf of Mexico shrimp trawl fishery. N. Am. J. Fish. Manag. 19(2):342-355. Graham, G. L. 1995. Finfish bycatch from the Southeastern shrimp fishery. In Solving bycatch: considerations for today and tomorrow, p. 115-119. Univ. Alaska Sea Grant College Program Report 96-03, Fairbanks, AK. Hill, B. J., and T. J. Wassenberg. 1990. Fate of discards from prawn trawlers in Torres Strait. Aust. J. Mar Freshw. Res. 41:53-64. 2000. The probable fate of discards from prawn trawlers fish- ing near coral reefs. A study in the northern Great Barrier Reef Australia. Fish. Res. 48(3):277-286. Hutchings, P. 1990. Review of the effects of trawling on macrobenthic epifaunal communities. Aust. J. Mar Freshw. Res. 41: 111-120. lUCN (International Union for Conservation of Nature and Natural Resources). 1996 World Conservation Congress, 1"' session. Resolutions and recommendations. 1.16 Fisheries by-catch; Montreal, Canada, 14-23 October 1996. lUCN, Gland, Switzerland. 2002. 2002 lUCN red list of threatened species. lUCN, Gland, Switzerland and Cambridge, UK. http://www.redlist. org/ [Accessed 21 January 2003]. Jennings, S., and M. J. Kaiser 1998. The effects of fishing on marine ecosystems. Adv. Mar Biol. 34:201-352. Julian, F, and M. Beeson. 1998. Estimates of marine mammal, turtle and seabird mortality for two California gillnet fisheries: 1990-1995. Fish. Bull. 96:271-284. Kaiser, M. J., and B. E. Spencer 1995. Survival of by-catch from a beam trawl. Mar Ecol. Prog. Ser. 126:31-38. Kvarnemo, C, G. I. Moore, A. G. Jones, W. S. Nelson, and J. C. Avise. 2000. Monogamous pair bonds and mate switching in the Western Australian seahorse Hippocampus subelongatus. J. Evol. Biol. 13:882-888. Larkin, S. L., and R. L. Degner. 2001. The U.S. wholesale market for marine ornamentals. Aquar Sci. Conserv. 3(l):13-24. Lourie, S. A., A. C. J. Vincent, and H. J Hall. 1999. Seahorses: an identification guide to the world's species and their conservation, 214 p. Project Seahorse, London, UK. Mensink, B. P., C. V. Fischer, G. C. Cadee, M. Fonds, C. C. Ten Hallers-Tjabbes, and J. P. Boon. 2000. Shell damage and mortality in the common whelk Buccinum undatum caused by beam trawl fishery. J. Sea Res. 43(1 ):53-64. Meyer, D. L., M. S. Fonseca, P. L. Murphey, R. H. McMichael, Jr. M. M. Byerly, M. W. LaCroix, P E. Witfield, and G. W. Thayer 1999. Effects of live-bait shrimp trawling on seagrass beds and fish bycatch in Tampa Bay Florida. Fish. Bull. 97:193-199. Milton, D. 2001. Assessing the susceptibility to fishing of populations of rare trawl bycatch: sea snakes caught by Australia's north- em prawn fishery. Biol. Conserv. 101:281-290. Nijhoff, M. 1993. Reproductive cycle of the seahorse Hippocampus reidi on the Bonaire coral reef M.S. thesis, 49 p. Rijksuniver- siteit, Groningen, Netherlands. Ortiz, M., C. M. Legault, and N. M. Ehrhardt. 2000. An alternative method for estimating bycatch from the U.S. shrimp trawl fishery in the Gulf of Mexico, 1972-1995. Fish. Bull. 98:583-599. Parrish, J. K. 1999. Using behavior and ecology to exploit schooling fishes. Environ. Biol. Fishes 55:157-181. Perante, N. C, M. G. Pajaro, J. J. Meeuwig, and A. C. J. Vincent. 2002. Biology of Hippocampus comes in the central Philippines. J. Fish Biol. 60(4):821-837. Perante, N. C, A. C. J. Vincent, and M. G. Pajaro. 1998. Demographics of Hippocampus comes seahorses in Bohol, central Philippines. In Proceedings of the third international conference on the Marine Biology of the South China Sea; Hong Kong 1996, p. 439-148. Hong Kong Univ Press, Hong Kong. Pettovello, A. D. 1999. By-catch in the Patagonian red shrimp {Pleoticus muelleri) fishery. Mar. Freshw. Res. 50:123-127. Pikitch, E. K., J. R. Wallace, E. A. Babcock, D. L. Erickson, M. Saclens, and G. Oddsson. 1998. Pacific halibut bycatch in the Washington, Oregon, and Baum et a\ Bycatch of Hippocampus erectus in a Gulf of Mexico shrimp trawl fishery 731 California groundfish and shrimp trawl fisheries. N. Am. J. Fish. Manag. 18:569-586. Polacheck.T. 1989. Harbour porpoises and the gillnet fishery: Incidental takes spur population studies. Oceanus 32:63-70. Probert, P. K., D. G. McKnight, and S. L. Grove. 1997. Benthic invertebrate bycatch from a deep-water trawl fishery, Chatham Rise, New Zealand. Aquatic conserva- tion: marine and freshwater ecosystems 7(l):27-40. Strawn, K. 1958. Life history of the pygmy seahorse, Hippocampus zos- ierae Jordan and Gilbert, at Cedar Key, Florida. Copeia 1958:16-22. Tabb, D. C, and N. Kenny 1969. A brief history of Florida's live bait shrimp fishery with description of fishing gear and methods. FAO Fish. Rep. 57:1119-1134. Teixeira, R. L., and J. A. Musick. 2000. Reproduction and food habits of the lined seahorse. Hippocampus erectus (Teleeostei: Syngnathidae) of Chesa- peake Bay, Virginia. Rev. Bras. Biol. 61(l):79-90. Thrush, S., and P K. Dayton. 2002. Disturbance to marine benthic habitats by trawling and dredging: implications for marine benthic biodiversity. Ann. Rev Ecol. Syst. 33:449-473. Vari, R. P. 1982. Fishes of the western North Atlantic, subfamily Hippo- campinae. The seahorses, p. 173-189. Sears Foundation for Marine Research Memoir, vol. 1, Yale Univ., New Haven, CT. Vincent, A. C. J. 1994. Operational sex ratios in seahorses. Behaviour 128: 153-167. 1995. A role for daily greetings in maintaining seahorse pair bonds. Anim. Behav. 49:258-260. 1996. The international trade in seahorses, vii + 163 p. TRAFFIC International, Cambridge, UK. Vincent, A. C. J., and L. M. Sadler 1995. Faithful pair bonds in wild seahorses, Hippocampus whitei. Anim. Behav. 50:1557-1569. Watling, L., and E. A. Norse. 1998. Disturbance of the seabed by mobile fishing gear: a comparison to forest clear cutting. Conserv. Biol. 12(6): 1180-1197. Weimerskirch, H., N. Brothers, and P. Jouventin. 1997. Population dynamics of wandering albatross Diomedea exulans and Amsterdam albatross D. amsterdamensis in the Indian Ocean and their relationships with long-line fisher- ies; conservation implications. Biol. Conserv. 79:257-270. Zar, J. H. 1996. Biostatistical analysis, 3"^ ed., 662 p. Prentice Hall, Upper Saddle River, NJ. 732 Abstract— Estimates of instantaneous mortality rates (Z) and annual appar- ent survival probabilities (.0) were generated from catch-curve analyses for oceanic-stage juvenile loggerheads (Caretta caretta) in the waters of the Azores. Two age distributions were analyzed: the "total sample" of 1600 log- gerheads primarily captured by sight- ing and dipnetting from a variety of vessels in the Azores between 1984 and 1995 and the "tuna sample" of 733 log- gerheads (a subset of the total sample) captured by sighting and dipnetting from vessels in the commercial tuna fleet in the Azores between 1990 and 1992. Because loggerhead sea turtles begin to emigrate from oceanic to neritic habitats at age 7, the best estimates of instantaneous mortality rate (0.094) and annual survival probability (0.911) not confounded with permanent emi- gration were generated for age classes 2 through 6. These estimates must be interpreted with caution because of the assumptions upon which catch-curve analyses are based. However, these are the first directly derived estimates of mortality and survival probabilities for oceanic-stage sea turtles. Estimation of survival probabilities was identified as "an immediate and critical require- ment" in 2000 by the Turtle Expert Working Group of the U.S. National Marine Fisheries Service. Estimates of survival probabilities for oceanic-stage loggerhead sea turtles {Caretta caretta) in the North Atlantic Karen A. Bjorndal Alan B. Bolten Archie Carr Center for Sea Turtle Research and Department of Zoology University of Florida P.O Box 118525 Gainesville, Florida 32611 E-mail address (for K, A. B|orndal): kab@zoology ufl edu Helen R. Martins Deparlamento de Oceanografia e Pescas Universidade dos Azores PT-9901-862 Horta Acores, Portugal Manuscript approved for publication 21 April 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:732-736(2003). A major gap in our understanding of sea turtle demography is the level of mortality — both natural and human- induced — experienced by wild popula- tions. Lack of directly derived estimates of mortality (or survival) probabilities for the juvenile oceanic-stage in sea turtle populations is a critical source of uncertainty in development of popula- tion models and evaluation of man- agement plans. In current population models, survival estimates for juvenile, oceanic-stage sea turtles are fitted parameters, not directly derived esti- mates, of survival (Chaloupka, in press; Heppell et al., in press). The Turtle Expert Working Group ( 2000) identified the estimation of survival probabilities as "an immediate and critical require- ment." Population models indicate that survival probability of juvenile oceanic- stage loggerhead sea turtles (Caretta caretta) has a substantial effect on overall population growth (Chaloupka, in press; Heppell et al., in press). Catch-curve analyses have been used for many years to estimate survival probabilities for species harvested in commercial fisheries and, less fre- quently, for other species (Seber, 1982). Estimates of survival probabilities have been generated from catch-curve analyses for subadult neritic-stage populations of loggerhead sea turtles (Frazer, 1987; Epperly et al., 2001) and Kemp's ridley sea turtles (Lepidochelys kempii) (Turtle Expert Working Group, 2000) based on stranding data. Catch- curve analyses confound mortality and permanent emigration, and thus gen- erate estimates of apparent survival probability (CP), 0 = S (1 - emigration ), where S = true survival probability; and emigration = the probability of perma- nent emigration. We estimated survival probabilities (both and S) for juvenile oceanic- stage loggerhead sea turtles in the wa- ters around the Azores, a developmental habitat for the population of loggerhead sea turtles that nest on beaches in the southeastern United States (Bolten et al., 1998). We applied catch-curve analyses to two age distributions of loggerhead sea turtles. Methods Two size distributions were compiled for this study The first ("total sample") comprised 1600 oceanic-stage logger- head sea turtles that were captured from 1984 through 1995 in the waters of the Azores. Except for a few of the smallest of these sea turtles found as stranded carcasses, they were collected in dipnets after being sighted at the surface of the ocean from the decks of a B|orndal et al.: Estimates of survival for Caretta caretta In the North Atlantic 733 variety of vessels. Loggerhead sea turtles captured on longline hooks were excluded from this sample to meet the requirement of equal probability of cap- ture across the age range ( see "Results" section ). The turtles were measured, tagged, and released soon after capture. The juvenile loggerhead sea turtles ranged in size from 8.5 to 82.0 cm curved carapace length (mean=33.1 cm, SD=11.6) measured from the anterior point at midline to the posterior notch at midhne between the supracaudals (Bolten, 1999). For 248 turtles, straight-line carapace measure- ments were converted to curved carapace length, as described in Bjorndal et al. (2000). The second age distribution ("tuna sample") was a subset of the total sample and comprised 733 log- gerhead sea turtles captured by crews of commercial tuna vessels in the Azores between 1990 and 1992. We analyzed the tuna sample in addition to the to- tal sample because the tuna sample was collected over a shorter interval ( 1990-92) than was the total sample (1984-95). In addition, sizes of vessels from which turtles were captured were more consistent for the tuna sample. This collaborative project with the tuna fleet is described in Bolten et al. (1993); sea turtles are not bycatch in the tuna fishery. Sea turtles were sighted at the surface while the crews were scanning for indications of tuna feeding ac- tivity. The turtles were then captured in dipnets, tagged and measured by a crew member, and re- leased at sea. The juvenile loggerhead sea turtles ranged from 11.0 to 82.0 cm curved carapace length ( mean=33.5, SD= 1 1.2 ). No conversion from straight to curved measurements was required. The size distributions were converted to age dis- tributions by using a size-at-age function developed for this population based on a skeletochronological study (Table 1; Bjorndal et al., 2003). Catch curves were generated for each age distribution by plot- ting the natural log of A^^ against x, where N^ is the number of turtles of age x. The catch curves were truncated by excluding age classes with fewer than five individuals, as recommended by Seber (1982). The age at which the population fully recruited to the capture method (threshold age) was identified as the age with the highest ln(A'',). Linear regression analy- ses of the values on the right-hand or declining slope of the distribution were used to generate estimates of total instantaneous mortality rate (Z), which is expressed on an annual basis and is the absolute value of the slope of the regression line. Annual apparent survival probability (46 cm — the size at which they begin to leave oceanic habitats and recruit to neritic habitats (Bjorndal et al., 2000). Thus, the sharper decline beginning with age 7 reflects the migration of turtles out of the sampling area. Estimates of total instantaneous mortality rate (Z) and annual apparent survival probability (4>) were generated for three age ranges for each sample: from threshold age to final age, from threshold age to pre-emigration age (6 years), and from pre-emigration age to final age. For the total sample, the age ranges were 2 to 12 years, 2 to 6 years, and 6 to 12 years, respectively; for the tuna sample, the ages were 4 to 11, 4 to 6, and 6 to 11 years (Fig. 2, Table 2). Discussion Estimates of mortality and survival generated from catch curves should be interpreted with caution for several reasons. First, the analysis assumes a stable age distribution, which we cannot confirm for North Atlantic loggerhead sea turtles. Second, the analysis assumes that mor- tality rates were consistent over the years of the study. The similarity of the mortality and survival estimates between the total sample ( 1984-95) and the tuna sample ( 1990-92) suggests that the esti- mates for the total sample are not greatly affected by heterogeneity among years. Third, converting size distributions to age distributions based on a size-at-age function introduces some level of error. We believe the error from our size-at-age function is small, as discussed in Bjorndal et al. (in press). Fourth, the analysis assumes no size or age effect on mor- tality rates. The catch curves for both the total sample and tuna sample reveal a size or age effect with a pivot point at age 6. This size or age effect reflects the beginning of emi- gration out of our study area. Loggerhead sea turtles begin to leave oceanic habitats around the Azores and recruit to neritic habitats at 7 years of age, at -46 cm curved carapace length (Bjorndal et al., 2000, 2003). This change in slope demonstrates the fifth difficulty in interpreting catch- curve estimates: permanent emigration and mortality are confounded in the estimates. That is, declines in numbers with age, whether they are due to emigration or mortality, are included in the estimate of mortality. The confounding of emigration and mortality can introduce a major error in estimates of mortality in populations — such as sea turtle populations in oceanic and neritic habitats — that undergo developmental migrations. — I 1 1 1 1 1 1 1 1 1 1 r- 01 23456789 10 11 12 - 6n 5 4 3 - 2 - 1 - 0 ^ B • •---^--_ 0 1 5 6 7 Age (years) 10 — r— 11 Figure 2 Catch curves for (A) total sample and (B) tuna sample. Dashed lines are linear regressions for entire sample, threshold age to age 6, and age 6 to final age (see Table 2 for regression statistics). Because little permanent emigration apparently occurs before the age of 7, the survival estimates for the ages prior to 7 years are our best estimates of true survival (S). As can be seen in Table 2, the estimates of total instanta- neous mortality and annual survival are similar for the two samples. We believe that the estimate of S (0.911 and 0.894, respectively, for the total sample and tuna sample) would apply to the entire life stage over the size range from 20 to 65 cm CCL for most sources of mortality other than fisheries biased to large sizes, such as longline fisheries. For predation, as sea turtles increase in size, they outgrow the prey size of some fish predators, but they also grow into the prey size of the largest predators, such as killer whales and humans (although the latter source is now very low in the Azores as a result of legislation and education [senior author, personal obs.]). Death from ingestion of or entangle- ment in marine debris would probably not vary substan- tially over this size range. However, mortality from inciden- Bjorndal et al.: Estimates of survival for Caretta caretta in the North Atlantic 735 Table 1 Size ranges of age classes and age distributions for tota of turtles in each age class x, and YOY is young of year. sample and tuna sam pie. CCL is curved carapace length N^ is the number Size range (cm CCL) Age Total sample Tuna sample K ln(N, 1 iV, \n{N^) < 15,0 YOY 9 2.197 — — 15.0-20.5 1 101 4.615 49 3.892 20.5-26.1 2 287 5.659 118 4.771 26.1-317 3 248 5.513 111 4.710 31.7-36.9 4 248 5.513 120 4.787 36.9-41.7 5 224 5.412 108 4.682 41.7-46.5 6 189 5.242 96 4.564 46.5-49.9 7 127 4.844 58 4.060 49.9-52.3 8 56 4.025 27 3.296 52.3-55.0 9 37 3.611 17 2.833 55.0-58.2 10 33 3.497 16 2.773 58.2-61.6 11 19 2.944 7 1.946 61.6-65.0 12 13 2.565 — tal capture in longline fisheries in the Azores does increase with size, with the 2 to 6 year age classes experiencing very little mortahty (Bolten,in press). Thus, if our estimate ofS (calculated for the age classes between threshold age and age 6) were applied to the entire oceanic stage, the effect of mortality in longline fisheries, or other fisheries biased to large size classes, would be underestimated. The estimate of 0.911 for annual survival probabilities for oceanic-stage loggerhead sea turtles in the waters of the Azores indicates high survival in this lifestage without mortality from longline fisheries. Species characterized by long life and late sexual maturity, such as loggerhead sea turtles, require very high survival throughout immature stages to maintain populations (Congdon et al., 1993; Grouse, 1999). This high probability of survival is also consistent with the theory that lower predation in oceanic habitats compared to neritic habitats is the selective pres- sure that maintains oceanic juvenile stages in most species of sea turtles (Bolten, 2003). In two updated matrix models for North Atlantic log- gerhead sea turtles (that differed in stage lengths). Hep- pell et al. (in press) derived fitted estimates of 0.745 and 0.875 for annual survival probabilities of the oceanic stage, which they defined as spanning 5 to 45 cm carapace length. Chaloupka (in press) derived an estimate of annual sur- vival probability for oceanic-stage loggerhead sea turtles in Australia of 0.67 sampled from a logistic probability density function that ranged from 0.60 to 0.76 and had a mode at 0.67. The tuned estimate of 0.67 was derived from a stochastic simulation model that incorporated em- pirically based survival probability estimates for all age classes in the model except the oceanic phase (Chaloupka and Limpus, 2002; Chaloupka, in press). The estimate of 0.67 was generated for a size range from posthatchlings that have left the waters directly adjacent to the nesting Table 2 Estimates of instantaneous mortality rates (Z) and annual apparent survival probabilities {0, estimated as e"^) for oceanic-stage loggerheads in the waters of the Azores generated from catch-curve analyses, r^ (coefficient of determination) and P values are from linear regression analyses (see Fig. 2). Age range (years Z 0 r2 P Total sample 2 to 12 0.333 0.720 0.935 < 0.001 2 to 6 0.094 0.911 0.923 = 0.009 6 to 12 0.441 0.643 0.974 < 0.001 Tuna sample 4 to 11 0.421 0.656 0.954 < 0.001 4 to 6 0.112 0.894 0.999 = 0.021 6 to 11 0.498 0.608 0.966 < 0.001 beach to subadults that begin to leave the oceanic habitats at a size of 69 cm curved carapace length (Chaloupka and Limpus, 2002). The fitted estimates for annual survival from the Heppell et al. (in press) models and the Chaloupka (in press) model are lower than the estimates in our study, but the size ranges are different. In the Heppell et al. (in press) models and the Chaloupka (in press) model, the oceanic stage includes the posthatchling phase during which loggerhead sea turtles migrate from nesting beaches to their oceanic habitats. We could not include this early posthatchling phase in our esti- mates of survival of oceanic-stage loggerhead sea turtles in the waters of the Azores because many turtles in this phase 736 Fishery Bulletin 101(4) have not reached the Azores and they are younger than our threshold ages. We beheve that mortality in this early transitional stage when loggerhead sea turtles first cross the Atlantic may be high. In addition to high rates of pre- dation. winds and currents can overwhelm the swimming and orientation abilities of the posthatchling sea turtles, transporting the turtles to habitats, such as waters off the British Isles, that cannot sustain them (Carr, 1986; Hays and Marsh, 1997). Generating directly derived estimates of survival probabilities of loggerhead sea turtles younger than 2 years of age should be a high priority. Acknowledgments This study would not have been possible without the sup- port of our colleagues in the Azores: "Equipa Tartaruga" at the Department of Oceanography and Fisheries (DOP), University of the Azores; the captains and crews of the commercial tuna fleet based in Horta and Pico; and J. and G. Franck of the MY Shanghai. We thank M. Chaloupka for encouragement to pursue catch-curve analysis. We thank M. Chaloupka and J. Seminoff for comments on earlier drafts of the manuscript and P. Eliazar for technical assistance. This project was funded by the U.S. National Marine Fisheries Service and the Disney Wildlife Con- servation Fund. All work was conducted in compliance with the Institutional Animal Care and Use Committee, University of Florida. Literature cited Bjorndal, K. A., A. B. Bolten, T. Dellinger, C. Delgado, and H. R. Martins. 2003. Compensatory growth in oceanic loggerhead sea turtles: response to a stochastic environment. Ecology 84:1237-1249. Bjorndal, K. A., A. B. Bolten, and H. R. Martins. 2000. Somatic growth model of juvenile loggerhead sea turtles Caretta caretta: duration of pelagic stage. Mar. Ecol. Prog. Ser 202:265-272. Bolten, A. B. 1999. Techniques for measuring sea turtles. In Research and management techniques for the conservation of sea turtles (K. L. Eckert, K. A. Bjorndal, F. A. Abreu-Grobois, and M. Donnelly, eds.), p. 110-114. lUCN/SSC Marine Turtle Specialist Group Publication 4, Washington, DC. 2003. Variation in sea turtle life history patterns: neritic versus oceanic developmental stages. In Biolog>' of sea turtles, vol. 2 (P. L. Lutz, J. A. Musick, and J. Wyneken, eds.), p. 243-257. CRC Press, Boca Raton, PL. In press. Active swimmers — passive drifters: the oceanic juvenile stage of loggerheads in the Atlantic system. In Loggerhead sea turtles (A. B. Bolten and B. E. Witherington, eds.). Smithsonian Institution Press, Washington, DC. Bolten, A. B., K. A. Bjorndal, H. R. Martins, T Dellinger, M. J. Biscoito, S. E Encalada. and B. W. Bowen. 1998. Transatlantic developmental migrations of logger- head sea turtles demonstrated by mtDNA sequence analysis. Ecol. Appl. 8:1-7. Bolten, A. B., H. R. Martins, K. A. Bjorndal, and J. Gordon. 1993. Size distribution of pelagic-stage loggerhead sea turtles (.Caretta caretta) in the waters around the Azores and Madeira. Arquipelago llA:49-54. Carr, A. 1986. Rips, FADS, and little loggerheads. BioScience 36: 92-100. Chaloupka, M. In press. Simulation modeling of population viability for log- gerhead sea turtles exposed to competing mortality risks in the western south Pacific region. In Loggerhead sea turtles (A. B. Bolten and B. E Witherington, eds.). Smith- sonian Institution Press, Washington, DC. Chaloupka, M. Y., and C. J. Limpus. 2002. Survival probability estimates for the endangered log- gerhead sea turtle resident in southern Great Barrier Reef waters. Mar Biol. 140:267-277. Congdon, J. D., A. E. Dunham, and R. C. van Loben Sels. 1993. Delayed sexual maturity and demographics of Blanding's turtle (Emydoidea blandingii): implications for conservation and management of long-lived organisms. Conserv. Biol. 7:826-833. Grouse, D. T 1999. The consequences of delayed maturity in a human- dominated world. Am. Fish. Soc. Symp. 23:195-202. Epperly S. P., M. L. Snover, J. Braun-McNeill, W. N. Witzell, C. A. Brown, L. A. Csuzdi, W. G. Teas, L. B. Crowder, and R. A. Myers. 2001. Stock assessment of loggerhead sea turtles of the western North Atlantic. In Stock assessments of logger- head and leatherback sea turtles and an assessment of the impact of the pelagic longline fishery on the loggerhead and leatherback sea turtles of the western North Atlantic, p. 3-66. NCAA Tech. Memo. NMFS-SEFSC-455. Frazer, N. B. 1987. Preliminary estimates of survivorship for wild juve- nile loggerhead sea turtles (Caretta caretta). J. Herpetol. 21:232-235. Hays, G. C, and R. Marsh. 1997. Estimating the age of juvenile loggerhead sea turtles in the North Atlantic. Can. J. Zool. 75:40^6. Heppell, S. S., L. B. Crowder, D. T Grouse, S. P Epperly and N. B. Frazer In press. Population models for Atlantic loggerheads: past, present and future. In Loggerhead sea turtles (A. B. Bolten and B. E. Witherington, eds.). Smithsonian Insti- tution Press, Washington, DC. Isaac, V. J. 1990. The accuracy of some length-based methods for fish population studies, 81 p. International Center for Living Aquatic Resources Management, Manila, Philippines. Seber, G. A. F 1982. The estimation of animal abundance and related parameters, 654 p. Macmillan Publishing Co., New York, NY. Turtle Expert Working Group. 2000. Assessment update for the Kemp's ridley and logger- head sea turtle populations in the western North Atlantic. U.S. Dep. Commer, NOAATech. Memo. NMFS-SEFSC-444, 115 p. 737 Abstract— The green sea urchin iStron- gyloccntrotus droebachiensis) is impor- tant to the economy of Maine. It is the state's fourth largest fishery by value. The fishery has experienced a con- tinuous decline in landings since 1992 because of decreasing stock abundance. Because determining the age of sea urchins is often difficult, a formal stock assessment demands the development of a size-structured population dynamic model. One of the most important com- ponents in a size-structured model is a growth-transition matrix. We developed an approach for estimating the growth- transition matrix using von Bertalanffy growth parameters estimated in previ- ous studies of the green sea urchin off Maine. This approach explicitly consid- ers size-specific variations associated with yearly growth increments for these urchins. The proposed growth- transition matrix can be updated read- ily with new information on growth, which is important because changes in stock abundance and the ecosystem will likely result in changes in sea urchin key life history parameters including growth. This growth-transition matrix can be readily incorporated into the size-structured stock assessment model that has been developed for assessing the green sea urchin stock off Maine. Developing a growth-transition matrix for the stock assessment of the green sea urchin (Strongylocentrotus droebachiensis) off Maine Yong Chen School of Marine Sciences 218 Libby Hall University of Maine Orono, Maine 04469 E-mail address ychen@maine.edu Margaret Hunter Maine Department of Marine Resources P.O. Box 8 West Boothbay Harbor, Maine 04575 Robert Vadas Department of Biological Sciences University of Maine Orono, Maine 04469 Brian Beal Division of Environmental and Biological Sciences University of Maine Machias, Maine 04654 Manuscript approved for publication 17 April 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish Bull. 101:737-744 (2003). The green sea urchin (Strongylocentro- tus droebachiensis) fishery is the state's fourth largest fishery by value, worth $20.3 million to harvesters in 1999. The fishery is managed by limited entry, a limited number of opportunity dates that are established each year by recommendation of the sea urchin zone council (SUZC), and minimum and maximum size limits. The fishery is further regulated seasonally by two zones that correspond to variation in spawning time along the coast (Vadas etal., 1997). The Maine sea urchin fishery began in the late 1980s and reached its peak in landings in 1992. It has since experi- enced a continuous decline in landings, mainly resulting from large decreases in sea urchin stock abundance (Fig. 1). Although the large decrease in abun- dance is evident in many studies (Ste- neck and Vadas'; Harris^) and apparent to the sea urchin fishing industry, the catch-per-unit-of-effort (CPUE) data derived from the fishery have shown no significant decreases over the last 10 years (Fig. 1). We need to perform a formal stock assessment to better understand the population dynamics of the sea urchin stock and to develop an optimal management strategy. A population dynamics model for the sea urchin stock should provide reliable estimates of model parameters with suitable statistical methods (Hilborn and Walters, 1992; Chen and Paloheimo, 1998; Walters, 1998). A size-structured population dynamics model is needed for the sea urchin fishery because sea urchins are difficult to age and growth varies widely among individuals (Quinn and Deriso, 1999). One of the key components of a size- structured population dynamics model is a growth-transition matrix, which describes the probability of an organ- 1 Steneck, R., and R. L. Vadas. 2002. Per- sonal commun. School of Marine Sciences, University of Maine, Orono, ME 04469. 2 Harris, L. 2002. Personal commun. Department of Zoology, University of New Hampshire, Durham, NH 03824. 738 Fishery Bulletin 101(4) ism growing from one size class to another size class in a given unit of time (Sullivan et al., 1990; Sullivan, 1992). In practice, two approaches can be used to incorporate a growth-transition matrix into a stock assessment: one is to incorporate the growth-transition matrix and simul- taneously estimate matrix parameters with parameters that describe other biological processes in the fishery (Sullivan et al., 1990), and the other approach is to esti- mate the growth-transition matrix independent of other stock assessment models (Chen et al., 2000). The former considers covariance among different processes by esti- mating all parameters simultaneously, but makes the analysis more complicated. The latter approach reduces the complexity of modeling but does not consider the covariance of growth and other biological processes. Be- cause size-structured models are often complicated and have many parameters to be estimated, the estimation of a growth-transition matrix outside the main modeling process may be preferable (Chen et al., 2000). In either case, the quality of the growth-transition matrix can greatly influence the quality of the stock assessment. It is thus essential to develop a growth-transition matrix for the Maine sea urchin stock that can capture the variations in growth increments among individuals. The information required in estimating a growth- transition matrix includes the mean growth increment in a given unit of time and its associated variation for sea urchins of different sizes. Because growth rates of sea urchins vary with size, growth increments also vary with size, and this variation in growth with size is rarely constant. This has been implicit in the statements of model assumptions in many papers (e.g. Sullivan et al., 1990; Sullivan, 1992, Quinn and Deriso, 1999). However, because the variance for growth increments is difficult to estimate, it is often assumed to be constant for organ- isms of different sizes (Quinn and Deriso, 1999). Such an assumption of constant variation in growth incre- ment is rather unrealistic and may introduce biases in estimating a growth-transition matrix. Thus, for the Maine sea urchin we need to develop an approach that can explicitly consider nonconstant variances for gi"owth increments of sea urchins of different sizes. Growth of the sea urchin along the Maine coast has not been studied extensively and the data are limited. The data we used for this study were from Vadas et al. (2002) who collected size-at-age data on sea urchins in two habitats (barren and kelp) from three areas along the coast of Maine. Methods and materials Previous studies have indicated that many environmental variables might influence the growth of the sea urchin (Meidel and Scheibling, 1998; Russell, 1998). Sea urchins in favorable habitats, feeding on preferred seaweeds, grow faster than those feeding on less favorable algae and mussels, and sea urchins on barren grounds grow slower Even in the same habitat, different rates of growth were identified (Vadas, 1977). Previous studies divided the LU Q. o 1985 1990 1995 2000 2005 80 60 40 20 0 I : . ^ - - - ' 1 1993 1994 1995 1996 1997 1998 1999 2000 2001 180 160 140 120 1993 1994 1995 1996 1997 Year 1998 1999 2000 2001 Figure 1 Observed catch measured in metric tons, effort measured in diver-hours, and catch per unit of effort measured in pounds per diver-hour for the sea urchin fishery in management zone 2 in Maine. Zone 1 has a similar temporal pattern. coast of Maine into three regions, northeast, center, and southwest (Vadas et al., 1997). For each region, sea urchin samples were randomly taken from two habitats, barren and kelp. Size-at-age data were collected in 1997-98 for sea urchins in each habitat and area (Vadas and BeaP). Detailed descriptions about the derivation of size and age '' Vadas, R. L., and B. F. Beal. 1999. Temporal and special vari- ability in the relationships between adult size, maturity and fecundity in green sea urchins: the potential use of a roe-yield standard as a conservation tool. Report to the Maine Depart- ment of Marine Resources, Augusta, Maine 04333. Chen et al : Developing a growth-transition matrix for stock assessments of Strongylocentrotus droebachiensis 739 information, justification for dividing the Maine coast, and selection of the habitats can be found in Vadas et al. ( 1997) and Vadas et al. (2002). Vadas et al. (1997) modeled the size-at-age data using the von Bertalanffy growth function (VBGF) described as L, = L„{\-e -All-I,,) (1) where L, = size at age t; L^ = defined as the mean asymptotic length that the sea urchin may attain; K = the Brody growth parameter; and tg = the hypothetical age of size 0 (Ricker, 1975). For each area and habitat, a VBGF was used to fit the size- at-age data. Three parameters in the VBGF (i.e. L„, K, and tf,) and their standard errors were estimated by using the nonlinear least squares method. These estimates were pre- sented in Vadas and BeaP and Vadas et al. (2002), and were made available to the authors of the present study (Table 1). Clearly there were large differences in the estimates of L^ and K and their associated variations among different areas and habitats (Table 1). The LJs estimated for different areas and habitats ranged from 63.1 (northeast region with barren habitat) to 95.2 mm (southeast region with kelp habitat) (Table 1) and tended to be smaller than some large individuals ob- served in the fishery (about 100 mm, Vadas, 1977; Hunter, unpubl. data). This likely resulted from relatively small sample sizes that covered relatively small areas, in a relatively short period, compared with the fishery catch, which targeted larger-size individuals. The exclusion of individuals in the fishery catch that were larger than the L.,'s estimated in Vadas and Beal'^ and Vadas et al. (2002) from the calculation of the growth-transition matrix may underestimate the variability in sea urchin growth, thus introducing errors in stock assessment. Based on the data collected in the Maine sea urchin fishery (Hunter, unpubl. data) and previous studies (Vadas, 1977), 100 mm was considered a reasonable value for the average asymptotic size (LJ for sea urchins on the coast of Maine. However, more extensive sampling needs to be done in the future to verify this estimate. We might be able to derive an estimate of L^ for the Maine sea urchin stock based on the examination of the data collected from the fishery and other studies (Ricker, 1975; Moreau, 1987; Chen et al., 1992). An estimate of K for the whole Maine urchin stock is, however, more difficult because K is an abstract rate describing how fast organ- isms approach the L_.^ and there are no observations or background information with which to compare estimates (Ricker, 1975; Moreau, 1987). We thus need to develop an approach to estimate K for the Maine sea urchin stock which corresponds to the value we assumed for the L^. Many studies have indicated that estimates of K and L„ tend to be highly and negatively correlated (e.g. Moreau, 1987; Chen and Harvey, 1994). Thus, a fish population or species with a large L^^ tends to have a low ii" value, and vice versa (Gallucci and Quinn, 1979; Chen et al., 1992). This suggests a strong relationship between L^ and /f estimates Table 1 The average asymptotic size (L_) and Broady growth coef- ficient (K) estimated for different areas and habitats along | the coast of Maine in the study done by '/adas et al .(1997, 2002). Coefficient of variation (CV) was calculated by using | Equation 2. Coefficient of Area Habitat Parameter variation (CV) ^„ A' CV(L„) CV(AD Northeast Barren 63.1 0.1404 0.242 1.209 Northeast Kelp 88.. 5 0.1263 0.224 0.543 Center Barren 67.0 0.2315 0.084 0.354 Center Kelp 6.3.4 0.3268 0.065 0.248 Southeast Barren 80.1 0.1776 0.099 0.397 Southeast Kelp 95.2 0.1181 0.128 0.338 (Pauly, 1980; Stergiou, 1993). Such a relationship may be used to estimate K for a given L„ or to estimate L„ for a given K. In this study we developed and used the follow- ing empirical approach to derive K for a given value of L^ and its associated uncertainties in the development of a growth-transition matrix: 1) conduct a regression analysis for K and L„ estimated for different areas and habitats along the coast of Maine (Table 1); 2) calculate coefficients of variation (CV) for each K and L„ (Table 1) as „ standard error for K CV(K) = and CV(L,) = K standard error for L^ (2) and conduct a regression analysis of CV(/0 and CV(L„) estimates of different areas and habitats (data in Table 1); 3) use 100 mm to approximate L^„ and use this L„ to esti- mate K from the regression analysis between K and Lj, and 4) calculate the average CV for LJs of different areas and habitats and then use the average CV(L„) to estimate CW(K) from the CW{K)-CV(LJ regression equation. Because K and L„ were estimated for different areas and habitats and had different precisions, outliers might arise in the regression analyses. To avoid possible bias in- troduced by outliers, we used a reweighted least squares (RLS) method for the regression analyses (Chen et al., 1994). This method involves conducting a robust least median of squares (LMS) analysis to identify outliers (Rousseeuw and Leroy, 1987) and justifying the identified outliers by using background information, followed by a weighted LS analysis where justified outliers are weighted by 0 and other data have a weight of 1 (Chen et al., 1994). In the two regression analyses (i.e. steps 1 and 2), L^ and CViLJ were used as the independent variables and K and CW(K) were used as the dependent variables. The reason for this choice (instead of the other way around) is thatL„ is often estimated more reliably and with much smaller 740 Fishery Bulletin 101(4) variations (Chen at al., 1992; also see Table 1), whereas K is often less reliably estimated (Moreau, 1987). One of the basic assumptions for a regression analysis is that the independent variable is error free. In practice, this assump- tion is often relaxed when the independent variable has a much smaller error than the dependent variable (McArdle, 1988). The violation of the normal distribution assump- tion for the errors in the regression analyses may bias the test for the significance of the regression model and its parameters using common parametric tests {F- or /-tests), but does not necessarily result in biases in the regression analysis (Sen and Srivastava, 1990)." Given K and L_, the growth increment during a unit of time (i.e. year) can be calculated as Ai. =(L„-L„)(l-e"'). (3) where K and L^ are the true values without errors; n indexes size class; and L,, is the middle point of the n"' size class. With Equation 3, we can develop two approaches to estimate the growth-transition matrix. One approach is a Monte Carlo simulation. We can randomly sample H sets ofK and L^ values from their joint distributions ( thus con- sider their covariance) and then use them in Equation 3 to calculate H sets of AL for each size group. We can then derive the probability distribution for AL from these H sets of AL values for each size group. The Monte Carlo simula- tion approach is straightforward but requires extensive calculations, in particular when there are a large number of size groups. It is also inconvenient to update the growth- transition matrix when there are new growth data or large changes in growth due to changes in the environment. The second approach is analytic and not so straightforward, but it is easy to update with new information and is less computationally intensive. It is likely that the growth- transition matrix for the Maine sea urchin fishery will need to be updated because of possible changes in growth caused by changes in the sea urchin population size and its ecosystem. Thus we used the second approach, which is described as follows. Assuming the uncertainties associated with the VBGF parameters L^ and K are AL„ and AK respectively, where, AZ._ e N{0.ai_ ) and AK e N{0.al ), we have L = L+AL ami K = K + AK. (4) where Z,„ and K^ are the estimated parameters. Replacing the true values of L„ and K in Equation 3 with Equation 4 and using the approximation e-^ = 1 + AX for small AX, we have AL„=(L„-L„)(\ -€-'') + [aL_(1 -e *)-(!„- Z.JAA:^"*' -AL„AA:e''l = aZ„ -I- f„, (5) where AL„=(L„-L„){\-e-'') (6) e„=ALJ\-e-^)-{L^-L„)AKe-^ -AL„AKe~^- (7) Thus, the expected (mean) value of AL,, is AL^ and vari- ance of AL^ can be estimated from Equation 7 as Var(AZ,„) = CT^d -e '^ ) +(L„ - L„)-(7j.e" 2Cov( L„. Ar)( 1 - e"'' )(ZL - i,„ )?'*". (8) Items with the order of three and above for AL„ and AK are omitted in deriving Equation 8 from Equation 7. From Equation 8, it is clear that the variance of the growth incre- ment varies among different size classes. From AL,, estimated in Equation 6, an expected average yearly growth increment was calculated for each size class. The variability for the average yearly growth increment was assumed to follow a normal distribution with a mean of AL„and variance ofVar (AL,,) estimated from Equation 8. This distribution was used to determine the vector of prob- abilities of growing from size class k to other size classes. If d,^„, and d„ are the lower and upper ends of size class d, the probability of a sea urchin growing from size class n to size class d can be computed as P..^., = I /( .v|AL„,Var(AL„)(/.v, (9) where x is a random variable having a density probability distribution defined by /(.vj AL„,Var(AL„ )) with its expected value of AL,;, and variance of Var(AL,^) (Quinn and Deriso, 1999). In the present study we assumed that the .v variable was a normal density distribution function with a mean of AL^, defined by Equation 6 and with a variance of Var ( AL,^ ) defined by Equation 8. The probability of a sea urchin grow- ing from one size to another was estimated for all size classes to form the matrix. Negative growth increments were not permitted. The largest size class acts as a plus group; there- fore sea urchins in this group have a probability of 1 of remaining in the group. The model contains 61 size classes, each with 1-mm interval width, ranging from 40 mm in size (midpoint value for size class from 39.5-40.5 mm) to 100 mm. Because no negative growth was allowed, the summation of the probabilities of a sea urchin of size class k growing into all other size classes was smaller than 1 (because the normal distribution is symmetric). This problem was avoided by standardization which involved dividing the probability of an urchin in a given size class n growing into each size class by the summation of the probabilities of growing from a given size n to all the size classes. All calculations were done in MS-Excel©( Microsoft Office 2000, Microsoft Corporation, Redmond, WA). A worksheet for estimating a growth-transi- tion matrix as described above is available upon request. Results The LMS analysis suggested that the logarithmic K and L^ data for the barren habitat in the Southwest area was an outlier in the K and L,, regression analysis (Fig. 2). The estimated K and L^ values for the barren habitat in the Southwest had CVs over 120% and 24%, respectively, much Chen et a\: Developing a growth-transition matrix for stock assessments of Strongylocentrotus droebachiensis 741 Ln(/.. 0.0 - 4 -0.5 -1.0 -1.5 -2.0 -2.5 J 4.2 4.3 4.4 4.5 4.6 ♦ Observed ■ Outlier — Predicted Figure 2 The regression analysis of logarithmic K and /-„ for different locations and habitats of Maine sea urchins. 1,4 1.2 ♦ Observed -m- Outlier ■ 1,0 be 0,8 - — Predicted £ 0,6 - S 0.4 - ^^^^ ^ 0.2 - ♦ 0.0 - 1 1 1 0 0 0.1 0.2 0.3 CV for /..., Figure 3 The regi location. ■ession analysis of CVs for K and L^ for different 5 and habitats of Maine sea urchins. larger than the estimates for other locations and habitats (Table 1). This was the only site where the K estimate was not significantly different from 0 (thus the VBGF was not significant). We thus concluded that this data point was an outlier because of the poor fit of the VBGF, and subse- quently it was given a zero weight in the RLS analysis. The RLS regression equation for K and L^ was estimated by LniK) = 8.653 - 2.3777 LniLJ, P=0"0038, adj. r2=0.94. (10) The standard deviations for the intercept and slope were 1.2605 and 0.28923, respectively. The P value for Equation 10 indicates that the regression model is significant. The adj. r^ is the coefficient of determination adjusted for the sample size, suggesting 94% of the variance in \n{K) could be explain by the model. The LMS analysis of the CVs of parameters K and L_ also suggested that the barren habitat in the southwest area was an outlier because it had an exceptionally large CV for K (Fig. 3). We thus concluded that this data point was an outlier and should be given a weight of zero in the RLS analysis. The RLS regression equation for the CVs of parameters K and L„ was estimated by F 7 h 6 r Q> F 5 ID 4 C sz 3 ^ o 2 CJ) Ti (IJ 1 CJ n II UJ 50 100 Midpoint of size class (mm) 150 Figure 4 The expected annual growth increment for Maine sea urchins of different size classes. CVm = 0.189 + 1.5602 CV (LJ, P=0.034, adj. r2 = 0.76. (11) The standard deviations for the intercept and slope were 0.0561 and 0.42319, respectively. The P value suggested the regression model was significant (P<0.05). The value of r^ suggests 76% of the variance in CV(K) could be explained by the model. The average CV for LJs of different areas and habitats was 15%. The L^ was assumed to have a value of 100 mm in this study as discussed previously. This gave the L^ a standard error estimate of 15.0 mm, making its 95% con- fidence intervals 70 mm to 130 mm. The iiT value was esti- mated to be 0.1006 using Equation 10 and L^ of 100 mm. Using Equation 11 and the CV for L,^ the CV for K was estimated to be 42.3%, which yielded the value of 0.0426 for the standard error for K. The annual expected growth increment decreased quick- ly with sea urchin size (Fig. 4). The largest expected annual increment was 6 mm for the smallest size class (39.5-40.5 mm) included in the study. The variance for annual growth increments calculated by using Equation 8 was large for small sea urchins. It decreased initially with size, reaching the smallest value at the 59 mm size class (58.5-59.5 mm), followed by a progressive increase with size (Fig. 5). The expected annual growth increment for the largest size class included in this study had the highest variance, which was over eight times as high as the smallest variance (Fig. 5). The probability distribution of annual growth increment varied among size classes (Fig. 6), reflecting the differences in variances associated with different size classes. The last size class was a plus class, with the probability of staying m the same size class being 1. Figure 6 clearly indicated that no negative growth was allowed. Discussion Great variation in growth was observed in the Maine sea urchin stock (Vadas et al., 2002). Such a pattern of variation was reflected in estimating the VBGF parameters for dif- 742 Fishery Bulletin 101(4) 60 80 Midpoint of size class (mm) Figure 5 The variances of growth increment estimated for dif- ferent sea urchin size classes by using Equation 8. ferent areas and habitats (Table 1). Large standard errors were estimated for the VBGF parameters for sea urchins of the same area and habitat, and large differences occurred in the estimated VBGF parameters between different areas and habitats (Table 1). The approach developed in the pres- ent study considered observations made in both the fishery and scientific studies and provided a systematic way to incorporate the large variation in growth into the estima- tion of a growth-transition matrix, and subsequently into the sea urchin stock assessment. It should be noted that the algorithm developed for esti- matmg the variance of growth increments is approximate, and violations of the assumptions used in deriving the algorithm may introduce errors in estimating a growth- transition matrix. For example, large errors in estimat- ing K and L„ will introduce errors in Equation 5, which was derived by assuming small errors for the two growth parameters. Nonnormal distribution of AL with its mean defined by Equation 6 and variance defined by Equation 8 will also result in errors in developing a growth-transition matrix. Other factors that may influence the quality of the estimated growth transition matrix include errors in esti- mating CVs for K, L^ estimated from Equations 10 and 11, and omitting high order items in deriving Equation 8. Unlike most studies in which the variance for the annual growth increment was assumed to be the same for all size classes (Quinn and Deriso, 1999), our study explicitly sug- gested that the variance for the annual growth increment changed with size (Fig. 4). The differences in the variance were large between size classes, and changed nonlinearly with size. If a constant variance were used for all size classes, the variance in growth increment would be se- verely underestimated for large and small fish. This could introduce large biases in a stock assessment. Size-dependent variation might better describe the variation in annual growth increment. Fish in small size classes tend to grow fast, but their growth tends to be more susceptible to environmental variation than adult growth, often resulting in large variation among individuals (Sum- merfelt and Hall, 1987). Fish in large size classes (older fish) have to divert some energy to reproduction but tend to have considerable variation in energy allocation strate- gies among individuals. Differences among adults in the ability to grow can also be considerable because of genetics, specific growth patterns during juvenile stages, and differ- ences in energy allocation between growth and matura- tion during younger ages (Nikolskii, 1969). This difference may cause large variations in growth for large and old fish (Summerfelt and Hall, 1987; Chen et al., 1988). Compared with old and young ages, growth rates for medium-size and medium-age fish may be less varied (Nikolskii, 1969). This pattern can be reflected realistically in the estimated varia- tion by using the approach derived in our study. Although the choice of L^ was a bit arbitrary in our study, it reflects observations from both the fishery and scientific studies. The largest sea urchins observed in the different scientific studies tend to be smaller than 100 mm, as in- dicated by the estimated L^ values for different areas and habitats (Fig. 1 ). The inability to observe larger sea urchins in scientific studies may result from relatively small sam- ple sizes, the focus of research (small areas), and the large growth variations even in small spatial scales. The data collected from the fishery were more extensive and covered more areas. This, together with the tendency for taking large individuals in the fishery, may suggest that large individuals are more likely to appear in the fishery, rather than in scientific studies. Thus, it may be reasonable to set the expected value of L^ at 100 mm. Also, this higher value corresponds more closely to the upper growth estimates for green sea urchins from the northeast Pacific ( Vadas, 1977). The CV was assumed to be 15% for L_, resulting in the 95% confidence interval of L^ ranging from 70 mm to 130 mm. This range was believed to be a reasonable estimate for the maximum attainable length for green sea urchins on the coast of Maine (Vadas, 1977). The approach developed in our study can be readily used to incorporate the VBGF parameters estimated from dif- ferent studies. This can be accomplished by rerunning the regression analyses between K and L^ and between CVs for K and L^. As more information about the growth of sea urchins on the coast of Maine becomes available, the growth transition matrix can be easily updated to reflect the variation identified in newer studies. The flexibility and ability to easily update and incorporate new information makes this approach desirable to the Maine sea urchin fishery, which is currently undergoing large changes in its population size and has only limited growth data. The value of 100 mm chosen for L^ was rather arbitrary. However, because we considered the negative correlation between K and L_ in deriving the growth transition ma- trix, a small error in the L^ estimate would not change the growth-transition matrix greatly. In the future, however, we can conduct a systematic sampling of the stock across its geographical range and derive some forms of weighted average size with a composite variance that captures the range of sizes exhibited by the species. Such an approach would provide us with a better estimate of L„. The growth-transition matrix developed in our study summarizes the growth patterns of sea urchins along the coast of Maine. It can be updated whenever new growth data become available. It can be readily incorporated into Chen et al : Developing a growth-transition matrix for stock assessments of Strongy/ocentrotus droebachiensis 743 55 o 06 0.5 0.4 03 0.2 0.1 0 72 73 74 / \ M ■' \ 1 \ \' '' 75 - - 76 77 / ; 1 U \ \ \ ' 78 ,^XO'.K^VV■■. - - - -79 - 78 83 88 0.5 0.2 63 80 / \ '^ 81 82 83 84 85 / ' A' \-' ■•/'■' -.'l \ '• 86 1 i\iynA\\ 87 .'.''// /)(y-/.V'A^v:^. 79 1 n 0.9 0.8 1 96 07 0.6 0.5 97 98 99 100 0.4 J 0.3 - 0.2 0.1 - n - 99 92 94 Midpoint of size class (mm) Figure 6 Probabilities of sea urchins growing from one size class to others. Each probability distribution was labeled with the midpoint value of the current size class of the sea urchin. 744 Fishery Bulletin 101(4) a size-structured stock assessment model to evaluate the status of sea urchin stock and to evaluate alternative man- agement strategies for the Maine sea urchin fishery (Chen and Hunter, 2003). Acknowledgments We would like to thank the Maine Department of Marine Resources, the Northeast Consortium, and the Maine Sea Urchin Zone Council for supporting this study. Construc- tive and detailed comments from two anonymous reviewers and the scientific editor greatly improved an early version of the manuscript, for which we are grateful. Literature cited Chen, S., S. Watanabe, and K. Takagi. 1988. Growth analysis on fish population in the senescence with special reference to an estimation of age at end of reproductive span and life span. Bull. Jpn, Soc. Sci. Fish. 54:1567-1572. Chen, Y., P. Breen, and N. Andrew. 2000. Impacts of outliers and mis-specification of priors on Bayesian fisheries stock assessment. Can. J. Fish. Aquat. Sci. 57:2293-2305. Chen, Y., and H. H. Harvey 1994. Maturation of white sucker, Catostomus commer- soni, populations in Ontario. Can, J. Fish. Aquat. Sci. 51: 2066-2076. Chen, Y., and M. Hunter 2003. Assessing the green sea urchin (Strongylocentrotus drobachiensis) stock m Maine, USA. Fish. Res. (Amst.) 60:527-537 Chen, Y, D. A. Jackson, and H. H. Harvey. 1992. A comparison for von Bertalanffy and polynomial functions in modeling fish growth data. Can. J. Fish. Aquat. Sci. 49:1228-1235. Chen, Y, D. A. Jackson, and J. E. Paloheimo. 1994. Robust regression approach to analyzing fisheries data. Can. J. Fish. Aquat. Sci. 51:1420-1429. Chen, Y., and J. E. Paloheimo. 1998. Can a more realistic model error structure improve parameter estimation in modelling the dynamics of fish populations? Fish. Res. (Amst.) 38: 9-19. Gallucci, V. F, and T. J. Quinn II. 1979. Reparameterizing, fitting, and testing a simple growth model. Trans. Am. Fish. Soc. 108:14-25. Hilborn, R., and C. Walters. 1992. Quantitative fisheries stock assessment: choice, dy- namics, and uncertainty, 570 p. Chapman and Hall, New York, NY. McArdle, B. H. 1988. The structural relationship: regression in biology. Can. J. Zool. 66:2329-2339. Meidel, S. K., and R. E. Scheibling. 1998. Size and age structure of the sea urchin Strongylocen- trotus droebachiensis in different habitats. In Echinoderms (R. Mooi, M. Telford, eds.), p 737-742. Proceedings of the 9"^ international echinoderm conference; San Francisco, 5-9 August 1996. A. A. Balkema, Rotterdam, Netherlands. Moreau, J. 1987. Mathematical and biological expression of growth in fishes: recent trends and further developments. In The age and growth offish (R. C. Summerfelt and 0. E. Hall (eds.), p 81-113. Iowa State Univ. Press, Ames, lA. Nikolskii, G. V. 1969. Theory of fish population dynamics, 323 p. Oliver & Boyd, Edinburgh, UK. Pauly, D. 1980. On the interrelationships between natural mortality, growth parameters and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor Mer 39:175-192. Quinn, T. J., II, and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Ricker,W. E. 1975. Computation and interpretation of biological statistics offish populations, 382 p. Bull. Fish. Res. Board Can., vol. 191. Rousseeuw, P. J., and A. M. Leroy 1987. Robust regression and outlier detection, 352 p. John Wiley & Sons, New York, NY. Russell, M. P 1998. Resource allocation plasticity in sea urchin: rapid diet induced, phenotypic changes in the green sea urchin, Strongylocentrotus droebachiensis (MiiUer). J. Exp. Mar Biol. Ecol. 220:1-14. Sen, A. K., and M. 8. Srivastava. 1990. Regression analysis: theory, methods and applications, 350 p. Springer- Verlag, New York, NY. Stergiou, K. I. 1993. Nutrient-dependent variation in growth and longev- ity of the red bandfish, Cepola macrophthalma (L.), in the Aegean Sea. J. Fish Biol. 42:633-644. Sullivan, P J. 1992. A Kalman filter approach to catch-at-length analysis. Biometrics 48:237-257. Sullivan, P J., H. L. Lai, and V. R Gallucci. 1990. A catch-at-length analysis that incorporates a stochas- tic model of growth. Can. J. Fish. Aquat. Sci. 47:184-198. Summerfelt, R. C, and G. E. Hall. 1987. The age and growth of fish, 530 p. Iowa State Univ. Press, Ames, lA. Vadas, R. L. 1977. Preferential feeding: an optimization strategy in sea urchins. Ecol. Monogr. 47:337-371. Vadas, R. L., B. Heal, S. Dudgeon, and W. Wright. 1997. Reproductive biology of green sea urchins along the coast of Maine: final report, 59 p. Maine Cooperative Extension Service and Maine Sea Grant Program, Orono, ME. Vadas, R. L., B. Smith, B. Beal, and T. Dowling. 2002. Sympatric growth morphs and size bimodality in the green sea urchin (Strongylocentrotus droebachiensis). Ecol. Monogr. 72:113-132. Walters, C. J. 1998. Evaluation of quota management policies for develop- ing fisheries. Can. J. Fish. Aquat. Sci. 55:2691-2705. 745 Abstract— Par tun us pelagicus was collected at regular intervals from two marine embayments and two estuaries on the lower west coast of Australia and from a large embayment located approximately 800 km farther north. The samples were used to obtain data on the reproductive biology of this species in three very different envi- ronments. Unlike females, the males show a loosening of the attachment of the abdominal flap to the cephalo- thorax at a prepubertal rather than a pubertal molt. Males become gonadally mature (spermatophores and seminal fluid present in the medial region of the vas deferentia) at a very similar carapace width (CW) to that at which they achieve morphometric maturity, as reflected by a change in the relative size of the largest cheliped. Logistic curves, derived from the prevalence of mature male P. pelagicus, gener- ally had wider confidence limits with morphometric than with gonadal data. This presumably reflects the fact that the morphometric (allometric) method of classifying a male P. pelagicus as mature employs probabilities and is thus indirect, whereas gonadal struc- ture allows a mature male to be read- ily identified. However, the very close correspondence between the CWj^'s derived for P. pelagicus by the two methods implies that either method can be used for management purposes. Portunus pelagicus attained maturity at a significantly greater size in the large embayment than in the four more southern bodies of water, where water temperatures were lower and the densities of crabs and fishing pressure were greater. As a result of the emigra- tion of mature female P. pelagicus from estuaries, the CWj^'s derived by using the prevalence of mature females in estuaries represent overestimates for those populations as a whole. Estimates of the number of egg batches produced in a spawning season ranged from one in small crabs to three in large crabs. These data, together with the batch fecundities of different size crabs, indi- cate that the estimated number of eggs produced by P. pelagicus during the spawning season ranges from about 78,000 in small crabs (CW=80 mm) to about 1,000,000 in large crabs (CW=180 mm). Manuscript approved for publication 19 June 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:745-757 (2003). Reproductive biology of the blue swimmer crab (Portunus pelagicus, Decapoda: Portunidae) in five bodies of water on the west coast of Australia Simon de Lestang Norman G. Hall Ian C. Potter Centre for Fish and Fisheries Research Division of Science and Engineering Murdoch University South Street Murdoch, Western Australia 6150 E-mail (for S de Lestang): simond@murdoch edu.au Portunid crabs, such as Portunus pelagi- cus, Scylla serrata, and Callinectes sapi- dus, form the basis of important commercial and recreational fisheries. The blue swimmer crab (P. pelagicus) is found in sheltered nearshore marine waters and estuaries throughout the Indo-West Pacific (Stephenson, 1962; Kailola et al., 1993). In Australia, the commercial catches of this portunid have increased greatly during the last 20 years, and annual catches in 1998 reached 4377 metric tons (t) (Anony- mous, 2000). The commercial fishery for P. pelagicus in Western Australia is the largest in Australia; the catch in the 1999-2000 financial year weighed 673 t and fetched a wholesale price of approximately $A3 million fCAES^). Large numbers of portunids fre- quently enter estuaries as juveniles and remain there for an extended period (Hill, 1975; Potter et al., 1983; Perkins-Visser et al., 1996; Potter and de Lestang, 2000). Although female portunids sometimes become ovigerous in estuaries, such individuals emigrate into coastal marine waters, where they release their eggs (Van Engel, 1958; Metcalf et al., 1995; Potter and de Les- tang, 2000). In contrast, the individuals of those assemblages of portunids that occupy marine embayments often do not leave these marine environments to spawn and, in cases where there is a salinity gradient, they spawn in the high salinity regions of those systems (e.g. Campbell, 1984; Sumpton et al., 1994; Prager, 1996; Potter and de Les- tang, 2000). The most common method for deter- mining the size at which male crabs at- tain maturity is to estimate the size at which the pattern of growth of one of its appendages changes from that which characterizes juvenile crabs to that which characterizes adult crabs (e.g. Hartnoll, 1974; Somerton, 1980; Reeby et al., 1990; Muino et al., 1999). Howev- er, this indirect approach is not precise and requires careful measurements of a considerable number of individuals covering a wide size range. Despite the fact that macroscopic characters can be used to distinguish sequential stages in the development of the vas deferentia of portunids (Ryan, 1967a; Meagher, 1971), few studies have attempted to use such staging to determine the body size at which the gonads of male crabs attain maturity (e.g. Reeby et al., 1990). Sumpton et al. (1994) considered that, as in female P. pelagicus, a marked loos- ening of the attachment of the abdomi- nal flap to the cephalothorax signaled the attainment of maturity in male P. pelagicus. However, this criterion has yet to be shown to be valid for the males of this species. Although varia- tions in the size at which crustaceans reach maturity among bodies of water and geographical regions may reflect, 1 CAES (Department of Fisheries, Catch and Effort Statistics). 2002. Unpubl. data. Western Australian Department of Fisheries, Catch and Effort Sta- tistics. Fisheries Western Australia, WA Marine Research Laboratories, West Coast Drive, Waterman, 6020, Perth, Australia. 746 Fishery Bulletin 101(4) in part, differences in such features as genetic composition and density, there is a strong overall tendency for the size at maturity of this crustacean to be inversely related to water temperature (Pillai and Nair, 1971; Jones and Simons, 1983; Polovina, 1989; Dugan et al., 1991; Miliou, 1996; Somerton and Donaldson, 1996; Fisher, 1999). Estimates of the fecundity of crabs have typically been based on the number of eggs in a single batch of eggs (e.g. Potter et al., 1983; Melville-Smith, 1987; Ingles and Braum, 1989). However, such an approach does not take into account the fact that female crabs often produce more than one batch of eggs during a spawning season (Van Engel, 1958; Pillai and Nair, 1971; Campbell, 1984). The aims of this study were as follows. 1) Compare the results of three methods directed at determining whether male P. pelagicus have attained maturity and elucidate whether each method produces reliable results. 2) Compare aspects of the reproductive biology oi P. pelagicus in two estuaries and two marine embayments in temperate Australia with those of this species in a large marine embayment in a much warmer and more northern subtropical environment. Particular emphasis will be placed on comparing the size at maturity of both sexes and the periods during which ovigerous females are present, and on proposing reasons for the significance of any differences between the assemblages in these five bodies of water 3) Use the data collected for one of the marine embayments to determine the age and time of year at which P. pelagicus becomes mature and develop a method for deriving the annual fe- cundity that takes into account the fact that the larger individuals of this species are believed to produce more than one batch of eggs in a spawn- ing season. ■ 25 S Indian Ocean 30 S Cockburn Sound Peel-Harvey Estuary , Leschenault Estuary 1 . >- Koombana Bay 110E 120 E Figure 1 Map showing locations of the five bodies of water in which Portunus pelagi- cus was sampled on the west coast of Australia. The map of Australia (insert top right) shows the location (box) of the five bodies of water, and the map of the lower west of Australia (insert bottom left) shows the location of the four bodies of water sampled in this region. Materials and methods Sampling regimes Up to 100 Portunus pelagicus were collected monthly for two years from the Leschenault Estuary (May 1997-April 1999), Koombana Bay, and Cockburn Sound (June 1998- May 2000), for three years from the Peel-Harvey Estuary (May 1995-April 1998), and bimonthly for two years from Shark Bay (July 1998-May 2000). The first four bodies of water are located on the lower west coast of Australia, approximately 800 km to the south of Shark Bay (Fig. 1). The nearshore, shallow waters in each of these bodies of water (water depth <1.5 m) were sampled for P. pelagicus by using a 21.5-m seine net with a bunt of 3-mm mesh; whereas offshore deeper waters were sampled by employ- ing a small otter trawl net with a codend of 25-mm mesh and crab traps consisting of either 12- or 76-mm mesh (see Potter and de Lestang, 2000 for further details of the nets and traps). The mean water depths at the deeper offshore sites of the above five bodies of water were 3, 9, 19, 3, and 10 m, respectively. The water temperature at the bottom of the water column at each site was recorded on each sampling occasion. Measurements and changes at puberty The carapace width (CW) of each crab, i.e. the distance between the tips of the two lateral spines of the carapace, was measured to the nearest 1 mm. The length and height of the propodus of the largest cheliped, the length of the merus of the second walking leg, and the length of the pri- mary pleopod of each male crab in Cockburn Sound and Shark Bay were measured to the nearest 0.1 mm. Because the relationship between the length of the dorsal propo- dus of the largest cheliped and the width of the carapace showed the greatest change over the size range of male crabs, that structure was chosen for allometric analysis to determine the size at which males become morphometri- cally mature. De Lestang et al.: Reproductive biology of Portunus pe/agicus 747 Sex of small crabs, i.e. with a CW < about 30 mm, was determined with a dissecting microscope to ascertain whether their pleopods bore setae and thus the crabs were females. At CWs > about 30 mm, the female crabs could readily be distinguished from male crabs by their posses- sion of a far wider abdominal flap (Van Engel, 1958; Warner 1977). During the pubertal molt of female portunids, the abdominal flap changes from a triangular to oval shape and from being tightly to loosely fixed to the cephalothorax (Ryan, 1967b; Fielder and Eales, 1972; Ingles and Braum, 1980; Fisher, 1999). The size and time of occurrence of all ovigerous females were recorded. The ovary of each crab was assigned to one of four stages by using macroscopic characters similar to those described for the development of the ovaries of P. pelagicus and other portunids (Ryan, 1967b; Meagher, 1971; Krol et al., 1992; Kumar et al.^). The assignment of these stages was augmented by examining the characteris- tics of a subset of 200 of these ovaries in 6-/.im histological sections that had been stained with Mallory's trichrome. For 5-10 ovaries of each macroscopic stage, the diameters of 30 randomly selected oocytes that had been sectioned through the nucleus were measured to the nearest 5 pm. Two measurements (the longest diameter and shortest di- ameter) for each oocyte were then averaged to provide an estimate of each oocyte diameter. Male crabs were designated as either morphometrically immature or mature by using differences in the regression equations for the relationships between the natural loga- rithms of the length of the dorsal propodus of their largest cheliped and carapace width in what were clearly either juvenile (small and gonadally immature) or adult crabs (large and gonadally mature). For full description of the method see Somerton (1980). On the basis of their macroscopic appearance, the vas deferentia of each male crab were assigned to one of three stages by using criteria derived from the description of go- nadal development for P. pelagicus by Meagher ( 1971) and for P. sanguinolentus by Ryan (1967a). Aquarium studies by Meagher (1971) showed that male crabs with gonads at stages I and II did not copulate and are thus considered im- mature, whereas those with gonads at stage III copulated successfully with females and thus have mature gonads. Ovaries and vas deferentia from a wide size range of at least 20 females and 20 males, respectively, from each sampling occasion in each of the five bodies of water were weighed to the nearest 0.01 g. The mean gonad weight at a constant carapace width for each sex in each month in each water body was determined by using analysis of covariance (ANCOVA) of the natural logarithm of the go- nad weight as the dependent variable, month as a fixed factor, and the natural logarithm of the carapace width as a covariate. The common constant carapace width of '^ Kumar, M., Y. Xiao, H. Williams, G. Ferguson, G. Hooper, and S. Venema. 1999. Blue crab fishery. South Australian Fisher- ies Assessment Series. 99/02, 64 p. South Australian Research Development Institute, Grenfell Centre Level 14. 2.5 Grenfell Street Adelaide 5000, Australia. crabs in all bodies of water was a default value calculated by the ANCOVA. Size frequency and reproductive data for the corre- sponding months in the different years in each water body were pooled for describing intra-annual trends in these variables. Size at maturity The percentages of female crabs of different carapace widths which, in each water body, had undergone a pubertal molt, were subjected to logistic regression to determine the size at which 50% of the female crabs would have become mature sensu Hartnoll (1974). Data for each assemblage were randomly resampled and analyzed to create 1000 sets of bootstrap estimates of the parameters of the logis- tic regression and estimates of the probability of maturity within the range of recorded carapace widths. The 95% con- fidence intervals of the CWgg's were derived by using this resampling technique, which produced slightly more con- servative estimates than those obtained from the Hessian matrix of the logistic regression and thus reflected better the uncertainty of the parameter that was associated with the data. The 95% confidence intervals of the probability of maturity at each specified carapace width were taken as the 2.5 and 97.5 percentiles of the corresponding pre- dicted values resulting from this resampling analysis. The point estimate of each parameter and of each probability of maturity at the specified carapace width were taken as the medians of the bootstrap estimates. The percentages of mature male crabs at different cara- pace widths in each of the five bodies of water, with matu- rity being assigned by using firstly morphometric and then gonadal criteria (see earlier), were subjected to logistic re- gressions to determine the CWgg's for these variables. The percentages of male crabs in Cockburn Sound and Shark Bay, which possessed an abdomen that was loosely fixed to the cephalothorax, were likewise subjected to logistic regression analysis. The logistic regressions relating ma- turity and carapace width for both the females and males in the different assemblages were compared by using a likelihood ratio test, as described by Cerrato (1990) and emplo3ring a Bonferroni correction. Fecundity The total wet weight of eggs in each batch of eggs of 40 early-stage ovigerous females, i.e. with yellow eggs, from Cockburn Sound and which covered a wide size range, was weighed to the nearest 0.001 g. The number of eggs in each of four replicate subsamples from each batch were recorded, after which each of those subsamples was weighed to the nearest 0.001 g. These data were then used to estimate the total number of eggs in each batch of eggs of each female. The relationship between batch fecundity (BF) and carapace width (CW) was described by using the equation \nBFj=m\nCWj+b. The number of batches of eggs produced by a full size range of mature females during the spawning period was estimated by determining the spawning period (SP), 748 Fishery Bulletin 101(4) defined as the time (days) when > 5% of all ma- ture females were ovigerous, and the proportions of ovigerous females among all mature females in se- quential 10-mm CW intervals during the spawning period. The proportion of ovigerous females (O^) in the^th size class during this period also represents the average time a mature female in this size class is ovigerous during that period and takes into ac- count the fact that an ovigerous female spawns at least once during a spawning period and that the brood period (BP) of an ovigerous female is about 18 days at 20°C (Meagher. 1971). Thus, the mean number of batches (NB ) produced by the mature female crabs in thejth size class during a spawn- ing period (average water temperature 20.4°C ) can be estimated with the equation NB=O^SP/BP. The relationship between number of broods and carapace width was described empirically by fitting a modified logistic curve, NB=l+Nb^^J 1 1 -i-exp[-ln(l 9 )(CW-a)/(6-a)), ranging upwards from a minimum of one batch to a maximum of l+NB^^^ batches, where a and b are parameters. The total fecundity of crabs at different carapace widths was calculated as the product of batch fecundity, BF, and the number of broods, NB, by using the relationships between BF and CW and NB and CW, as described above. Results o -•— Shark Bay -o — Cockburn Sound -i- Peel-Harvey Estuary -♦— Leschenault Estuary 30 28 26 24 a 22 k 20 E (D r 18 0) 14 12 10 JFIylAMJJASOND Month Figure 2 Mean monthly water temperatures for sampling sites in Shark Bay, Cockburn Sound, Peel-Harvey Estuary, and Leschenault Estuary (water temperatures in Koombana Bay were essentially the same as those in Leschenault Estuary). Mean monthly water temperatures in each water body were derived from data pooled for at least two years. The black rectangles on the .v axis refer to summer and winter months and open rectangles to autumn and spring months. Water temperature Mean monthly water temperatures at the bottom of the water column in the Leschenault Estuary, Peel-Harvey Estuary, Cockburn Sound, and Shark Bay followed the same trends, with values rising to a maximum in mid to late summer and declining to a minimum in mid-winter (Fig. 2). Water temperatures in Koombana Bay were essentially the same as those in Leschenault Estuary. Although the mean monthly water temperatures in the Leschenault and Peel-Harvey estuaries and Koombana Bay in corresponding months were similar, they were lower in these bodies of water than in Cockburn Sound in eight of the twelve months of the year (Fig. 2). However, the mean water temperatures in each month in Shark Bay were greater than those in the corresponding months in each of the above four more southern bodies of water. Thus, for example, although the maximum mean monthly water temperature was 28°C in Shark Bay, it never reached 25°C in any of the other bodies of water (Fig. 2). Likewise, the minimum monthly water temperature was greater in Shark Bay (19°C) than in either Cockburn Sound ( 16°C) or the Leschenault and Peel-Harvey estuaries (12-13°C). Macroscopic and histological gonad staging Macroscopic examination of the gonads of a large number of females and males oi P. pelagicus, covering a wide size range, and, in the case of females, an histological examina- tion of the ovaries of a subset of these crabs, showed that the ovaries and vas deferentia could be classified into four and three developmental stages, respectively (Tables 1 and 2). Size at sexual maturity The minimum carapace widths of female crabs that had undergone their pubertal molt ranged from 61 mm in both the Peel-Harvey Estuary and Shark Bay to 84 mm in the Leschenault Estuary. Although the CWg^'s derived for females at maturity in Cockburn Sound (86.4 mm) and Koombana Bay (86.9 mm) were not significantly dif- ferent (P>0.05), both of these values were significantly less (P<0.05) than the 92.0 mm for females in Shark Bay (Fig. 3). The high CWg^'s for female crabs in the Peel- Harvey (97.5 mm) and Leschenault estuaries (98.0 mm) were not representative of females in their populations as a whole (see "Discussion" section). The relationships between the dorsal length of the larg- est cheliped and carapace width of male P. pelagicus in each of the five bodies of water were described better by using two log-log lines (Fig. 4A) rather than a single log-log line. The CWjjg's of male crabs at morphometric maturity in the four bodies of water on the lower west coast, esti- mated with data obtained from an allometric approach and employing the above log-log regressions, ranged only from 86.2 mm in the Peel-Harvey Estuary and Cockburn De Lestang et al : Reproductive biology of Portunus pelagicus 749 Table 1 Morphological characteristics of macroscopic stages in the development of the ovaries of Portunus pelagicus and the types of oocytes found in each of those stages. Mean diameters of oocytes at different stages in development are shown in parentheses. Maturity stage Macroscopic appearance of ovary Types of oocytes I Immature Relatively small, flattened and off white to ivory in color Anterior region is small, and does not displace the hepa- topancreas. The central "H" shaped region, located in the gastric region, is loosely joined to the dorsal surface of the spermathecae. The posterior section, located in the cardiac and intestinal regions, forms two parallel lobes. II Early Conspicuously larger than stage-! ovaries, pale yellow, oval development in cross section and slightly nodulated. The anterior region marginally displaces the hepatopancreas and the central region envelops the dorsal surface of the spermathecae, and the two lobes of the posterior region are starting to become convoluted. III Late Large, yellow, and nodulated. Anterior region displaces the development hepatopancreas, and the central and posterior regions occupy almost all of the space in the gastric, posterior and intestinal cavities. Most of the spermathecae are enveloped by ovarian tissue. IV Fully mature Very large, deep yellow to orange, and highly nodulated. Hepatopancreas is now completely displaced from its former position by the enlargement of the anterior region of the ovary. The gastric, posterior, and intestinal cavities are completely filled with the enlarged central and posterior sections of the ovary. The spermathecae are totally enveloped by the ovary. Loosely packed oocytes, comprising oogo- nia (5 ^m) and, to a lesser extent, chromatin nucleolar oocytes (10 ^m) and perinucleolar oocytes (30 ^m). These three types of oocytes are found in each of the next three ovarian stages. Yolk-vesicle oocytes (90 ^m) are present for the first time. Early yolk-granule oocytes (130 fjm) sur- round small areas of early stage oocytes, and some late yolk vesicle oocytes are present. Advanced oocytes all at the late yolk- granule stage (250 /jm). Table 2 Morphological characteristics of stages in the development of the vas deferens of Portunus pelagicus and the location of spermato- phores in those stages. Maturity stage External appearance of vas deferentia Histological appearance of the vas deferentia I Immature I Vas deferentia not detectable macroscopically. NA II Immature II Anterior vas deferentia (AVD) becoming enlarged, middle Spermatophores present in AVD. MVD and PVD and posterior vas deferentia (MVD and PVD, respectively) contain no spermatophores. straight and opaque. III Mature I AVD and MVD enlarged and white and PVD enlarged and Spermatophores present in AVD and MVD PVD convoluted but still opaque. contains no spermatophores. Sound to 87.2 mm in the Leschenault Estuary (Fig. 4B). The CWjq's for each of these bodies of water, which were not significantly different (P>0.05) from each other, were significantly less at P < 0.05 or 0.001 than the 96.0 mm determined for male crabs in Shark Bay. The CWjq's for males at gonadal maturity in each water body, derived from the prevalence of males with mature go- nads, i.e. stage III (Fig. 4C), differed by only 0.3 to 2.2 mm from those derived for males in each corresponding water body by using the prevalence of morphometrically mature males (Fig. 4B). The CWg^'s derived for male crabs from gonadal data in the four southern bodies of water, which ranged only from 86.5 to 88.4 mm, did not differ signifi- cantly (P>0.05 ). However, on the basis of gonadal data, each of these CWjg's differed significantly at P<0.05, 0.01, or 0.001 from the 97.0 mm estimated for male crabs in Shark Bay (Fig. 4C). These trends were parallel to those derived from morphometric data (Fig. 4B). The logistic curves derived from gonadal data in each of the four southern bodies of water were significantly differ- 750 Fishery Bulletin 101(4) Female size at maturity 180 100 80 60 40 20 0 100 80 60 40 20 0 Koombana Bay Ij III 86 9 mm —-^■- \ (358) 20 60 100 140 180 Cockburn Sound 180 60 100 140 180 Carapace width (mm) 60 100 140 180 Carapace width (mm) Figure 3 Logistic regressions and their 95% confidence limits fitted to percentage contribu- tions of those adult females which, at each size, had undergone their pubertal molt in each of the five bodies of water sampled in Western Australia. Arrows and measure- ments denote CW^q's and the numbers in parentheses refer to the number of crabs used to create the regressions. ent (P>0.05) and had steeper slopes than those determined by using morphometric data (Fig. 4). The confidence limits for the logistic curves constructed from gonadal data were also usually tighter than those constructed from morpho- metric data. The CWr,Q for male crabs with a loose abdominal flap in Cockburn Sound, i.e. 72.1 mm, differed significantly (/'<0.05) from that in Shark Bay, i.e. 76.2 mm (data not shown). However, all of the male crabs in Cockburn Sound with carapace widths of 70 to 75 mm and loosely attached abdominal flaps contained gonads at stage I or II and were thus immature. Trends exhibited by gonad weights and proportions of ovigerous females The mean monthly gonad weight of mature female crabs with a standard carapace width ( 104 mm), as determined by ANCOVA (see "Material and methods" section), rose to a sharp peak of about 5 g in October in Koombana Bay and in September in Cockburn Sound (Fig. 5). In contrast, the mean monthly gonad weights of mature female crabs in the Leschenault and Peel-Harvey estuaries remained at <1.5 g and did not tend to peak sharply at any time of the year. The mean monthly gonad weights of mature female De Lestang et al.: Reproductive biology of Portunus pelagicus 751 T3 Q. 0) Leschenault Estuary Koombana Bay Peel-Harvey Estuary ./ / Cockburn Sound Shark Bay 100 50 0 100 50 0 100 50 = 0 100 50 0 100 50 2 3 4 5 6 Log carapace width (mm) 96 0 mm (602) 100 50 0 100 50 0 100 50 = 0 100 50 0 100 50 88.0 mm J/^ (76) J 86 5 mm (67) 97,0 mm f (813) 0 40 80 120 160 Carapace width (mm) 0 40 80 120 160 Carapace width (mm) Figure 4 Maturity data for male Portunus pelagicus in the five bodies of water sampled in Western Australia. (A) Relationship between the natural logof length of the dorsal propodus of the largest cheliped and the natural log of carapace width. Logistic regressions and their 95% confidence limits were fitted to percentage contributions of those adult males, which, at each size, were (B) morphometrically mature and (C) possessed mature gonads. Arrows and measurements denote CW^^'s and the numbers in parentheses refer to the number of crabs used to create the regressions. crabs in Shark Bay did not peak sharply at any time and were > 1 g in all but two of the ten months in which this embayment was sampled (Fig. 5). The mean monthly gonad weights of male crabs with a standard carapace width of 118.4 mm, as determined by ANCOVA, varied little and never exceeded 1 g in any of the five bodies of water (data not shown). However, they did reach their maxima at a similar time of the year, i.e. late summer (February) or early autumn (March), in the four bodies of water on the lower west coast of Australia. The monthly percentage contributions made by oviger- ous female crabs from all mature female crabs in Koom- 752 Fishery Bulletin 101(4) 10 8 6 4 2 0 10 8 6 4 2 0 10 g, g Peel-Harvey i 6 ■ ^s'^^^y 2 0 10 8 6 4 2 0 10 8 6 4 2 0 Leschenault Estuary Koombana Bay Cockburn Sound Shark Bay iir~i-i — m V \, O -1' % '^ \ '^^'^ ^<5 \ ' Figure 5 Mean monthly gonad weights +95'/f confidence intervals for female Por- tunus pelagicus and the percentage of ovigerous females among all adult female P. pelagicus in the five bodies of water sampled. The black rectangles on the -V axis refer to summer and winter months and open rectangles to autumn and spring months. bana Bay and Cockburn Sound peaked at 32-36% in No- vember and December in Koombana Bay and at 35-45% in October to December in Cockburn Sound. Although the corresponding percentage contributions of ovigerous female crabs were far lower in the Leschenault and Peel- Harvey estuaries than in the above two embayments, they still reached their maxima at the same time of the year (Fig. 5). Few or no ovigerous female crabs were caught in either of those estuaries or embayments between March and August. Ovigerous females were found in Shark Bay in each of the ten months in which that embayment was sampled and, unlike the situation in the two more southern marine embayments, their monthly contributions to the overall number of adult female crabs did not vary markedly throughout the year (Fig. 5). Trends exhibited by oocyte development The maximum diameter of the oocytes increased progres- sively from 95 pm in stage-I gonads to 315 pm in stage-IV gonads (data not shown). The modal oocyte diameter of the distinct and largest cohort of oocytes in stage IV (240-259 ^m) was only slightly less than that of the fertil- ized yellow external eggs found under the abdominal flap of ovigerous females (300-319 /jm). The most advanced oocytes in ovaries at stages III and IV were at the early and late yolk-granule stage, respectively. The presence of two distinct size cohorts of oocytes in the ovaries of large females with grey eggs under their abdomen (i.e. eggs that had been fertilized for several days) is consistent with the view that large female P. pelagicus are multiple spawners. De Lestang et al : Reproductive biology of Portunus pelagicus 753 20 15 0 '■ 30 15 0 30 t Female crabs r-Th-T-r-n 10 0 30 r 15 0 ^ 30 0' 30 15 0 30 r Male crabs 40 80 120 160 0 Carapace width (mm) 40 80 120 160 Figure 6 Frequency histogi-ams for the carapace widths of juvenile and adult females and males of Portunus pelagicus in Cockburn Sound, n = number of crabs measured. Designation of males as mature was based on gonadal criteria described in the text. Age and time of sexual maturation of Portunus pelagicus The carapace-width frequency data for P. pelagicus in inshore and offshore waters in Cockburn Sound demon- strated that, in this marine embayment, males are repre- sented by two main size cohorts in January and February (Fig. 6). The first size cohort represents the 0+ age class that resulted from the spawning period that commenced in the previous August-September, whereas the second cohort corresponds to l-i- crabs, which start to decline markedly in numbers after February and are rarely represented after June (Fig. 6). Although similar trends are exhibited by the data for females, the numbers of 1+ individuals of this sex remained higher for a longer period, i.e. until May. The above trends are entirely consistent with those reported in detailed studies of the age composition and growth of P. pelagicus in the Peel-Harvey Estuary (Potter et al., 1983) and Leschenault Estuary (Potter and de Lestang, 2000). None or very few of the female and male 0+ crabs caught in Cockburn Sound in January, February, and March were mature. However, some of the larger 0-t- crabs had become mature by May, i.e. when they would mostly have been 754 Fishery Bulletin 101(4) O 350 o ° 300 • • X - 250 r^^^^ "O 200 1 '50 ■g 100 S 50 4 03 ■5 3 m ^ • ^ o 2 y^ Number 0 5- 1000 0 0 '- 800 X ^^^^--"^^ Total fecundity 0000 ^ ^--^^ 80 90 100 110 120 130 140 150 160 Carapace width (mm) Figure 7 Batch fecundity, mean number of batches, and total fecundity of Por- tunus petagicus in Cockburn Sound. between four and eight months old (Fig. 6). The prevalence of mature crabs subsequently increased, with the result that the vast majority of crabs in the following January, i.e when they had just entered their second year of life, were mature (Fig. 6). Thus, all crabs have typically become mature when they are just over one year old. Fecundity In Cockburn Sound, the number of eggs recorded for a single batch of eggs under the abdomen of a female, ranged from 68,450 in a crab with a CW of 84 mm to 324,440 in a crab with a CW of 154 mm (Fig. 7). The rela- tionship between batch fecundity (BF) and carapace width (CW) is described by the following equation: lnfiF= 1.82081nCH'+3.2862. The estimated mean number of egg batches, produced by female crabs in the different size classes over the spawn- ing period, ranged from about one in crabs of 100-109 mm CW to about three in crabs of 150-159 mm CW (Fig. 7). A range of one to three batches per instar corresponds to that recorded by Campbell ( 1984) for P. pelagicus in aquaria ex- periments. The empirical relationship between the number of batches (NB) and carapace width (CW) is described by A^Bj=l+2/{l+exp[-ln(19)(CW,-113.7)/13.8]|. A combination of the equations for the relationships between batch fecundity and CW and the number of egg batches and CW was then used to determine the relation- ship shown between total fecundity (TF) and carapace width (CW) and which is shown in Figure 7. Discussion Designation of maturity in male crabs Aquaria studies by Meagher (1971) demonstrated that male crabs with gonads at stage 111, i.e. with spermato- phores and seminal fluid in the medial vas deferentia, can copulate successfully with females. Because this parallels the situation recorded by Comeau and Conan (1992) for the snow crab iChionoecetes opilio), we likewise regard such gonads as mature. Our study also showed that, because male P. pelagicus still possess immature gonads (stage II) when their abdominal flap becomes loosely attached to the cephalothorax, the latter change occurs at a prepubertal De Lestang et a\ : Reproductive biology of Portunus pe/agicus 755 molt and thus, unlike the supposition of Sumpton et al. (1994), does not coincide with the attainment of maturity. The situation in males thus contrasts with that in female P. pelaglcus, in which the abdominal flap becomes loose as an outcome of the pubertal molt (Fielder and Eales, 1972; Campbell, 1984; Potter and de Lestang, 2000; Smith^). The very close similarity between the corresponding CWjq's derived for male P. pelagicus in each of the five bodies of water by using morphometric and gonadal data demonstrates that morphological and gonadal maturity are attained by this species at essentially the same cara- pace width. However, the question of whether a male crab of about the size of maturity has become morphometrically mature depends on determining whether the relative length of one of its appendages is closer to the regression line which relates the length of that appendage to the cara- pace width in either juvenile or adult crabs. Because the overall relationship between cheUped length and carapace width of P. pelagicus does not undergo a marked shift at around the attainment of maturity, the use of the allome- tric method never enabled us to determine with absolute certainty whether, in the region of size overlap, a male was morphometrically immature or mature. The lack of precision, when determining maturity with morphometric data, could account for the slopes of the logistic curves for the prevalence of "mature" individuals of P. pelagicus de- rived from these morphometric data in the four southern bodies of water being shallower than those obtained from gonadal data. From the above, it follows that there would be an advan- tage in determining the CW^q's for male P. pelagicus at ma- turity by using data on gonadal state obtained by the simple and direct procedure of examining the vas deferentia, rather than relying on data obtained by an allometric method that is indirect and relies on a careful measurement of the ap- pendage lengths and carapace dimensions of a considerable number of individuals. However, the remarkable similari- ties between the CWg^'s derived by using morphometric and gonadal data show that, if it is desirable to avoid damaging the crabs, the data obtained from allometric analysis does yield a close approximation of this important measure for P. pelagicus. Thus, the CW^q derived from either gonadal or morphometric data for male P. pelagicus can be used for developing management plans for this species. The very close correspondence between the size at which gonadal and morphometric maturity are attained by the males of P. pelagicus contrasts with the situation recorded by Comeau and Conan (1992) and Sainte-Marie et al. ( 1997) for the males of the snow crab Chionoecetes opilio. In this latter species, the males attain gonadal maturity at a smaller body size than that at which morphometric ma- turity is attained following the terminal molt. The males of C. opilio with large cheliped and large body size are at a competitive advantage over smaller males when courting (Comeau and Conan, 1992; Sainte-Marie et al., 1997). Be- •^ Smith, H. 1982. Blue crabs in South Australia — their status, potential and biology. Report 6, p. 33-51. South Australian Fisheries Industry Council Grenfell Centre Level 14, 25 Grenfell Street Adelaide 5000, Adelaide, Australia. cause the aquaria studies of Campbell (1984) have shown that the large males of P pelagicus also have a similar com- petitive advantage during courting, a male of this species with mature gonads and differentiated chelipeds may not be able to compete for females successfully if larger males are present. Influence of migration on estimates of CW50 for female crabs The mean monthly gonad weights recorded for post- pubertal individuals were less for females in estuaries than in marine embayments, strongly indicating that females often tend to emigrate from the estuaries to their spawning grounds before their gonads are fully developed (Van Engel, 1958; Potter and de Lestang, 2000). Such an emigration from estuaries by mature female P. pelagicus reduces the proportion of mature individuals within each carapace width interval, thereby increasing the proportion of immature females in these class intervals. This shifts the logistic curve to the right and consequently increases the CW5Q, which accounts for the significantly greater CWgg's derived for females in estuaries than in marine environ- ments on the lower west coast of Australia. For this reason, subsequent comparisons of the CW^q's for female crabs in the different bodies of water will focus on those derived for assemblages in the three marine embayments. In contrast to the CWgg's for females, the CWgg's for males at maturity in the two estuaries and the two marine embayments on the lower west coast of Australia were not significantly different. This presumably reflects the fact that, unlike mature females, the large males of P. pelagicus tend to remain in estuaries during the spawning period (Potter and de Lestang, 2000). Influence of temperature on reproductive biology The CWjq's derived for males at "maturity" in each of the five bodies of water never differed by more than 2.2 mm, irrespective of whether gonadal or morphometric data were used. However, the maximum CWgg's determined for males in the two estuaries and two embayments by using gonadal and morphometric data, i.e. 88.4 mm for Cockburn Sound and 87.2 mm for the Leschenault Estuary, respec- tively, were 8.6 and 8.8 mm less than the corresponding CWgg's determined for males in Shark Bay Furthermore, the CWgg's for female P. pelagicus in Koombana Bay and Cockburn Sound were 5.6 and 5.1 mm, respectively, less than that of females in Shark Bay. The greater CWgg's for P. pelagicus in Shark Bay than in the other four bodies of water, which are located approxi- mately 800 km farther south, runs counter to the general- ization that the CWgg's for decapods tend to be inversely related to water temperature (e.g. Campbell and Robinson, 1983; Jones and Simons, 1983; Dugan et al., 1991). Howev- er, the opposite situation has sometimes been recorded and, in those cases, has been attributed to differences among populations of one or more of the following: density, preda- tion pressure, and food availability (Hines, 1989; Polovina, 1989; Pollock, 1995; McGarvey et al., 1999). It thus appears relevant that the mean density of P. pelagicus was far lower 756 Fishery Bulletin 101(4) in the sampling sites in Shark Bay, 0.6 crabs/100 m'-, than those in Cockburn Sound and Koombana Bay, 2.80 and 2.94 crabs/100 m-, respectively. The mean density in Shark Bay was also far lower than those recorded in the Leschenault and Peel-Harvey estuaries between the middle of spring and middle of autumn, when P. pelagicus colonizes estu- aries (Potter et al., 1983; Potter and de Lestang, 2000). Furthermore, commercial or recreational fishing pressure (or both), which leads to a reduction in CWgy's at maturity in the spiny lobster (Polovina, 1989), is far greater for P. pelagicus in the southern bodies of water than in Shark Bay (Bellchambers''). Recent work with microsatellite DNA has also shown that the assemblages of P. pelagicus in Shark Bay are genetically distinct from those in more southern bodies of water, such as Cockburn Sound and the Peel-Harvey Estuary (Chaplin et al.^). The marked differences between the CWjq's at maturity for P. pelagicus in Shark Bay and bodies of water farther south emphasize the need for managers to take into ac- count this type of variation when determining a minimum legal carapace width (MLCW) for capture. However, the current MLCW for P. pelagicus in Western Australia, 127 mm, is well above even the CW^^ for this species at maturity in Shark Bay. The prevalence of ovigerous females did not peak sharply at any time of the year in Shark Bay, whereas ovigerous females were found predominantly during spring and summer in Cockburn Sound and Koombana Bay. Moreover, the mean monthly gonad weights of a female P. pelagicus of standard carapace width lay within a relatively narrow range of 0.9 to 1.8 g in Shark Bay, whereas they rose to a sharp peak of about 5 g in spring and fell below 1 g in some months in Cockburn Sound and Koombana Bay. The trends exhibited by the reproductive variables of female P. pelagi- cus thus provided strong evidence that reproductive activity extends over much or all of the year in Shark Bay, whereas it occurs predominantly in spring and summer in the two southern embayments. The more protracted spawning period in Shark Bay presumably reflects the presence of higher water temperatures throughout the year and in par- ticular during winter and early spring. Such a conclusion is consistent with the results of other studies, which have shown that water temperature influences ovulation and egg development in P. pelagicus and other decapods (Rahaman, 1980; Campbell, 1984; Pollock, 1995; Kumar et al.^). Fecundity The vast majority of previous estimates of the fecundity of crustaceans have been based on the number of eggs borne by females at a particular time which, in the case of multiple spawners, does not take into account the fact that ■' Bellchambers, L. 2002. Personal conimun. Fisheries West- ern Australia, WA Marine Research Laboratories, West Coast Drive, Waterman, 6020, Perth, Australia. 5 Chaplin, J., E.S.Yap.E.Sezmis.andl. C.Potter. 2001. Genetic (microsatellite) determination of the stock structure of the blue swimmer crab in Australia. FRDC project 98/1 18. 84 p. Mur- doch University, South Street, Murdoch, 6150, Perth, Australia. larger crabs can produce two or more batches of eggs within a spawning period. The few previous attempts to obtain the total fecundity of crustaceans have involved tracking the number of batches of eggs borne by particular individuals at different times (e.g. Chubb et al.^). The advantage of the approach developed during the current study is that it uses a combination of batch fecundity and an estimate of the number of batches produced during the spawning period by female P. pelagicus of different carapace widths to determine the relationship between the total fecundity and body size of this species in a given population. Because the older crabs have a far longer intermolt period between copulation and egg extrusion than younger crabs, i.e. eight versus four months, they have a far greater amount of time to accumulate the energy reserves required to produce eggs. This difference accounts for the greater number of egg batches produced by larger than small crabs. Acknowledgments Thanks are expressed to many colleagues and friends, and particularly R. Melville-Smith, D. Fairclough, M. Pember, T Linke, M. Travers, and W. White, who assisted with sampling. Our thanks are also expressed to the three anonymous referees for their constructive criticisms. Fund- ing was provided by the Australian Fisheries Research and Development Corporation and Murdoch University. Literature cited Anonymous. 2000. FAO yearbook. Fisheries statistics. Capture produc- tion, 86/1, 713 p. FAO, Rome, Italy Campbell, G. R. 1984. A comparative study of adult sexual behaviour and larval ecology of three commercially important portu- nid crabs from the Moreton Bay region of Queensland, Australia. Ph.D. diss. 253 p. Univ. Queensland, Bris- bane, Queensland, Australia. Campbell, A., and D. G. Robinson. 1983. Reproductive potential of three American lobster (Homarus americanus) stocks in the Canadian Maritimes. Can. J. Fish. Aquat. Sci. 40:1958-1967. Cerrato, R. M. 1990. Interpretable statistical tests for growth comparisons using parameters in the von BertalaniTy equation. Can. J. Fish. Aquat. Sci. 47:1416-1426. Comeau, M., and G. Y. Conan. 1992. Morphometry and gonad maturity of male snow crab, Chionoecetes opilio. Can. J. Fish. Aquat. Sci. 49: 2460-2468. Dugan, J. E., A. M. Wenner, and D. M. Hubbard. 199 1 . Geographic variation in the reproductive biology of the sand crab Emerita analoga (Stimpson) on the California coast. J. Exp. Mar Biol. Ecol. 150:63-81. '^ Chubb, C, C. Dibdcn, and K. Ellard. 1984. Studies on the breeding stock of the western rock lobster, Paniilirus cygnus. in relation to stock and recruitment. FIRTA project 85/87, 37 p. Fisheries Western Australia, WA Marine Research Laboratories, West Coast Drive, Waterman, 6020, Perth, Australia. De Lestang et al : Reproductive biology of Portunus pelaglcus 757 Fielder, D. R., and A. J. Bales. 1972. Observations on courtship, mating and sexual matu- rity in Portunus pelagicus (L., 1766). J. Nat. Hist. 6: 273-277. Fisher, M. R. 1999. Effect of temperature and salinity on size at maturity of female blue crabs. Trans. Am. Fish. Soc. 128:499-506. Hartnoll, R. G. 1974. Variation in growth pattern between some secondary sexual characters in crabs (Decapoda, Brachyura). Crus- taceana 27:131-136. Hill, B. J. 1975. Abundance, breeding and growth of the crab Scylla serrata in two South African estuaries. Mar. Biol. 32: 119-126. Hines, A. H. 1989. Geographic variation in size at maturity in brachy- uran crabs. Bull. Mar. Sci. 45:356-368. Ingles, J., and E. Braum. 1989. Reproduction and larval ecology of the blue swim- ming crab Portunus pelagicus in Ragay Gulf, Philip- pines. Int. Rev Hydrobiol. 74:471-490. Jones, M. B., and M. J. Simons. 1983. Latitudinal variation in reproductive characteristics of a mud crab, Helice crassa (Grapsidae). Bull. Mar Sci. 33:656-670. Kailola, P. J., M. J. Williams, P C. Stewart, R. E. Riechelt, A. McNee, and C. Grieve (eds.). 1993. Australian fisheries resources, 422 p. Bureau of Resource Sciences, Canberra, Australia. Krol, R. M., W. E. Hawkins, and R. M. Overstreet. 1992. Reproductive components. In Microscopic anatomy of invertebrates (F. W. Harrison and A. G. Humes, eds.), p. 295-343. Wiley-Liss, New York, NY. McGarvey, R., G. J. Ferguson, and J. H. Prescott. 1999. Spatial variation in mean growth rates at size of southern rock lobster, Jasi/s edwardsii, in South Australian waters. Mar Freshw. Res. 50:333-342. Meagher, T. D. 1971. Ecology of the crab Portunus pelagicus (Crustacea: Portunidae) in south Western Australia. Ph.D. diss, 227 p. Univ Western Australia, Perth, Australia. Melville-Smith, R. 1987. The reproductive biology of Geryon maritae (Decapoda, Brachyura) off south west Africa/Namibia. Crustaceana 53:259-275. Metcalf K. S., J. van Montfrans, R. N. Lipcius, and R. J. Orth. 1995. Settlement indices for blue crab megalopae in the York River, Virginia: temporal relationships and statistical efficiency Bull. Mar. Sci. 57:781-792. Miliou, H. 1996. The effect of temperature, salinity and diet on final size of female Tisbe holothuriae (Copepoda, Harpacti- coida). Crustaceana 69:742-754. Muifio, R., L. Fernandez, E. Gonzalez-Gurriaran, J. Freire, and J. A. Vilar. 1999. Size at maturity of Liocarcinus depurator (Brachy- ura: Portunidae): a reproductive and morphometric study. J. Mar Biol. Assoc. U.K. 79:295-303. Perkins-Visser, E., T. G. Wolcott, and D. L. Wolcott. 1996. Nursery role of seagrass beds: enhanced growth of juvenile blue crabs iCallinectes sapidus Rathburn). J. Exp. Mar Biol. Ecol. 198:155-171. PiUai, K. K., and N. B. Nair 1971. The annual reproductive cycles of Uca annulipes, Portunus pelagicus and Metapenaeus affinis (Decapoda: Crustacea) from the south-west coast of India. Mar Biol. 11:152-166. Pollock, D. E. 1995. Changes in maturation ages and sizes in crustacean and fish populations. S. Afr J. Mar. Sci. 15:99-103. Polovina, J. J. 1989. Density dependence in spiny lobster, Panulirus mar- ginatus, in the northwestern Hawaiian Islands. Can. J. Fish. Aquat. Sci. 46:660-665. Potter, I. C, P. J. Chrystal, and N. R. Loneragan. 1983. The biology of the blue manna crab Portunus pelagi- cus in an Australian estuary. Mar. Biol. 78:75-85. Potter, I. C, and S. de Lestang. 2000. Blue swimmer crab Portunus pelagicus in Leschen- ault Estuary and Koombana Bay, south-western Aus- tralia. J. R. Soc. West. Aust. 83:221-236. Prager, M. H. 1996. A simple model of the blue crab, Callinectes sapidus, spawning migration in Chesapeake Bay. Bull. Mar. Sci. 58:421-428. Rahaman, A. A. 1980. Ecological observations on spawning of a few inver- tebrates of the Madras coast. J. Madurai Kamaraj Univ. 9:71-77. Reeby, J., P. N. Prasad, and M. S. Kusuma. 1990. Size at sexual maturity in the male crabs of Portu- nus sanguinolentus and P. pelagicus. Fish. Technol. 27: 115-119. Ryan, E. P 1967a. Structure and function of the reproductive system of the crab Portunus sanguinolentus (Herbst) (Brachyura: Portunidae). I. The male system. Mar. Biol. Assoc. India Symp. Ser. 2:506-521. 1967b. Structure and function of the reproductive system of the crab Portunus sanguinolentus (Herbst) (Brachyura: Portunidae). II. The female system. Mar. Biol. Assoc. India Symp. Ser 2:522-544. Sainte-Marie, B., J. M. Sevigny, and Y Gauthier. 1997. Laboratory behaviour of adolescent and adult males of the snow crab (Chionoecetes opilio) (Brachyura: Majidae) mated noncompetitively and competitively with primipa- rous females. Can. J. Fish. Aquat. Sci. 54:239-248. Somerton, D. A. 1980. A computer technique for estimating the size of sexual maturity in crabs. Can. J. Fish. Aquat. Sci. 37: 1488-1494. Somerton, D. A., and W Donaldson. 1996. Contribution to the biology of the grooved and triangle tanner crabs, Chionoecetes tanneri and C. angulatus, in the eastern Bering Sea. Fish. Bull. 94:348-357. Stephenson, W 1962. The evolution and ecology of portunid crabs, with espe- cial reference to Australian species. In The evolution of living organisms (G. W. Leeper, ed.), p. 34-67. Melbourne Univ. Press, Melbourne, Australia. Sumpton, W. D., M. A. Potter, and G. S. Smith. 1994. Reproduction and growth of the commercial sand crab, Portunus pelagicus (L.) in Moreton Bay, Queensland. AsianFish. Sci. 7:103-113. VanEngel.W.A. 1958. The blue crab and its fishery in Chesapeake Bay Part 1. Reproduction, early development, growth and migration. Commer. Fish. Rev. 20:6-17. Warner, G. F. 1977. The biology of crabs, 202 p. Paul Elek (Scientific Books), London. 758 Abstract— Culture of a non-native species, such as the Suminoe oyster iCrassostrea ariakensis). could offset the harvest of the declining native eastern oyster (Crassostrea virginica) fishery in Chesapeake Bay. Because of possible ecological impacts from intro- ducing a fertile non-native species, introduction of sterile triploid oysters has been proposed. However, recent data show that a small percentage of triploid individuals progressively revert toward diploidy, introducing the possibility that Suminoe oysters might establish self-sustaining populations. To assess the risk of Suminoe oyster populations becoming established in Chesapeake Bay, a demographic popu- lation model was developed. Parameters modeled were salinity, stocking density, reversion rate, reproductive potential, natural and harvest-induced mortal- ity, growth rates, and effects of vari- ous management strategies, including har\'est strategies. The probability of a Suminoe oyster population becoming self-sustaining decreased in the model when oysters are grown at low salinity sites, certainty of harvest is high, mini- mum shell length-at-harvest is small, and stocking density is low. From the results of the model, we suggest adopt- ing the proposed management strate- gies shown by the model to decrease the probability of a Suminoe oyster population becoming self-sustaining. Policy makers and fishery managers can use the model to predict potential outcomes of policy decisions, supporting the ability to make science-based policy decisions about the proposed introduc- tion of triploid Suminoe oysters into the Chesapeake Bay. A model for assessing the likelihood of self-sustaining populations resulting from commercial production of triploid Suminoe oysters iCrassostrea ariakensis) in Chesapeake Bay Jodi R. Dew Jim Berkson Eric M. Hallerman Department of Fisheries and Wildlife Sciences 106 Cheatham Hall Virginia Polytechnic Institute and State University Blacksburg, Virginia 24061-0321 E-mail address (for J. Berkson, conlaci auttior). |berkson(gvt.edu Standish K. Allen Jr. School of Manne Science Virginia Institute of Manne Sciences Gloucester Point, Virginia 23062 Manuscript approved for publication 16 June 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. lOhT.'JS-TeS (2003). The native eastern oyster (Crassostrea virginica) population in Chesapeake Bay has declineti because of habitat degradation, over-harvest, and dis- ease- and parasite-mediated mortahty. Efforts to restore the eastern oyster pop- ulation in Maryland and Virginia have been hindered by persistent diseases and habitat degradation (Mann et al., 1991; Gottlieb and Schweighofer, 1996). Recent restoration efforts have included intensified reef building programs. In addition to restoring the native oyster, discussions about introducing non- native disease- and parasite-resistant oyster species into the Chesapeake Bay have gone forward since the early 1990s (Mann et al., 1991; Lipton et al., 1992; Gottlieb and Schweighofer, 1996; Hallerman etal., 2002). In 1997, in-water testing of non-na- tive oyster species (sterile triploids) began in Virginia, first with the Pa- cific oyster (Crassostrea gigas), then with the Suminoe oyster (Crassostrea ariakensis) (Calvo et al., 1999; Calvo et al., 2001). Field studies with Pacific oysters showed poor performance under Chesapeake Bay conditions (Calvo et al., 1999). However, field studies with Suminoe oysters demonstrated disease resistance and rapid growth, and indi- viduals reached minimum harvest shell length of about 77 mm in approximately one year (Calvo et al., 2001). These re- sults, and subsequent small-scale trials by industry, evoked strong interest in the commercial culture of Suminoe oys- ters to supplement the eastern oyster fishery. Ideally, aquaculture with 100% tri" ploid oysters would pose no risk of es- tablishment of a self-sustaining oyster population (Guo and Allen, 1994a). However, a number of factors make the use of triploids imperfect. For example, recent data have shown that a small percentage of triploid oysters progres- sively revert toward diploidy with age (Calvo et al., 2001; Zhou. 2002). Rever- sion of triploids leads to mosaicism in which individuals comprise both dip- loid and triploid cells. Mosaics them- selves are innocuous unless the re-es- tablishment of diploid cells leads to re- covered reproductive capability, which could in turn lead to the establishment of a self-sustaining Suminoe oyster population. We define this hazard, "reproductively effective reversion," as the process of yielding mosaics with re- covered reproductive capability. Repro- ductively effective reversion introduces the possibility that triploid Suminoe Dew et al: Model for assessing populations of Crassostrea anakensis in Cfiespeake Bay 759 oysters planted for aquaculture could become a self-sustaining population of diploid Suminoe oysters and introduce numerous unknown eco- logical consequences. Another hazard associated with deployment of triploid Suminoe oysters is the possibility that nontriploids might be stocked inadver- tently because of failure to detect them in a mixed batch of triploid and diploid individu- als. Although technology to produce "100%" triploids is now available, as practiced on Pacific oyster (Guo and Allen, 1994b; Guo et al., 1996), the reliability of the approach for producing "100%" triploids in Suminoe oyster is yet undetermined. Diploids may enter the population from several sources: chromosomal nondisjunction in tetraploid males producing haploid gametes, low level hermaphrodism in diploid females yielding self-fertilized embryos, and cross-contamination between diploid and triploid cultures (cf Guo and Al- len, 1997). Typically, flow cytometry has been used to determine the presence or absence of diploid cells (Allen, 1983). Flow cytometry has the sensitivity to detect one diploid among a thousand triploid oysters (Allen and Bushek, 1992); thus, the detection threshold is 0.001 with current technology. Should the (nonzero) frequency of diploids be greater than zero but less than one in a thousand, then the batch would be certified 100% triploid. This failure to detect diploid individuals in a mixed batch poses a hazard for stocking other fertile diploid oysters in that batch into culture systems. Before substantial commercial introduction of triploid Suminoe oysters into the Chesapeake Bay, any environ- mental hazards of reproduction associated with a range of management scenarios should be assessed. Hazards are defined as undesirable outcomes from an activity (Hallerman and Kapuscinski, 1995). Stocking triploid Suminoe oysters produces two hazards in this model: the inadvertent stocking of diploids and the reproductively effective reversion of triploids. These two hazards may lead to the establishment of a self-sustained Suminoe oyster population and the probability of this occurring is defined as a risk. Risk assessment is the process of 1) identifying hazards posed by management actions, such as deployment of triploid Suminoe oysters, 2) quantifying the associated risks of hazards being realized (Hallerman and Kapuscinski, 1995), such as the population becoming self-sustaining, and 3) evaluating the consequences of the hazards. Quantitative models often are used to as- sess risk (Lackey, 1994). Building upon data collected on growth, mortality, and reproductively effective reversion for Suminoe oysters, we have developed a quantitative model to estimate the risk associated with large-scale deployment of triploid Suminoe oysters under a range of management scenarios. The model predicts the likelihood of out-planted triploid Suminoe oysters giving rise to a self-sustaining population at a given site in the Chesa- peake Bay given user-specified stocking, reproductively Slocking, age-class zero | Surviving stocked age-class zero oysters Starting population size Next year Grow to mean shell length Surviving juvenile oysters ^ Mortality ^ Reproduction Natural mortality Reproductively effective reversion and detection threshold Surviving adult oysters Final population Figure 1 Flow chart depiction of the annual time step in the model for estimating likelihood of estabhshing self-sustaining reproduction in triploid Suminoe oysters (C. ariakensis). effective reversion, reproduction, growth, and mortality rates (both natural and harvest), as well as user-specified management options. Methods Overview of model A quantitative population model of the Suminoe oyster was developed to evaluate the consequences of hazards asso- ciated with introducing triploid Suminoe oysters under a range of environmental conditions and management strategies. The model includes set demographic parameters (length-fecundity, oyster density-fertilization efficiency, and salinity-fecundity relationships) and user-specified variables (reproduction, growth, and natural and harvest mortality rates). It includes options for varying stock- ing rates, harvest rates, and other management actions. Because little is known about Suminoe oyster reproduc- tion, we assumed that Suminoe oysters would behave like the congeneric eastern oysters in Chesapeake Bay; hence, an eastern oyster fecundity model (Mann and Evans, 1998) was used to estimate fecundity of Suminoe oysters. The model assumes that the Suminoe oyster population is closed, i.e. that natural immigration and emigration do not occur The model is age-structured, and a yearly time step is used. The state variable tracked through time is population size. Intrinsic population growth rate is exponential and without density dependence. The final output of the model is the predicted population size of Suminoe oyster assum- ing specified demographic parameters and environmental and management variables. The model was programmed in Visual Basic (Microsoft Corp., Redmond, WA). 760 Fishery Bulletin 101(4) Modeling approach In each annual time step for age classes one through six, growth occurs to the mean shell length of the age class, then natural mortality and harvest are imposed, and then reproduction occurs (Fig. 1). Because Suminoe oysters grow quickly in autumn (Cahn, 1950), the annual time step begins in September Harvest occurs from October to April. Natural mortality occurs at the greatest rates during the summer months. Because an annual time step is being used, the model is designed so that natural mortality and harvest are imposed simultaneously. Reproduction occurs during the summer months. The model simulates repro- duction for fertile individuals in all mature age classes. The final population size for a particular age class after natural mortality and harvest becomes the starting value for population size for the next age class in the next time step. All individuals stocked each year are age-class zero individuals. The starting population size for age-class one in the next time step is equal to the sum of all individuals less than one-year old produced by all age classes, plus the number of individuals stocked. Model variables, parameters, and equations The initial conditions for the model are determined by the user's choice of specific values for several variables (Tables 1 and 2). The key abiotic variable driving population growth is salinity, because fecundity is highly dependent upon salinity (Mann and Evans, 1998). Biotic variables of the model include mean shell length for each age class, mor- tality (natural and harvest) for each age class, disease prevalence, total mortality of oysters less than one year old, oyster population density, sex ratio for each age class, and reproductively effective reversion rate for each age class (Table 1). Other variable inputs are stocking rates, harvest regulations, and management strategies. Stochasticity is programed into the model to incorporate both the uncertainty involved in estimating variable values and environmental variation. Some variables are regarded as stochastic variables because they vary around some mean value from year to year, whereas other variables (such as salinity and sex ratio of the population for each age class) are deterministic in the model because they fluctuate over a longer period of time in the absence of a catastrophe (Kennedy et al., 1996). Stochasticity affects shell length, natural mortality, and reproductively effective reversion rates at each age, and the degree of variance is set by the user as a constant for each year At each time step, a mean shell length, mortality rate, and reproductively effective reversion rate for each age class is randomly drawn from a log normal distribution around a mean with an associated variance. We assume that the mean shell length of each age class at the current time step does not affect the mean length of the age class at a subsequent time step, because of large, highly variable growth rates per year (Calvo et al., 2001). Default mean shell length for each age-class values were obtained from Cahn ( 1950). The user may, of course, specify other mean shell lengths. Growth affects the potential for Table 1 Definitions of model parameters and variables. Symbol Parameter and variable definition A Area (square meters) C Certainty in obtaining the desired harvest rate (00.05) revealed little popu- lation structure among samples. Mantel tests demonstrated that the genetic relationships among samples did not correspond to an isolation-by-distance model for either class of marker. Four of eight comparisons of coastal and offshore samples revealed differences of allele frequencies at the Gpi-F* locus (P<0.05), although none of these differ- ences was significant after correction for multiple testing (P>0.001). Results are consistent with the hypothesis that the CYRA yellowfin tuna samples com- prise a single genetic stock, although gene flow appears to be greater among coastal samples than between coastal and offshore samples. Allozyme and RAPD variation in the eastern Pacific yellowfin tuna Whunnus albacares) Pindaro Dfaz-Jaimes Instituto de Ciencias del Mar y Limnologia Unlversidad Nacional Autonoma de Mexico Circuito exterior de Ciudad Universitaria Apdo. Postal 70-305 Mexico, D F 04510 E-mail address; pindaro@maricmyl-unam.mx Manuel Uribe-Alcocer Instituto de Ciencias del Mar y Limnologia Universidad Nacional Autonoma de Mexico Circuito exterior de Ciudad Universitaria Apdo, Postal 70 305 Mexico, D.F 04510 Manuscript approved for publication 19 June 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:769-777. Yellowfin tuna (Thunnus albacares) is a cosmopolitan species inhabiting tropical and subtropical waters in the Atlantic, Pacific, and Indian oceans. This species has accounted for more than a third of the world's tuna production since 1970. The eastern Pacific has contributed from 21% to 26% of the global catch from 1993 through 1997, represent- ing 273,329 metric tons (t) in 1990 to 264,426 t in 1998 (lATTC, 1999). Yellowfin tuna is a large pelagic fish with a common size of 150 cm (Collette and Nauen, 1983). Spawning occurs throughout the year in the tropical oceans, preferably near islands and coasts (Leis et al., 1991). Growth is rapid and individuals reach maturity by the end of the second year (Suzuki et al., 1978). Schooling of individuals of similar size is observed near surface waters and is often associated with floating objects (Wild, 1994). Yellowfin tuna is currently consid- ered to comprise a single species (Gibbs and CoUete, 1967), although significant morphometric and meristic differences, limited fish movements, and differences among catch data, have been reported for the different regions of the Pacific Ocean (Godsil and Greenhood, 1951; Schaefer, 1955; Joseph et al., 1964; Suzuki et al., 1978; Schaefer, 1991). Population structure in yellowfin tuna has been addressed in the Pacific Ocean by using several independent methods. Morphometric and meristic based stud- ies have shown significant differences (Godsil and Greenhood, 1951; Schaefer, 1955; Kurogane and Hiyama, 1957), and at least three stocks or discrete units (western, central, and eastern Pa- cific) have been proposed. More recent studies using morphometric multivari- ate analysis suggest the presence of dif- ferent stocks between north and south regions in the eastern Pacific (Schaefer, 1991), as well as across the Pacific Ocean (Schaefer, 1991). Additionally, differences in larval distribution, catch rates, and size composition data of yel- lowfin tuna caught along the equatorial Pacific by longline and purse-seine have been used by Suzuki et al. ( 1978) to dis- tinguish between western, central, and eastern Pacific groups. Tagging experiments have shown limited movement of yellowfin tuna between western and eastern Pacific waters (Joseph et al., 1964; Fink and Bayliff, 1970). In the eastern Pacific, the presence of two groups has been suggested: a northern group off Baja California coast and the Revillagigedo Islands and a southern group from the Maria Islands through Chile. Some mixing occurs between them (Fink and Bayliff, 1970). There seem to be marked movements between north and south groups along the coast with limited westward movements (Joseph et al., 1964). 770 Fishery Bulletin 101(4) Some studies of population structure using genetic anal- yses have not revealed the presence of discrete stocks along the Pacific Ocean. Barret and Tsuyuki (1967) used transfer- rin analysis and did not find differences in allele frequen- cies between samples from Hawaii and eastern Pacific samples, although heterogeneity was detected within the eastern Pacific samples (lATTC, 1975). Allozyme variation studies in the esterase locus (Fujino, 1970) did not show enough evidence of genetic differentiation between east- ern Pacific and Hawaii samples. Furthermore, Scoles and Graves (1993) used restriction fragments length polymor- phisms (RFLP) and analysis of mitochondrial (mt) DNA to examine five samples collected across the Pacific Ocean and one from the Atlantic Ocean. Although they detected 34 haplotypes and considerable genetic variation, no evidence of genetic differentiation among samples was found. However, more recent genetic studies have provided lim- ited evidence of genetic heterogeneity. Ward et al. (1994) analyzed four polymorphic allozyme loci and 18 mtDNA haplotypes in yellowfin tuna from the Pacific Ocean. Al- though no unique haplotypes were found in the analyzed populations through RFLP analysis, the eastern Pacific samples were found to be different from the central and western Pacific samples in frequency differences at a sin- gle locus GPI-F*, suggesting that the signal of population structure exhibited is due to selective factors contributing to the divergence. Eastern Pacific samples (n=41) were collected in the northeast Pacific off California and at an unspecified site off Mexico (n=40). Comparisons of GPI-F* allele frequencies from eastern Pacific also included two samples previously analyzed by Sharp (1978) from Roca Partida (Central America) and Ecuador. Their results showed population homogeneity at the GPI-F* locus for this region. To date, the methods and logistics used to study diver- gence in the Pacific yellowfin tuna have been focused on a global rather than a local scale, and sampling has been focused on the wide areas of the west and central Pacific. Local structure in the eastern Pacific yellowfin tuna has not been addressed through a more intense sampling strategy to examine genetic homogeneity in this region. Because tagging studies have shown restricted longitudi- nal movements by yellowfin tuna, population structure and isolation by distance hypotheses can be tested. To evaluate the stock structure of yellowfin tuna in eastern Pacific, we employed analyses of allozymes and of randomly amplified polymorphic DNA (RAPDs). RAPDs have proven to be useful genetic markers because of their high levels of polymorphisms (Williams et al., 1990; Welsh et. al., 1991). They have been used to estimate population structure in fishes, including the cod (Kenji, 1998), red mullet (Mamuris et al., 1998), and striped bass (Bielawski and Pumo, 1997). The use of RAPDs, considered as neutral markers, and the simultaneous use of allozyme analyses with intense sampling in a more local area, might provide evidence about the relationship between gene flow and spatial distribution of the eastern Pacific yellowfin tuna, as well as evidence of the presence of local selective factors responsible for the divergence suggested by Ward etal.(1994). Materials and methods Sampling A total of 327 tissue samples from specimens of ten loca- tions were obtained from commercial tuna boats fishing in the tropical eastern Pacific from 1994 to 1996 (Fig. 1). Muscle tissue samples were dissected from specimens at the time of landing and were transported in liquid nitrogen or on dry ice to the Laboratorio de Genetica de Organismos Acuaticos of the Instituto de Ciencias del Mar y Limnologia in Mexico City. Samples were maintained at -70°C until processing. Allozyme analysis For allozyme analysis, 1 cm-* (about one gram) of tissue sample was ground with a manual homogenizer in 1.5 mL of extraction buffer (O.OIM Tris-O.OOIM EDTA, pH 6.8, and 1% NADP) and centrifuged at 2500 g at 4°C. Electrophoretic runs were performed in 12% (w/v) starch gels ( Sigma Chemicals, St. Louis, MO ). Four buffer systems were used to analyze nineteen enzymes that resolved 28 loci, eight of which showed polymorphism: Aat-S* (aspar- tate aminotransferase), Glud (glutamate dehydrogenase), Gpi-F* and Gpi-S* (glucose phosphate isomerase). La (leucil-L-alanine), Lgg (L-leucil-glycil-glycine), Pap-F* (L-leucil-L-proline) and 6-Pgd (phosphogluconate dehy- drogenase). Enzymes AK (adenilate kinase), CK (cre- atinine kinase), GAPDH (glyceraldehyde-3-phosphate dehydrogenase), LDH (lactate dehydrogenase), MDH (mMalate dehydrogenase), ME (malic enzyme) and SOD (superoxide dismutase), displayed twenty more loci that were presumably monomorphic. Buffer systems for enzyme analysis were 1) amino-citrate: 0.04M citric acid, 15mL/L of N-3-aminopropyl-morpholine, pH 6.5 (AAT, GPI, and LA); 2) 0.008M Tris, 0.003 M citric acid, pH 6.7 (GLUD and 6-PGD); 3) 0.025 M Tris, 0.192 glycine, pH 8.5 (GPI and LGG); 4) 0.0.076 M Tris, 0.005 M citric acid, pH 8.7 (PAP). Enzyme assays were performed following Harris and Hopkinson (1976). Enzymes showing polymorphism were analyzed for all samples and subjected to population genetic analysis. RAPD analysis For RAPD analyses, genomic DNA was extracted from muscle tissue by using standard phenol-chloroform proto- cols (Sambrooketal., 1989), resuspendedinTE buffer (lOmM Tris-O.lmM EDTA pH 8.0), and quantified with a Hoefer DyNA quant 200 fluorometer. DNA was amplified with primer F-10 (Operon® Alameda, CA; 5'-GGAAGCTTGG-3'). Amplifying reactions were performed in a final volume of 22 //L consisting of 0.7 to 1 ng///L of DNA in amplification buffer, 10 mM Tris-HCl, 50 niM KCl, 1.5 mM MgCl.^, 33 ng of primer, 10 mM dNTPs, and 1 U of Taq polymerase. Amplification of genomic DNA was performed in a Perkin Elmer®, Foster City, CA (mod. 480), thermal cycler. The pro- gram was set for 1 cycle of 1 min. at 36°C, followed by 44 cycles of 1 min. at 36°C; 1 min. at 94°C; 2 min. at 72°C, and DIaz-Jaimes and Uribe-Alcocer; Allozyme and RAPD variation in Thunnus albacares 771 126° 119° Date (Nl OCY Jun94 (42) WRE Jun94 (49) NCL Aug96(6) SCL Feb9S(14) CUE Feb9f) (50) MCH Feb96(2()) COL Feb96(30) NAY Apr94(25) CSL May94 (40) GC May94(51) 28° 21° 7° - OCY "O WRE € 126° 19° Revillagigedo ," Islands Gpi- F alleles 700 Figure 1 Thunnus albacares. Location of sampling sites, dates, and sample sizes, and Gpi-F* gene frequencies of eastern Pacific yellowfin tuna. The sites are as follows: OCY = out of CYRA area; WRE = west of the Revillagigedo Islands; NCL = North Clipperton islands; SCL = South Clipperton Islands; MCH = Michoacan; QUE = Guerrero; COL = Colima; CSL = Cape San Lucas; NAY = Nayarit; GC = Gulf of California. Symbols are • coastal, A intermediate, ■ offshore. a final cycle of 15 min. at 72°C. Optimal DNA concentra- tions for amplification were determined by testing several dilutions, one of which was taken as the standard for every subsequent amplification. Amplified fragments were resolved by electrophoresis in 1.5% agarose gels (Sigma Chemicals) for 3 to 4 h. at 90 mA (100 V). A lOObp DNA Ladder (GibcoBRL, Gaithersburg, MD, 15628-019) was used as size standard. After electro- phoresis, gels were stained with ethidium bromide and photographed in a UV light transilluminator. Data analysis Allelic frequencies, test of conformity of genotype distribu- tions with Hardy-Weinberg, and heterozygous deficit were determined by using Genepop version 3.3 (Raymond and RoussetM. Homogeneity of allozyme and RAPD allele fre- quencies was evaluated by using the exact probability test (Raymond and Rousset, 1995) consisting of a contingency analysis for every polymorphic locus and an estimation of their probability values by the combined probability of Fisher (Sokal and Rohlf, 1995) as implemented in the TFPGA program (Miller^). Pairwise comparisons were conducted to determine allele frequency differences among samples in order to define sources of variation. Based on the longitudinal differentiation pattern observed by Ward et al. (1994) and the morphological latitudinal differences within eastern Pacific samples reported by Schaefer (1991) 1 Raymond, M. L., and F. Rousset. 1995a. GENEPOP (ver- sion 1.2): population genetics software for exact tests and ecumenicist. J. Heredity 86:248-249. 2 Miller, M. P. 1997. Tools for genetic populations analyses (TFPGA) 1.3: a windows program for the analysis of allozyme and molecular population genetic data, 29 p. Computer software distributed by the author at http://bioweb.usu.edu/ mpmbio. 772 Fishery Bulletin 101(4) at north-south of the 15-20°N range, spatial homogene- ity was tested at the following levels: overall samples (O), among longitudinal regions (L; coastal, intermediate, and offshore localities), and latitudinal regions (N; north-south of the 15-20° range). The Gulf of California sample was excluded from this analysis because the large variation found in this sample (reflected in the significant differ- ences shown in allele frequency homogeneity from pairwise comparisons) would not allow an accurate assessment of whether longitudinal or latitudinal differentiation exists or not. Significance levels were adjusted for multiple testing through the Bonferroni sequential method (Rice, 1989). Population subdivision was estimated by using the Weir and Cockerham (1984) method through the TFPGA program. Standard error and confidence intervals were obtained through jackknife and bootstrapping procedures, respectively, with Fgj, Pro 1.0 (Weir, 1990). Estimates of pop- ulation subdivision were partitioned into the following lev- els: O, over all the samples; L, longitudinal regions (coastal, intermediate and offshore); and N, latitudinal regions. We used the 6 statistic to estimate gene values between sample pairs (Slatkin, 1993) that are defined as Mg. An "isolation by distance" model was evaluated from the cor- relation between the distance between localities measured as geographic separation in nautical miles (nmi), and the M(j values by means of the Mantel test (Hellberg, 1994) in both allozymes and RAPDs. The patterns of the amplification products resulting from the RAPD analysis were subjected to the same analyses as allozymes with the procedures described in Lynch and Milligan (1994). RAPD fragments were interpreted under the following assumptions: 1) fragments were considered to behave as dominant genes (Williams et al., 1990); 2) every polymorphic fragment was considered derived from a two-allele locus; 3) the equilibrium of Hardy- Weinberg was assumed for all genotypes, and 4) each fragment was considered to be an independent locus. Only those fragments clearly defined and having consis- tent intensity were recorded. Because of this, the Michoacan sample, with poor consistency in the banding patterns, was excluded from the RAPD analysis. The allele frequency of every fragment was calculated on the basis of the inferred homozygous recessive genotypes. Because of the dominant nature of the alleles, and in order to correct the bias originat- ed by calculating the recessive allelic frequencies, we chose the estimation based on the Taylor expansion (Kendall and Stuart, 1977, cited m Le Corre et al., 1997) as implemented in TFPGA program. This reduction on the bias is based on the equation resulting from the second order expansion of Taylor (see details in Lynch and Milligan, 1994). Results Allozymes eight polymorphic loci detected are shown in Table 1. After adjusting levels of significance by the Bonferroni procedure, significant deviation of genotypic frequencies from those expected under Hardy-Weinberg was found in the loci Lgg and Pap-F* for the Gulf of California sample and in the Aat- S* locus for two localities — west of Revillagigedo Islands, and the Gulf of California (P<0.0006, Table 1). Deviations displayed for both locations corresponded to a heterozygous deficit (P<0.0006, after Bonferroni correction). Comparison of allozyme allele frequencies among all collections (overall) by the exact probability test revealed significant heterogeneity at loci Glud, La, and Lgg, after Bonferroni adjustment (P<0.006). Pairwise comparisons among samples to test allele homogeneity gave signifi- cant differences for nine of 45 comparisons after correc- tion for multiple tests (P<0.001), seven of which involved comparisons with the Gulf of California (GC) sample. The remaining significant differences were between Guerrero- Nayarit, and Cape San Lucas-Nayarit comparisons (Table 2), resulting from significant heterogeneity at Glud, La, and Lgg loci. In general, allozyme analysis displayed low levels of dif- ferentiation. The 9 value over all loci was different from zero (P<0.05) and showed that 4.8% of the variance was attributable to differences among samples (Table 3). Individual loci showed d values ranging between 0.0037 and 0.27. Highly significant values at loci La (9=0.13 ±0.089) and Lgg (0=0.27 ±0.253) evidently resulted from their weak polymorphism in some samples (Table 3). Allele frequency homogeneity was tested among coastal, intermediate, and offshore regions (Table 4). Significant heterogeneity was found by exact test between coastal and offshore comparisons (P=0.0043) but was found to be not significant between coastal and intermediate regions (P=0.0632). Subdivision as measured by 0 among coastal, intermediate, and offshore localities was not different from zero. However, for the Gpi-F* locus the population subdivi- sion among regions (0.0058) was twice as large as that noted among samples (0=0.003), but neither value was significant (Table 3). No latitudinal differentiation by the exact test or population subdivision estimated by the 9 index was found between north and south regions (data not shown). The gene flow values (Mg) were high (mean 24.8 mi- grants per generation). A lack of correlation between gene flow estimations and geographic distance by means of the Mantel test was observed (r'-=-0.144; P=0.22), resulting in a rejection of the isolation by distance model. For the Gpi-F* locus, paired tests of significance (data not shown) showed discrepancies in the Gpi-F* 175 allele frequencies among localities from the coast with those located at the CYRA limits (Fig. 1, Table 5). Four of eight comparisons of coastal and offshore samples revealed differences of allele frequencies at this locus (P<0.05), although none of these differences was significant after correction for multiple testing (P>0.001). Of the 28 analyzed allozyme loci, eight (28.5%) showed polymorphism under the 0.95 criterion. The observed het- erozygosities per sample over all allozyme loci ranged from 0.027 to 0.083 (mean 0.052). Allozyme frequencies for the RAPDs The primer OPF-10 produced 11 amplified fragments, with sizes from 200 to 600 bp (base pairs). Four of the fragments Diaz-Jaimes and Unbe-Alcocer: Allozyme and RAPD variation in Thunnus albacares 773 Table 1 Allele frequencies for allozymes and RAPDs, samples size (n) and agreement to the Hardy-Weinberg equilibrium (HW) for every loci and sample of Thunnus albacares. Significance values for HW tests were adjusted for multiple comparisons with an initial a level of 0.0006 ((0.05/(8 loci x 10 samples)]. P = probability of significance for allele frequency heterogeneity per locus, * = significant at a = 0.006. — = no data. OCY = out of CYRA area; WRE = west of the Revillagigedo Islands; NCL = North Clipperton islands; SCL = South Clipperton Islands; MCH = Michoacan; GUE = Guerrero; COL = Colima; CSL = Cape San Lucas; NAY = Nayarit; GC = Gulf of California. Collection Locus Aat-S* Glud Gpi-F* Gpi-S* La Lgg Pap-F* 6-Pgd FIO-1 FlO-2 FlO-3 FlO-4 Allele OCY WRE NCL SCL GUE MCH COL CSL NAY -90 0.171 -100 0.829 n 41 HW yes 100 0.700 85 0.300 n 40 HW ves 135 0.071 100 0.262 75 0.667 n 42 HW yes -60 0.402 -100 0.598 n 41 HW yes 120 0.0 100 1.0 n 42 HW yes 115 0.0 100 1.0 n 42 HW yes 110 0.0 100 1.0 n 42 HW — 100 0.902 90 0.098 n 41 HW yes a b n HW a b n HW a b n HW a b n HW 0.406 0.594 9 0.406 0.594 9 0.406 0.594 9 0.115 0.885 9 0.266 0.734 32 no 0.634 0.366 41 yes 0.134 0.280 0.586 41 yes 0.536 0.464 42 yes 0.024 0.976 42 0.024 0.976 42 0.0 1.0 42 0.903 0.097 31 yes 0.282 0.718 8 0.603 0.397 0.475 0.525 0.282 0.718 8 0.250 0.750 6 yes 0.667 0.333 6 yes 0.250 0.250 0.500 6 yes 0.417 0.583 6 yes 0.0 1.0 6 0.0 1.0 6 0.0 1.0 6 0.667 0.333 6 yes 0.234 0.766 7 0.441 0.559 7 0.329 0.671 7 0.234 0.766 7 0.231 0.769 13 yes 0.857 0.143 14 yes 0.091 0.500 0.409 11 yes 0.571 0.429 14 yes 0.0 1.0 14 0.0 1.0 14 0.0 1.0 14 0.821 0.179 14 yes 0.322 0.678 9 0.322 0.678 9 0.625 0.375 9 0.178 0.822 9 0.154 0.846 39 yes 0.663 0.337 49 yes 0.190 0.230 0.580 50 yes 0.360 0.640 50 yes 0.0 1.0 50 0.0 1.0 50 0.02 0.98 50 yes 0.939 0.061 49 yes 0.275 0.725 20 yes 0.775 0.225 20 yes 0.079 0.316 0.605 19 no 0.425 0.575 20 yes 0.0 1.0 20 0.0 1.0 20 0.05 0.95 20 yes 1.0 0.0 20 0.083 0.917 6 0.278 0.722 6 0.397 0.603 6 0.083 0.917 6 0.077 0.923 26 yes 0.914 0.086 29 yes 0.185 0.370 0.444 27 yes 0.414 0.586 29 yes 0.0 1.0 24 0.0 1.0 24 0.0 1.0 29 0.966 0.034 29 yes 0.505 0.495 9 0.092 0.908 38 yes 0,788 0.212 40 yes 0.125 0.363 0.512 40 yes 0.449 0.551 39 yes 0.0 1.0 40 0.0 1.0 24 0.0 1.0 40 1.0 0,0 40 0.355 0.645 10 0.322 0.355 0.678 0.645 9 10 0.406 0.529 0.594 0.471 9 10 0.246 0.219 0.754 0.781 9 10 0.0 1.0 25 0.680 0.320 25 yes 0.160 0.260 0.580 25 yes 0.479 0.521 24 yes 0.1 0.9 20 yes 0.0 1.0 24 0.0 1.0 25 0.840 0.160 25 yes 0.282 0.718 8 1.0 0.0 0.371 0.629 0,138 0.862 8 GC 0.234 0.766 47 no 0.395 0.605 43 yes 0.200 0.314 0.486 35 yes 0.338 0.663 40 yes 0.2 0.8 50 yes 0.349 0.651 43 no 0.083 0.917 48 no 0.938 0.062 48 yes 0.246 0.754 9 0.322 0.678 9 0.406 0.594 9 0.246 0.754 9 0011 0.526 0.015 0.055 0.007 0.572 0.369 0.912 0.929 774 Fishery Bulletin 101(4) Table 2 Painvise-sample comparisons of allele frequency homogeneity for Thunnus albacares. Probability values in allozymes are above the diagonal { — ) and RAPDs are below the diagonal ( — ). * = significant values after Bonferroni correction for multiple tests (initial a was 0.05). na = data not available. OCY = out of CYRA area; WRE = west of the Revillagigedo Islands; NCL = North Clipperton islands; SCL = South Clipperton Islands; MCH = Michoacan; GUE = Guerrero; COL = Colima; CSL = Cape San Lucas; NAY = Nayarit; GC = Gulf of California. Sample OCY WRE NCL SCL GUE MCH COL CSL NAY GC OCY 0.5807 0.7391 0.3414 0.4386 0.8254 0.0295 0.1525 0.0194 *<0.001 WRE 0.9316 — 0.9482 0.6047 0.4330 0.1180 0.0012 0.0084 0.0192 *<0.001 NCL 0.9797 0.9770 — 0.9232 0.3494 0.7307 0.1567 0.0859 0.4779 0.0174 SCL 0.9798 0.8088 0.8915 — 0.3989 0.0717 0.3414 0.1093 0.0287 *<0.001 GUE 0.5935 0.4732 0.9095 0.7757 — 0.4137 0.1723 0.5932 *<0.001 *<0.001 MCH na na na na na — 0.0500 0.1532 0.0096 *<0.001 COL 0.9385 0.7864 0.9155 0.9794 0.4569 na — 0.8237 0.0082 *<0.001 CSL 0.9475 0.9703 0.9824 1.000 0.6576 na 0.9849 — 0.0012 *<0.001 NAY 0.9656 0.9984 0.8418 0.4965 0.4101 na 0.4206 0.5707 — *<0.001 GC 0.9458 0.9637 0.9987 0.9977 0.9743 na 0.9715 0.9874 0.6518 — Table 3 Estimates of population subdivision 9 (Weir and Cocker- ham, 1984) for allozymes of Thunnus albacares partitioned into longitudinal regions 0^ (regions) (i.e. coastal-interme- diate-offshore) and samples 9^ (overall), n.v. = negative values. P = probability of significance of subdivision esti- mations. Significance of single-locus values was corrected with an initial level of 0.006 (0.05/8 loci). Means and stan- dard error were obtained by the jackknife method. Con- fidence intervals obtained by 1000 resamplings through bootstrapping are also shown. * = signifcant values after Bonferroni correction. Locus L (regions) "O(overall) Aat-2* Glud Gpil* Gpi-2* La Lgg Pap-1* 6-Pgd Mean CI 95% 0.0152 0.0009 0.0058 0.0014 n.v. 0.0073 n.v. 0.044 0.0067 ±0.0042 0.0003-0.0199 0.072 0.227 0.086 0.780 0.012 0.004 0.068 >0.006* 0.029 0.086 0.003 0.0037 0.13 0.27 0.024 0.02 0.048 ±0.022 0.019-0.101 0.007 >0.006* 0.210 0.007 >0.006* >0.006 0.001 >0.006* were polymorphic for all samples (Table 1). No significant heterogeneity of RAPD allele frequencies was found for any locus between any paired sample comparison, among all collections, nor among latitudinal or longitudinal regions (P=0.4806). The mean lvalue for all fragments and samples (over- all), as well as regional estimations derived from liAPDs (0.0302), were not significantly different from zero and displayed some negative values. Klstimations of gene flow between sample pairs (Mg) from RAPU data aver- Table 4 Pairwise-regions comparison of allele frequencies for Thunnus albacares. Probabilities of nonheterogeneity for allozymes (based on exact tests) are above the diagonal ( — ) and RAPDs are below the diagonal ( — ). * = significant after corrected for multiple tests (Rice, 1989). Region Coastal Intermediate Offshore Coastal Intermediate Offshore 0.9998 0.9039 0,0632 0.9126 0.0043* 0.5384 aged 29.2 migrants per generation. The evaluation of the relationships between geographic distances and the gene flow estimations in pairwise collections (Afg), through the Mantel test, showed a nonsignificant correlation (r-=0.413, P=0.984). Discussion The test of conformance to the Hardy-Weinberg frequen- cies showed significant differences in Lgg and Pap-F* loci only in the Gulf of California sample, where polymorphism at those loci was also consistently found. Similar results were obtained, with smaller differences in locus Aat-S* from the west of the Revillagigedo Islands and the Gulf of California samples. Considering the fact that our samples were provided by the commercial fleet, they could have included representatives of different schools with differ- ences in genotypic distributions originated by differences in age classes or sexual ratios (or for both) among schools because recruitment of individuals into new schools has been reported to be mainly by aggregating individuals of Diaz-Jaimes and Uribe-Alcocer: Allozyme and RAPD variation in Thunnus albacares 775 Table S Comparison 31,(1994).— of Gpi-F* allele frequencies for Thunnus = data absent. albacares among data from the present study and those reported in Ward et Locus Allele Western/Central' Eastern' Eastern (present data) Offshore Intermediate Coastal Pooled Gpi-F* 135 0.026 0,100 0.103 0.147 0.163 0.145 100 0.640 0.269 0.301 0.412 0.301 0.307 75 0.332 0.631 0.596 0.441 0.566 0.548 40 0.002 — — — — — n 346 178 83 17 196 296 ' Allele frequ -ncies for Gpi-F* reported in Ward et al, ( 1994) similar sizes (Collette an(i Nauen, 1983). Because recruit- ment to the original tuna schools has been reported as well (Kimley and Holloway, 1999), random processes could also induce differences in genotypic frequencies that favor aggregation of some genotypes, while segregating some others, causing a kind of Wahlund effect that is reflected by a heterozygous deficit as shown by the homozygous excess for loci and locations having HW deviations, especially as shown in the Gulf of California sample. The estimations of population structure based on allo- zymes showed a small but significant value different from zero (6=0.048; P<0,01). The Gulf of California sample con- tributed to the significant subdivision value as shown when that collection was excluded from the regional subdivision analysis, as well as to significant heterogeneity of its allele frequencies when paired comparisons were made. The small value of 0 for overall estimations on RAPD data is probably due to the small sample size. The nega- tive values of 6 from overall and regional estimations re- sulted from subtracting the large value of the correction derived from the variation expected of the sample sizes from the small value of variation due to fluctuations in al- lele frequencies. The fact that RAPD data are considered dominant could reduce information about the true allele distributions by subestimation of null allele frequencies notwithstanding the correction applied to recessive geno- types, which is dependent on the sample sizes (Lynch and Milligan, 1994). Other assumptions for RAPD data limit the value of this marker, especially when estimations are derived from a small number of loci and sample sizes. Ad- ditional constraints are related to the limited number of alleles (two) to estimate dominant markers, which tend to subestimate the polymorphism and thus reduce the significance of relatively small discrepancies in allele distributions. No differentiation between coastal and offshore samples was found in our study because of the slight, nonsignifi- cant differences in the estimation of the subdivision by regions. Although the overall estimation was not different from zero, the allele homogeneity analysis showed allele- frequency heterogeneity between coastal and offshore samples, and nonheterogeneity between coastal and inter- mediate samples. These results are consistent with the migration reports through tagging studies; evidence exists for the presence of two main yellowfin tuna groups in the eastern Pacific that mix to some extent (Fink and Bayliff, 1970) and that migrate longshore from around the 20°N to the mouth of the Gulf of California and to the zone between the Revilla- gigedo and the Clipperton islands, and back again (Joseph et al, 1964; Fink and Bayliff, 1970), although longitudi- nal movements are restricted to the limits of yellowfin regulatory area (CYRA). Similarly, important northward movements along the coasts to the mouth of the Gulf of California, and subsequently to the western coasts of Baja California, have been reported. Although the estimation of 6 for allozymes showed a significant value, it was notably influenced by the heterogeneity found between the Gulf of California sample and all other samples. Discarding the variation displayed by loci La, Lgg, and Pap-F*, originating mainly from Gulf of California sample, the estimation of subdivision was still marginally significant after Bonferroni correction, which should be considered as evidence that the Gulf of California sample may represent a partially isolated population with differ- ent allele frequencies. Oceanographic conditions inside the Gulf are somewhat different from those of the Pacific Ocean where there are warmer waters at the end of the year, es- pecially during yellowfin tuna spawning seasons. There is also high productivity characterized by the presence of sig- nificant biomass abundance of sardine or anchovy schools (Cisneros-Mata et al., 1995), which represents opportuni- ties to establish the feeding and consequently the spawn- ing grounds for eastern Pacific yellowfin tuna. Likewise, there is a trend of migratory movements through the Gulf of California by different groups of yellowfin tunas (Fink and Bayliff, 1970). These movements promote stock mix- ing and help to explain the wide polymorphism displayed in this sample, in contrast to the weak variation found in other samples from the coast and offshore regions. Further genetic research, including sequential temporal sampling of young fishes in order to ensure the presence of individu- 776 Fishery Bulletin 101(4) als that originated in discrete spawning grounds, should be undertaken to prove the presence of an independent unit inside the Gulf of California, which, if confirmed, might necessitate new stock management strategies. Allele frequencies for Gpi-F* locus found in the present study, apparently, correspond to those reported by Ward et al. (1994 and 1997). These authors reported a higher pro- portion of the allele Gpi-F*75 (0.571) in the eastern Pacific region and a gradual decrement of the same allele toward the central (0.423) and western Pacific regions (0.330), where allele Gpi-F*100 (0.650) had the higher proportion. In the present study, the highest frequencies for the allele Gpi-F*75 corresponded to the region of the eastern Pacific, situated in the limits of the yellowfin tuna regulatory area (offshore region), and there was a slight decrease in fre- quencies towards the coastal area (Table 5). Furthermore, allele frequencies for the Gpi-F*75 allele from the coastal locality, Colima (0.444), and the intermediate locality southeast, Clipperton Islands (0.409), have coincidences with those reported by Ward et al. ( 1994) for the collection Hawaii 92 (0.423) in the central Pacific region. The similarities in the Gpi-F* allelic frequencies be- tween eastern (Colima and Clipperton) and central Pacific samples (Hawaii 92) might possibly be attributed to the extended migrations of yellowfin tuna in the eastern Pa- cific brought about by the strong influence of warm waters on tuna movements because of the increased depth of the thermocline layer in that area, which was reflected by a decrease in catches (Joseph and Miller, 1988; Wild, 1994) and which possibly led to the mixing of the eastern and central Pacific stocks. The low number of RAPD loci analyzed and the uncer- tainty of fulfilling some assumptions, such as the genetic identity of each band needed for qualitative and quantita- tive interpretation of data in terms of allelic frequencies, do not allow us to consider our estimations of subdivision reliable with the RAPD method. Additionally, the lack of reliability of estimations associated with high sampling variances by using randomly collected fishery samples highlights the need to design more efficient spatial and temporal sampling strategies in local and wide areas, as well as the need for alternative hypervariable markers to assess the divergence patterns observed in highly migra- tory species. Acknowledgments We are grateful to Ernesto Escobar from Pescados Industri- alizados S. A. PINSA for allowing the sampling, and Robert Olson from lATTC for providing the Gulf of California samples. We thank Monica Dominguez-Lopez, Yolanda Hornelas-Orozco, Evangelina Castillo, and Alma Hernan- dez-Perez, for collection and processing of samples, Luis Eguiarte and Valeria Souza for the facilities provided in their laboratory and three anonymous reviewers for their valuable comments. This manuscript benefited from the critical reading of John Graves, Jan McDowell, and Bar- bara Rutan. Funding for this project was provided by the Programa de Apoyo a Estudiantes de Posgrado (PADEP) and by the project IN20598 of the Programa de Apoyo a Proyectos de Investigacion e Innovacion Tecnologica. Direccion General de Asuntos del Personal Academico, Universidad Nacional Autonoma de Mexico. Literature cited Barret, I., and H. Tsuyuki, 1967. Serum transferrin polymorphism in some scombroid fishes. Copeia 1967:551-.557. Bielawski, J. P. and D. E. Pumo. 1997. Randomly amplified polymorphic DNA ( RAPD ) analy- sis of Atlantic coast striped bass. Heredity 78:32^0. Cisneros-Mata, M. A., Nevarez-Martinez, M. O., and Hammann, M. G. 1995. The rise and fall of the Pacific sardine, Sardinops sagax caeruteus Girard, in the Gulf of California, Mexico. CALCOFI Rep. 36:136-143. Collette, B. B., and C. E. Nauen. 1983. Scombrids of the world: an annotated and illustrated catalogue of tunas, mackerels, bonitos and related species known to date. In FAO species catalogue, vol. 2, 137 p. FAO Fish. Synop. 1325. FAO, Rome. Fmk, B. D., and W. H. Bayliff. 1970. Migrations of yellowfin and skipjack tuna in the east- ern Pacific Ocean as determined by tagging experiments, 1952-1964. Inter-Am. Trop. Tuna Comm. Bull. 15:1-227. Fujino, K. 1970. Immunological and biochemical genetics of tunas. Trans. Am. Fish. Soc. 99:152-178. Gibbs, R. H., and B. B. Collette. 1967. Comparative anatomy and systematics of the tunas, genus Thunnus. Fish. Bull. 86:835-838. Godsil, H. C, and E. C. Greenhood. 1951. A comparison of the populations of yellowfin tuna (Neothunnus macropterus) from the eastern and central Pacific, 33 p. Calif Dep. Fish Game Fish. Bull. 82. Harris, H., and D. A. Hopkinson. 1976. Handbook of enzyme electrophoresis in human genet- ics, 273 p. Am. Elsevier Publishing Co., New York, NY. Hellberg, M. E. 1994. Relationships between inferred levels of gene flow and geographic distance in a philopatric coral, Ballanophyllia elegans. Evolution 48:1829-1854. lATTC (Inter- American Tropical Tuna Commission). 1975. Annual report of the Inter- American Tropical Tuna Commission, 169 p. lATTC. LaJolla, CA. 1999. Annual report of the Inter-American Tropical Tuna Commission, 357 p. lATTC, LaJolla, CA. Joseph, J., F G. Alverson, B. D. Fink, and E. B. Davidoff. 1964. A review of the population structure of yellowfin tuna, Thunnus albacares. in the eastern Pacific Ocean. Inter- Am. Trop. Tuna Comm. Bull. 9(2):53-112. Joseph, J., and F. R. Miller 1988. El Nino and the surface fishery for tunas in the east- ern Pacific. Proceedings of the Tuna Fisheries Research Conference, Japan Fish. Far Seas Fish. Res. Lab. Maguro Gyogyo Kyogikai Gijiroku, Suisancho-Enyo Suisan Kenky- usho: 199-207 Kcnji, S, 1998. Genetic variation and local differentiation in the Pacific cod Gadua macrocephalus around Japan revealed by mtDNA and RAPD markers. Fish. Sci. 64(5):673-679. Diaz-Jaimes and Unbe-Alcocer: Allozyme and RAPD variation in Thunnus a/bacares 777 Kimley, A. P., and C. F. Holloway. 1999. School fidelity and homing synchronicity of yellowfin tuna, Thunnus albacares. Mar. Biol. 133:307-317. Kurogane, K., and Y. Hiyama. 1957. Morphometric comparison of the yellowfin tuna taken from the Equatorial Pacific. Jpn. Soc. Sci. Fish. Bull. 23: 388-293. Leis, J. M., T. Trnski, M. Harmelin-Vivien, J.-P. Renon, V Dufour, M. K. El Moundi, and R. Galzin. 1991. High concentrations of tuna lai-vae (Pisces: Scombri- dae) in near-reef waters of french Polynesia (Society and Taumotu islands). Bull. Mar. Sci. 48(1):150-158. Le Corre, S., S. Dumolin-Lapegue. and A. Kremer 1997. Genetic variation at allozyme and RAPD loci in sessile oak Quercus petraea (Matt.) Liebl.: the role of history and geography Mol. Ecol. 6:519-529. Lynch, M., and B. G. Milligan. 1994. Analysis of population genetic structure with RAPD markers. Mol. Ecol. 3:91-99. Mamuris, Z., A. P. Apostodilis, A. J. Theodoru, and C. Triantaphyllidis. 1998. Application of random amplified polymorphic DNA (RAPD) markers to evaluate intraspecific genetic varia- tion in red mullet {Mullus barbatus). Mar Biol. 132: 171-178. Raymond, M. L., and F. Rousset. 1995. An exact test for population differentiation. Evolu- tion 49:1280-1283. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-225. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed., 999 p. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Scoles, D. R., and J. E. Graves. 1993. Genetic analysis of the population structure of yellow- fin tuna, Thunnus albacares, from the Pacific Ocean. Fish. Bull. 91:690-698. Schaefer, K. M. 1991. Geographical variation in morphometric characters and gill-raker counts of yellowfin tuna Thunnus albacares, from the Pacific Ocean. Fish. Bull. 9:690-698. Schaefer, M. B. 1955. Morphometric comparison of yellowfin tuna from southeast Polynesia, Central America, and Hawaii. Inter- Am. Trop. Tuna Comm. Bull. 1:89-136. Sharp, G. D. 1978. Behavioral and physiological properties of tunas and their effects on vulnerability to fishing gear In The physi- ological ecology of tunas (G. D. Sharp and A. E. Dizon, eds), p. 397-449. Academic Press, New York, NY. Slatkin, M. 1993. Isolation by distance in equilibrium and non-equilib- rium populations. Evolution 47:264-279. Sokal, R., and F. J. Rohlf 1995. Biometry, 3rd ed., 776 p. W. H. Freeman and Co., New York, NY. Suzuki, Z., P. K. Tomlinson, and M. Honwa. 1978. Population structure of Pacific yellowfin tuna. Inter- Am. Trop. Tuna Comm. Bull. 17:273-441. Ward, R. D., N. G. Elliott, P. M. Grewe, and A. Smolenski. 1994. Allozyme and mitochondrial DNA variation in yellow- fin tuna {Thunnus albacares) from the Pacific Ocean. Mar Biol. 118:531-539. Ward, R. D., N. G. Elliott, B. H. Innes, A. J. Smolenski, and P. M. Grewe. 1997. Global population structure of yellowfin tuna, Thun- nus albacares, inferred from allozyme and mitochondrial DNA variation. Fish. Bull. 95:566-575. Weir, B. S. 1990. Genetic data analysis, 377 p. Sinauer, Sunderland, MA. Weir, B. S., and C. C. Cockerham. 1984. Estimating F statistics for the analysis of population structure. Evolution 38:1358-1370. Welsh, J., C. Petersen, and M. McCelland. 1991. Polymorphisms generated by arbitrary primed PCR in the mouse: application to strain identification and genetic mapping. Nucl. Acids Res. 19:303-306. Wild, A. 1994. A review of the biology and fisheries for yellowfin tuna, T. albacares, in the eastern Pacific Ocean. FAO Fish. Tech. Pap. 336(2):52-107. Williams, J. G, A. R. Kubelik, J. Livak, J. A. Rafalaski, and S. V. Tingey 1990. DNA polymorphisms amplified by arbitrary prim- ers are useful as genetic markers. Nucl. Acids Res. 18: 6531-6535. 778 Abstract— Lengths and ages of sword- fish (Xiphias gtadius) estimated from increments on otoliths of larvae col- lected in the Caribbean Sea, Florida Straits, and off the southeastern United States, indicated two growth phases. Larvae complete yolk and oil globule absorption 5 to 6 days after hatching (DAH). Larvae <13 mm preserved standard length (PSD grow slowly (-0.3 mm/d); larvae from 13 to 115 mm PSL grow rapidly (-6 mm/d). The accel- eration in growth rate at 13 days fol- lows an abrupt (within 3 days) change in diet, and in jaw and alimentary canal structure. The diet of swordfish larvae is limited. Larvae <8 mm PSL from the Caribbean, Gulf of Mexico, and off the southeastern United States eat exclusively copepods, primarily of one genus, Corycaeus. Larvae 9 to 11 mm eat copepods and chaetognaths; larvae >11 mm eat exclusively neustonic fish larvae. This diet indicates that young larvae <11 mm occupy the near-surface pelagia, whereas, older and longer larvae are neustonic. Spawning dates for larvae collected in various regions of the western North Atlantic, along with the abundance and spatial distribution of the youngest larvae, indicate that spawning peaks in three seasons and in five regions. Swordfish spawn in the Caribbean Sea, or possibly to the east, in winter, and in the western Gulf of Mexico in spring. Elsewhere swordfish spawn year-round, but spawning peaks in the spring in the north-central Gulf of Mexico, in the summer off southern Florida, and in the spring and early summer off the southeastern United States. The western Gulf Stream fron- tal zone is the focus of spawning off the southeastern coast of the United States, whereas spawning in the Gulf of Mexico seems to be focused in the vicinity of the Gulf Loop Current. Larvae may use the Gulf of Mexico and the outer con- tinental shelf off the east coast of the United States as nursery areas. Some larvae may be transported northward, but trans-Atlantic transport of larvae is unlikely. The early life history of swordfish (Xiphias gladius) in the western North Atlantic John Jeffrey Govoni Elisabeth H. Laban Jonathan A. Hare Center for Coastal Fisheries and Habitat Research National Oceanic and Atmospheric Administration 101 Rivers Island Road Beauloa North Carolina 28516-9722 E-mail address (for J, J, Govoni): Jeff Govoni(5>noaa gov Manuscript approved for publication 17 April 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Offfice. Fish. Bull. 101:778-789 (2003). Swordfish (Xiphias gladius) live in warm waters of the world's oceans, as well as in large enclosed basins such as the Caribbean and Mediterranean seas, and the Gulf of Mexico (Berkeley, 1983). Swordfish are highly migratory throughout their global range. The worldwide population structure, as currently understood, has at least three breeding units: Mediterranean, north- western Atlantic to the tropical South Atlantic, and Indo-Pacific (Kotoulas et al., 1995; Chow and Takeyama, 2000; Reeb et al., 2000). For the purpose of fishery management, the International Commission for the Conservation of Atlantic Tunas (ICCAT) recognizes only North Atlantic and South Atlan- tic stocks. Possible genetic exchange between eastern and western North Atlantic populations is incompletely documented. Swordfish reportedly spawn year- round in the western North Atlantic in different seasons and regions. Spawn- ing season and location has been inferred from the abundance of small larvae (Gorbunova, 1969; Richards and Potthoff, 1980; Potthoff and Kel- ley, 1982; Grail et al, 1983; Govoni et al., 2000), gonad maturation (LaMonte, 1944; Beckett, 1975), or oocyte cytol- ogy (Taylor and Murphy, 1992; Arocha, 1997; Arocha, 2002). The observation of live females with running eggs, hooked on long-lines, and followed to the fishing vessel by several smaller males (Lee^; Berkeley''^) corroborates spawning in some seasons and locations. Although gonad condition and oocyte status can indicate spawning season, the resolu- tion of spawning location can be am- biguous with these methods because mature gonads and hydrated oocytes can be found in several seasons within the range of these highly migratory fishes. The determination of age and the distribution of young larvae, along with realistic estimates of water veloc- ity and trajectory, help to resolve this ambiguity. Beyond spawning, the early life his- tory of swordfish in any ocean is incom- pletely described (Palko et al., 1981). Larvae undergo a stark change in physical appearance between ~8 and 13 mm preserved standard length (PSL), from a typical scombroid larval form to a juvenile istiophorid one (Collette et al., 1984). At this juncture in develop- ment, larvae develop characteristic preorbital, supraorbital, posttemporal, and preopercular spines; enlarged and spinous dorsal, ventral, and lateral scale anlagen; and a continuous long dorsal fin that extends along most of the dorsal aspect. Swordfish retain these larval characters until they are at least 188 mm PSL (Arata, 1954; Pot- thoff and Kelley, 1982), a size at which most fishes are considered juveniles. By using the ages and lengths of larvae hatched in the laboratory and reared through yolk and oil globule absorp- tion (Sanzo, 1910; Yasuda et al., 1978), along with length frequencies of larvae caught in the western North Atlantic, ' Lee, D. J. 1995. Personal commun. Southeast Fisheries Science Center, NMFS, 75 Virginia Beach Drive, Miami, FL 33149. '^ Berkeley, S. A. 1998. Personal commun. Long Marine Laboratory, Univ. California, Santa Cruz, 100 Shaffer Road, Santa Cruz, CA 95060. Govoni et al,: Early life history o\ Xiphias gladius in the western North Atlantic 779 Arata (1954) inferred the age and growth of larvae from 6 to 192 mm preserved total length (PTT). Aside from this effort, the age of larval swordfish has been undetermined and growth has not been described. Diets of lai-vae have been reported (Gorbunova, 1969), but the apparent transi- tion in diet has neither been detailed nor reconciled with changes in physical features and growth. Similarly, the vertical distribution of larvae has not been reconciled with their diet or growth. Most larvae are collected near the surface of the ocean, typically in neuston nets, but some larvae are collected in nets that sample below the surface (Govoni et al., 2000). In the present study, we resolve and summarize the early life history of swordfish in the western North Atlantic. We report estimated age, describe growth, relate growth to feeding, morphological features, and vertical distribution, and infer spawning time and location and the sources and fates of swordfish larvae. This study supplements that of Govoni et al. (2000) by providing the dimension of time, i.e. age of larvae, to the spatial distribution and possible transport of these larvae. Methods Collections of larvae Ichthyoplankton collections from cruises in 1989 (in the northeastern Caribbean about the Lesser Antilles), 1991 and 1997 (off the southeastern United States), and 2000 (in the Straits of Florida and off the southeastern United States), produced 63 larvae that were preserved in 95% ethanol (for examination of otolith microstructure). Sam- ples were collected either from the neuston (i.e. taken with a 1.0x0.5 m neuston net) or from depth intervals (i.e. taken with a 1-m MOCNESS [multiple opening and closing net and environmental sampling system] [Wiebe et al., 1976]). Larvae were measured for preserved standard length (PSL), the conventional length measure for larval fishes (Kendall et al, 1984), and lower-jaw-fork-length (LJFL), the measure in common use for juvenile and adult sword- fish (Megalofonou et al., 1995). Otolith excision and examination Of the 63 larvae collected, sagittae were found and success- fully excised from 37 larvae, lapilli from 32, and asterisci from six. Otoliths were mounted on glass slides and dried before examination. Broken sagittae and lapilli, and some large sagittae, were embedded in plastic, sectioned with a saw, and polished (Secor et al., 1991). Otolith growth increments were counted along the lon- gest axis of each sagitta and lapillus by using a compound light-transmission microscope; increments in asterisci were fewer and less defined and were not counted. Three blind counts were made by the same observer. Although increments were consistently visible on both sagittae and lapilli, counts from individual larvae were greater on sagittae (Student's Mest; P<0.001). Standardized counts (the standard deviate of each repeated count) on the right and left sagittae were not significantly different (nested ANOVA; P<0.05). Increment counts from either the left or right sagitta, decided by coin toss, were used for age and growth rate determination. The mean of three replicate counts was rounded to a whole number. Increment counts from sagittae were used to estimate larval age. Increments were assumed to form daily (Cam- pana, 2001). The core increment was assumed to form at hatching (Jones, 1986). The first increment outside of the core increment was counted as one. Age from hatching (AFH) was the number of increments counted from the core increment on sagittae. This definition differs from that of Prince et al. (1991) who counted the core increment as increment one for the istiophorid blue marlin (Makaira nigricans) and for a single larval swordfish that was 8.5 PSL. The radius of each otolith was measured by image analysis. Growth model The best empirical fit among a suite of regressions of estimated age (AFH) and length (PSL) — linear, polyno- mial, and piece-wise, and moving — was chosen to describe somatic growth (Forbes and Lopez, 1989; Hare and Cowen, 1995; Rogers et al., 2001). Criteria for best fit were the following: the interpretation of fit from graphical display; regression coefficients (r^), and dispersion or convergence of regression residuals. The model of best fit was also applied to the estimated age and lower jaw fork length (LJFL) to allow comparison with published accounts of juvenile swordfish growth. Diet Sixty-eight specimens from the present collections and from Govoni et al. (2000), all with undamaged alimentary canals, were examined for gut contents. These specimens were taken in the northeastern Caribbean (3 specimens), Gulf of Mexico (5), and off the southeastern coast of the United States (60). Food was identified to the lowest taxon possible following Govoni et al. (1983). Physical features of larvae Histological sections of three larvae, 21.5, 30.0, and 52.0 mm PSL, were cut as a preliminary aid to the loca- tion of otoliths within the cranium and to determine the histological constitution of the larval alimentary canal. Time and location of spawning Spawning dates were estimated from the ages of larvae (estimated from the growth model (AFH)), plus 3 days (the incubation period at 25°C for swordfish eggs given by Yasuda et al. [1978]). Spawning dates are thus days from fertilization (DFF). Spawning location was inferred by applying DFF to larval swordfish lengths reported in the present study, as well as lengths given in Govoni et al. (2000), taking into consideration the time that eggs and larvae were at large and adrift (DFF) and the location where 780 Fishery Bulletin 101(4) Figure 1 Otoliths of larval swordfish (Xiphiasgladius): (A) histological section through the braincase of a 21.0-mm larva; (B) left sagitta from a 5.4-inni-PSL larva; (C) histological section of a right sagitta of a 33.3-mm-PSL larva; and (D) transverse section of a right sagitta of a 47.5-mm-PSL larva ( C=core increment; l=first check; 2=second check; M=macula within the otolith vestibule; 0=otolith; x's (B-D) mark intervals of measurement). larvae were collected, and back-calculating the geographic origin of eggs with mean axial trajectories and velocities of water currents for the Yucatan, Gulf Loop, Florida Cur- rents, and the Gulf Stream (-1.5 m/s (Maul and Vukovich, 1993; Olson et al., 1994; Boicourt et al., 1998]) and the Caribbean Sea (-0.2 m/s [Mooers and Maul, 1998]). Results Otolith structure and increment counts Sagittae and lapilli were round, extremely small, and lacked rostra or sulci in larvae <5 mm PSL(Fig. 1, A and Bi. A rostrum developed on sagittae at -5.5 mm PSL (Fig. IC). Lapilli did not develop rostra, and remained symmetrical with growth. Two checks, distinct zones of irregular increment spacing and opacity, were evident on most sagittae (Fig. 1, B-D). The first check was evident at the third increment on all sagittae examined. The second check was found on sagit- tae from larvae >3.8 PSL but varied from the seventh to tenth increment. Growth model A piece-wise regression (Table 1; Equations 5-7) with two linear segments provided the best fit with biologically real- istic parameters. An assigned intercept of 3.2 mm PSL was used for the first segment; this value was obtained by adjust- ing the length at hatching with the scale given by Yasuda et al. ( 1978) and by accounting for shrinkage due to pres- ervation. Growth rate for the first segment was 0.3 mm/d and 5.9 mm/d PSL for the second segment (Fig. 2A). The intersection of the two linear segments was at an estimated age of 13.3 d AFH., 3 to 6 d after the observed second check. The PSL of larvae at the intersection was 11.0 mm. Growth rate in LJFL, also modeled with a piecewise regression, was 0.2 mm/d for the first segment (the upper and lower jaws of larvae <11 mm PSL are of equal length. Govoni et al : Early life history o\ Xiphias gladius in the western North Atlantic 781 Table 1 Summary of models evaluated for describing growth of larval swordfish. Models 1-6 were preserved standard length (PSD as a function of estimated age from hatching (AFH); model 7 was lower jaw fork length (LJFL) as a function of AFH. There were three model types; linear regression (models 1 and 2), 2"'' order polynomial (models 3 and 4), and piecewise regression (models 5-7). Y-intercepts (model parameter a) were estimated by the regression in Equations 1, 3, and 5 but were fixed at an observed length at hatching of 3.2 mm from Yasuda et al. (1978) in Equations 2, 4, 6, and 7 ("ns" denotes that a was not significantly different from an assigned a of 3.2 at a=0.05; * =y-intercept significantly different from 3.2 at a=0.05; na = not applicable; b and c are slopes; and d is the inflection point). Model 1. PSL = a + 6AFH 2. PSL = 3.2 +6AFH 3. PSL = a +6AFH+ fAFH2 4 . PSL = 3.2+6AFH + c AFH^ 5. PSL = a + 6AFH + c(AFH - d) x (AFH d) 6. PSL = 3.2 + 6AFH + c(AFH-d)x(AFH>d) 7. LJFL = 3.2 + 6AFH+C (AFH- d)x (AFH >d) Model parameters Model fit a b c d r2 Residuals -39.11* 4.15 na na 0.81 not normal 3.20 na 1.47 na na 0.81 not normal 11.31ns -2.77 0.22 na 0.92 normal 3.20 na -1.75 0.19 na 0.91 normal -2.22* 0.76 5.10 13.53 0.89 normal 3.20 na 0.26 5.60 13.29 0.89 normal 3.20 na 0.16 3.28 11.24 0.77 ns Table 2 Diet composition of 68 larval swordfish (Xiph ias gladius) in the western North Atlantic. % Frequency % total of occurrence number (among larvae (among all Diet item with food) food items) Copepodites and adult copepods (unidentifiable) 5 2 calanoids Eucalanus 2 1 cyclopoids Corycaeus spp. 59 74 Oithona spp. 4 2 Chaetognaths 2 1 Larval and juvenile fishes 33 16 Invertebrate eggs 2 2 Chyme 5 2 hence PSL=LJFL) and 3.4 mni/d for the second segment (Fig. 2B). The fit of the piecewise regression for PSL and LJFL was unchanged by inclusion or exclusion of the estimated age of the largest larva. Diet The diet of larvae is limited and transitional. Larvae <8.3 mm PSL ate copepods exclusively, primarily a single cyclopoid genus, Corycaeus spp., but also another cyclopoid, Oithona spp., and the calanoid Eucalanus (Table 2). Larvae 9.0 to 11.0 mm PSL ate copepods (Fig. 3A) and chaeto- 120 A © / e" / E / £ / E" 80 / m / •g / CO / CO CO • f • 1 40 /• a> CO /?* CD D. ^T* i-i-i^^ia — — en^co*'',*' 1 1 1 0 7 14 21 28 35 120 B e" E. , £ 80 • / O) c cu / ■=£ / £ / 5 / a • / S 40 S A^ _i •Z7 • J^^** ©rartBi*,* 0 ( ) 7 14 21 28 35 Estimated age from hatching (days) Figure 2 Growth of larval swordfish from the western North Atlantic: (A) estimated age and PSL; (B) estimated age and LJFL. 782 Fishery Bulletin 101(4) B ■-«»Si,~, <5 5 to 10 10 to IS Standard lengtti (mm) >1S Figure 3 Diet composition of larval swordfish from the western North Atlantic: (A) a 3.9-mm-PSL larva and cyclopoid copepods (genus Corycaeus) that were removed from the gut; (B) a 12.7-mm larva with a larval fish in its gut; and (C) the frequency of occurrence of copepods and larval fishes in the guts of larval swordfish. gnaths. Larvae > 1 1 .0 mm PSL ate almost exclusively larval and juvenile fishes (Fig. 3, B and C). Remnant jaws and heavy pigmentation of many of the fishes eaten, indicated that most were neustonic. One exocoetid was identified by intact pectoral fins and counts of vertebrae. Jaw and alimentary canal structure The structure of the alimentary canal and jaws changed concomitantly. The alimentary canal began to change from three segments (foregut, midgut and hindgut), typical of larval fishes (Govoni et al., 1986a), to four segments (esoph- agus, stomach, anterior intestine, and posterior intestine) between 9.0 and 12.0 mm PSL. Jaws change during this period from the beak-like jaws to the elongate rostral bill of the istiophorids (Fig. 3, A and B). Gastric glands were evident in the fundic region of the stomach (ventricili-gas- tric cecum), close to the junction with the esophagus in the 30.0-mm-PSL larva (Fig. 4, A and B). The pyloric region of the stomach (pars pylorica) was evident in the 21.5-mm- PSL larva and the 30.0-mm larva. Time and location of spawning Back-calculated spawning dates demonstrated year-round spawning and peaks in three seasons and five regions (Fig. 5, A-C). Larvae collected in the eastern Caribbean were spawned in the winter (northern hemisphere) only. Larvae collected in the western Gulf of Mexico were spawned in spring. In the north-central Gulf of Mexico, larvae were spawned in all seasons, but spawning peaked in spring. Off south Florida, larvae were spawned in all seasons, and spawning peaked peaked in spring. Lar- vae collected off the southeastern United States were spawned throughout the year whereas larvae collected in the north-central Gulf of Mexico and in southern Florida waters were spawned mostly in spring and early summer. Modes of the number of larvae collected and their esti- mated DFF advanced slightly in day of the year from the north-central Gulf of Mexico to off the southeastern United States. Larvae <10 DFF were collected both in the north- central Gulf of Mexico and off the southeastern United States (Fig. 5D), but not off South Florida. Govoni et al.: Early life history of Xiph/as gladius in the western North Atlantic 783 Figure 4 Photomicrographs of a 30.0-mm-PSL swordfish larva: (A) fundic stomach; (B) gastric glands in stomach mucosa (Gg=gastric glands; M=mucosa; L=stomach lumen with larval fish remains; E=eye of eaten larval fish). Discussion The first check on sagittae apparently corresponds with the completion of yolk and oil globule absorption and the beginning of feeding. Sanzo (1910) and Yasuda et al. (1978) reported hatching 3 days after fertilization when larvae were ~4 mm live total length (LTD, or 3.8 mm live stan- dard length (scaled from their drawings), and complete yolk and oil globule absorption 8 d after fertilization (or 5 DAH) when larvae were -5 mm LTL, or 4.3 mm live standard length. Larvae from the present material had completed yolk and oil absorption between 3.8 and 3.9 mm PSL and had the first check 3 increments after the core increment. Temperature and feeding influence the growth rate of fish larvae and their otoliths, but larvae are typically collected in water 25 ±1''C (Arata, 1954; Taning, 1955; Tibbo and Lauzier, 1969; Markle, 1975; Govoni et al., 2000), as were the larvae collected for age determination. This tempera- ture is common to the Gulf Stream and its progenitor cur- rents (Schmitz et al., 1993; Hitchcock et al., 1994), and is similar to the temperature used to rear larvae (Sanzo, 1910; Yasuda et al., 1978). The difference in length at complete yolk and oil absorption between Sanzo (1910) and Yasuda et al. (1978) and the present collections probably owes to shrinkage of larvae with death and preservation (e.g. Thei- lacker, 1980). The second check follows concomitant changes in diet and morphological features that take place between 8 and 13 mm PSL or from 7 to 11 DFH. An acceleration in so- matic growth follows the second check within a day or so. Young swordfish larvae eat copepods; older larvae other larval fishes. The most striking morphological change of larval swordfish is in the jaws. Swordfish larvae <13 mm SL have beak-like jaws that are typical of the larval scom- broid fishes (Collette et al., 1984), particularly those of the wahoo (Acanthocybium solandrl) and scaleless tuna (Gymnosarda unicolor); older larvae develop bill-like jaws with elongate rostral cartilages anterior of the premaxillar- ies and equally elongate mandibles (McGowan, 1988). The constitution of the alimentary canal changes as well. The development of a functional stomach with gastric glands in larval swordfish, which typically arises during the meta- morphosis of fishes (Govoni et al., 1986a), is evident in the larvae of other scombroid fishes where it is accompanied similarly by a change in diet from zooplankton to fish (Kaji et al., 1999; Shoji et al., 1999). A switch from zooplankton to fish is common among istiophorid larvae, but it is neither as exclusive nor abrupt (Voss, 1953; Lipskaya and Gorbu- nova, 1977) as it is with swordfish. Accelerated growth af- ter such a dietary shift is also a common trait of scombroid larvae (Shoji etal, 1999). Swordfish larvae grow rapidly, faster than other larval fishes with reportedly rapid growth. Growth rates reported in the present study are for larvae that have survived pre- dation and possibly variable feeding success; these rates do not necessarily represent average larval growth of the overall population. Growth rates of larvae >11 mm PSL (13 DAH), 5.6 mm/d, are nonetheless faster than the lar- val growth rates of other fast-growing larvae that survive in the sea, e.g. sablefish (Anoplopoma fimbria) (Boehlert and Yoklavich, 1985), and the oceanic-pelagic common dol- phinfish (Coryphaena hippurus) reared in the laboratory at high food densities without predation (Hassler and Hog- arth, 1977; Kraul, 1991). The growth rate of larval sword- fish <13 mm LJFL, 3 mm/d, is slower than the maximum (16 mm LJFL/d) and sustained (10 mm LJFL/d) growth 784 Fishery Bulletin 101(4) #h.n ^JT^ 200 Estimated Spawning date 300 400 78 Longitude (°W) 40 30- 20 10 0 16 Southeast US iifn iMilniiFl] [1 0 12- South Florida 0 60r 50- 40 30 20 10 0 25r 20 15 10 5 0 12 10 8 6 4 k Northcentral Gulf of Mexico iJIL.^ r Z c 3 Western Gulf of Mexico 0 Eastern Caribbean 0 100 200 300 400 Estimated spawning date Figure 5 Estimated spawning dates (day of the year) for swordfish larvae collected from the Caribbean Sea, the Gulf of Mexico, and off the southeast coast of the United States (hatched bars and circles depict winter in the north- ern hemisphere, filled bars and triangles spring-summer, and cross-hatched bars and squares summer-fall): (A) number of larvae and estimated spawning dates; (B) geographical distribution of larval collections; (C) number of larvae and estimated spawning date by region; (D) number of larvae by region and season. rate oflarval blue marlin over the first 100 days or <10G0 mm LJFL (Prince et al., 1991). Larval and juvenile swordfish from the western North Atlantic and Mediterranean exhibit four growth phases. Growth is linear for larvae <11 mm PSL, for larvae 11 to 115 PSL, and for juveniles 510 to 740 mm LJFL (Megalo- fonou et al., 1995). Growth becomes allometric for larger juveniles (Ehrhardt, 1992). Larvae <11 mm grow -0.1 mm LJFL/d. After the acceleration of growth, larvae in the western North Atlantic grow at -3 mm LJFL/d, whereas young juveniles in the Mediterranean grow at 23 mm LJFL/d (Megalofonou et al., 1995). Growth slows in older juveniles <250 mm LJFL to -2.5 mm LJFL/d (Ehrhardt, 1992). Adult growth (Berkeley and Houdc, 1983; Tserpes and Tsimenides, 1989) may constitute a fifth phase, as rec- ognized by Yabe et al. (1959) for Pacific swordfish. The limited diet oflarval swordfish is unusual; few larval fishes prey almost exclusively upon either copepods or lar- val fishes. Swordfish larvae 12.0 mm total length (TL) eat zooplankton, and larger larvae >12.0 mm TL eat other fish larvae (Gorbunova and Lipskaya, 1975), including conspe- cifics (Arata, 1954). Larval fishes as a whole are selective feeders; Corycaeus is selected by larval percoids in the Gulf of Mexico (Govoni et al., 1986b). Young larval istiophorids from the Florida Current eat primarily cyclopoid copepods of the genera Corycaeus, Farranula, and Oithona (Post et al, 1997), before they become piscivorous (Gorbunova and Lipskaya, 1975). Closely related genera of fishes exhibit Govoni et al.: Early life history o\ Xiphias gladius in the western North Atlantic 785 20 D 2()r 15 10 5 20|- 15 10 Winter spawning ai tru 20 40 no larvae 20 40 60 60 40 20- 15^ loi- 60 20| Spring/summer spawning ou Summer/fall spawning 40 - 20 - >i^ /_' •.■.1-.-.I 20r 10k 5^ 20 40 no larvae no larvae no larvae Estimated age from spawning (days) Figure 5 (continued) c CO en o Pz 60 O 0) IT) S 3 0) m cr u> different diets, even when occupying the same space (e.g. Govoni et al., 1983; 1986b). Larvae of the related istiophorids have limited diets, but these are not as exclusive or abruptly changing as that of swordfish. Diets of larvae examined in our study showed no evidence of cannibalism. The diet of larval swordfish helps to resolve their vertical distribution. Most larvae have been collected at the surface in neuston or dip nets (Taning, 1955; Yabe et al., 1959; Gor- bunova, 1969; Nishikawa and Ueyanagi, 1975), although some have been collected in plankton nets that fished principally below the surface (Grail et al., 1983; Govoni et al., 2000). The diet of swordfish larvae indicates that larvae <11 mm PSL may live in the near surface water, whereas larvae >11 mm are neustonic. Corycaeus is a common ne- ritic copepod of the Caribbean, the Florida Current, and the continental shelf off the southeast coast of the United States; Corycaeus is not neustonic (Owre and Foyo, 1967; 1972; Paffenhofer, 1983; 1985). That Corycaeus is eaten almost exclusively by young swordfish larvae implies that these larvae occupy the near-surface pelagia. Istiophorid larvae undertake dietary shifts (Voss, 1953; Gorbunova and Lipskaya, 1975; Lipskaya and Gorbunova, 1977) and changes in vertical distribution (Bartlett and Haedrich, 1968; Leis et al., 1987) that are similar to those of sword- fish larvae, but conflicting evidence exists for vertical dis- tribution of larval istiophorids. Gorbunova and Lipskaya (1975) implied that istiophorid larvae accumulate in sur- face waters during the day and disperse below the surface 786 Fishery Bulletin 101(4) at night, but Bartlett and Haedrich (1968) indicated the reverse. Large swordfish larvae are caught during the day and night in the neuston. The restricted diets of both large and small larvae implies little vertical movement. The overall modal increase in larval age from the Gulf of Mexico to the north indicates that spawning takes place in the north-central Gulf of Mexico and off the southeast coast of the United States and that there is possible northward transport. Estimated ages, along with velocities and trajec- tories of currents, indicate that larvae could be transported from considerable distances, but only if they remain within the axes of major currents. The smallest larva collected off the Carolinas was 3.9 mm PSL and had an estimated age of 7 d AFH (this specimen was previously reported as be- ing approximately 4 days old in Govoni et al. [2000]). With 3 days incubation at 25°C added to this estimated age, a swordfish egg and larva would be planktonic for 10 days. With a mean axial trajectory and velocity of the Florida Current and Gulf Stream of 1.5 m/s (Olson et al., 1994), a larva 3.9 mm PSL could be transported from as far away as 910 km, which could place the origin of this larva in the Straits of Florida, if its northward progress had not been checked by eddies so typical of the Gulf Loop Current (Maul and Vukovich, 1993) and Gulf Stream, particularly after the Gulf Stream exits the Straits of Florida (e.g. Lee et al., 1991; Govoni and Hare, 2001). Off the southeast coast of the United States, swordfish larvae aggregate in the west- ern Gulf Stream frontal zone (Govoni et al., 2000), where northward current velocities are considerably slower than 1.5 m/s (Marmorino et al., 1999). Because larvae reside primarily in the western Gulf Stream frontal zone where northward velocities are slower and where the front itself is so frequently distorted by meanders and eddies, it is unlikely, but not impossible, that a larva as young as 7 d APH collected off the Carolinas of the United States could have been transported from the Straits of Florida. The largest and oldest larva examined, one collected off South Carolina (Govoni and Hare, 2001), was 115 mm PSL and had an estimated age of 30 d AFH; with 3 days incuba- tion this fish could have been at large for 33 days and would have traveled 4290 km, given the mean axial trajectories and velocities of the Caribbean Sea (-0.2 m/s) and the Yucatan, Gulf Loop, and Florida Currents (-1.5 m/s). This calculation might place the spawning origin of this larva in the eastern Caribbean Sea or south of the Sargasso Sea if a direct, unchecked passage is assumed. Inference of the seasonality and geography of spawn- ing is limited and biased by the unsystematic temporal and spatial distribution of the present collections of larval swordfish and by uncertainties about the rate and trajec- tory of transport of eggs and larvae. Yet, taken as a whole, spawning dates, back-calculated from larvae collected in various regions of the western North Atlantic, and the abundance and spatial distribution of the youngest larvae indicate a spawning distribution with modes in three seasons and five regions. The western Gulf Stream frontal zone is the focus of spawning off the southeastern coast of the United States. Spawning in the Gulf of Mexico seems to be focused in the vicinity of the northern most arc of the Gulf Loop Current. Estimated spawning dates and the spatial distribution of young larvae offer an alternative to gonad condition and oocyte status as a means of resolving spawning season and location. Spawning season and location resolved in the present study corroborate the scenario recently proposed by Arocha (1997). Rather than the single breeding unit cur- rently recognized for the western North Atlantic by ICCAT, Chow and Takeyama (2000) and Arata ( 1997) proposed two spawning groups: one south of the Sargasso Sea and east of the Lesser Antilles, and the other in the Windward Pas- sage of the Antilles, the Yucatan Channel, and the Straits of Florida up to 35°N latitude. Accordingly, spawning begins in December south of the Sargasso Sea. Larvae from this spawning group transit into the Caribbean, are retained there by its anticyclonic circulation, and use the southeast- em Caribbean as a nursery area. Arocha (1997) implied that the second group spawns progressively later in the year and that larvae are transported with the Gulf Loop and Florida Currents, and the Gulf Stream. Arocha (1997) speculated that larvae and juveniles use the Gulf of Mexico and waters inshore of the Gulf Stream as a nursery area. Spawning dates and abundances of young larvae corroborate Arocha's (1997) proposed scenario for the seasonality and location of spawning and confirm spawning off the southeastern United States in the late spring and summer in the north- ern hermisphere. Spawning dates and abundance of young larvae also indicate the Gulf of Mexico as a nursery area. Further, large numbers of juveniles discarded from the long- line fishery prosecuted in the vicinity of the Charleston Gyre (Cramer, 2001) and the collection of larvae there (Govoni and Hare, 2001) indicate that the waters off the southeast- ern coast of the United States serve as a nursery area. Swordfish larvae are collected elsewhere in the western North Atlantic, although they are rarely caught north of Cape Hatteras, North Carolina (Tibbo and Lauzier, 1969). The trajectory of the Gulf Stream north of Cape Hatteras is convoluted and its velocity is slower, ~1 m/s (e.g. Bow- ers and Rossby, 1989; Flierl and Davis, 1993; Hare et al., 2002); the transit period of plankton from Cape Hatteras to the Azores is 120-300 days (Scheltema, 1971). Swordfish spawned in the western North Atlantic would be juvenile fish by the time they reached the eastern North Atlantic. North of Cape Hatteras, the Gulf Stream sheds eddies into the Sargasso Sea (McGuillicuddy et al., 1998); thus, the general location for juvenile swordfish that are spawned and not retained in the western North Atlantic may well be the central Atlantic and Sargasso Sea. Swordfish are multiple spawners (Arocha, 2002) and adults may move and spawn among regions of the western North Atlantic. Movement of spawning adults, along with transport of larvae, may result in the genetically well-mixed population of the western North Atlantic ( Alvarado Bremer et al., 1995a). There is apparently no genetic exchange be- tween northwestern Atlantic and Mediterranean reproduc- tive populations (Alvarado Bremmer et al., 1995b; Chow and Takeyama, 2000). Transoceanic migration of adult fish is possible, but cross-Atlantic transfer of swordfish larvae is not likely. Swordfish lars'ae, collected principally in water >25°C (Govoni et al., 2000), probably perish as Gulf Stream water cools when it traverses the northern western North Govoni et al.: Early life history of Xiphias gladius in the western North Atlantic 787 Atlantic (Scheltema, 1971; Cowen et al., 2000; Gaylord and Gaines, 2000). Larvae that are not retained by eddies of the Gulf Loop Current and the Gulf Stream may be diverted to the central Atlantic (McGuillicuddy et al., 1998) and prob- ably do not transit to the eastern North Atlantic. Acknowledgments We thank L. R. Settle [NOAA - NOS, Center for Fisheries and Habitat Research (CCFHR)] for the collection of some of the swordfish specimens reported in our study. M. M. Leiby (Florida Department of Natural Resources) loaned SEAMAP collections. F. Arocha (Universidad de Oriente, Venezuela), S. A. Berkeley (Oregon State University), and D. L. Lee and J. L. Cramer (NOAA- Fisheries, Southeast Fish- eries Science Center) provided invaluable counsel on sword- fish biology. D. W. Ahrenholz and M. H. Prager (CCFHR, National Marine Fisheries Service, NOAA), W. J. Richards (Fisheries, Southeast Fisheries Science Center, NOAA), and F Arocha provided reviews of the manuscript. Literature cited Alvarado Bremer, J. R., A. J. Baker, and J. Mejuto. 1995a. Mitochondrial DNA control region sequences indi- cate extensive mixing of swordfish (Xiphias gladius) in the Atlantic Ocean. Can. J. Fish. Aquat. Sci. 52:1720-1732. Alvarado Bremer, J. R., J. Mejuto, T. W. Greig, and B. Ely. 1995b. Global population structure of the swordfish (Xiphias gladius) as revealed by analysis of the mitochondrial DNA control region. J. Exp. Mar Biol, Ecol. 197:295-310. 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. Arocha, F. 1997. The reproductive dynamics of swordfish Xiphias gla- dius L. and management implications in the northwestern Atlantic. Ph.D. diss., 350 p. Univ Miami, Miami, FL. 2002. Oocyte development and maturity classification of swordfish from the north-western Atlantic. J. Fish Biol. 60:13-27. Bartlett, M. R., and R. L. Haedrich. 1968. Neuston nets and South Atlantic larval blue marlin {Makaira nigricans). Copeia 1968:469-474. Beckett, J. S. 1975. Biology o^ sworAfish, Xiphias gladius L., in the north- west Atlantic ocean. In Proceedings of the international billfish symposium Kailua-Kona, Hawaii, 9-12 August 1972. part 1: Report of the symposium (R. S. Shomura and F Williams, eds.), p. 103-106. NOAA Tech. Rep. NMFS- SSRF-675. Berkeley, S. A. 1983. Atlantic swordfish stock structure data and sugges- tions for its interpretation. Int. Comm. Conserv. Atl.Tuna, Collect. Vol. Sci. Pap. 18:839-845. Berkeley, S. A., and E. D. Houde. 1983. Age determination of broadbill swordfish, Xiphias gladius, from the Straits of Florida, using anal fin spine sections. In Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, billfishes. and sharks (E. D. Prince and L. M. Pulos, eds.), p. 137-143. NOAA Tech. Rep. NMFS-SSRF-8. Boehlert, G. W., and M. M. Yoklavich. 1985. Larval and juvenile growth of sablefish, Anoplopoma fimbria, as determined from otolith increments. Fish. Bull. 83:475-481. Boicourt, W. C, W. J. Wiseman Jr, A. Valle-Levinson, and L. P. Atkinson. 1998. Continental shelf of the southeastern United States and the Gulf of Mexico: in the shadow of the western bound- ary current. In The sea (A. R. Robinson and K. H. Brink, eds.), vol. 11, p. 135-183. John Wiley & Sons, New York, NY. Bowers, A. S., and T Rossby. 1989. Evidence of cross-frontal exchange processes in the Gulf Stream based on isopycnal RAFOS float data. J. Phys. Oceanogr. 19:1177-1190. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish Biol. 59:197-242. Chow, S., and H. Takeyama. 2000. Nuclear and mitochondrial DNA analyses reveal four genetically separated breeding units of the swordfish. J. Fish Biol. 56:1087-1098. Collette, B. B., T. Potthoff W. J. Richards, S. Ueyanagi, J. L. Russo, and Y. Nishikawa. 1984. Scombroidei: development and relationships. In Ontogeny and systematics of fishes (H. G. Moser, W. J. Richards, D. M. Cohen, M. P Fahay, A. W. Kendall, and S. L. Richardson, eds), p. 591-620. Am. Soc. Ichthyol. Herpetol. Spec. Publ. 1. Cowen, R. K., K. M. M. Lwiza, S. Sponaugle, C. B. Paris, and D. B. Olson. 2000. Connectivity of marine populations: open or closed? Science 287:857-859. Cramer, J. 2001. Geographic distribution of longline effort and sword- fish discard rates in the Straits of Florida and oceanic waters of the continental shelf slope, and Blake Plateau off Georgia and the Carolinas from 1991 to 1995. In Island in the stream: the ecology and fisheries of the Charleston Bump (G. R. Sedberry, ed.), p. 97-103. Am. Fish. Soc. Symp. 25. Ehrhardt, N. M. 1992. Age and growth of swordfish, Xiphias gladius, in the northwestern Atlantic. Bull. Mar Sci. 50:292-301. Flierl, G. R., and C. S. Davis. 1993. Biological effects of Gulf Stream meandering. J. Mar Res. 51:529-560. Forbes, T. L., and G. R. Lopez. 1989. Determination of critical periods in ontogenetic trajectories. Funct. Ecol. 3:525-632. Gaylord, B., and S. D. Gaines. 2000. Temperature or transport? Range limits in marine species mediated solely by flow. Am. Nat. 155:769-789. Gorbunova, N. N. 1969. Breeding grounds and food of the larvae of the sword- fish Xiphias gladius Linneus (Pisces, Xiphiidae). Probl. Ichthyol. 9:375-387. Gorbunova, N. N., and N. Ya. Lipskaya. 1975. Feeding of larvae of the blue marlin, Makaira nigri- cans (Pisces, Istiophoridae). J. Ichthyol. 15:95-101. Govoni, J. J., G. W. Boehlert, and Y. Watanabe. 1986a. The physiology of digestion in fish larvae. Environ. Biol. Fishes 16:59-77. 788 Fishery Bulletin 101(4) Govoni, J. J., and J. A. Hare. 2001. The Charleston Gyre as spawning and larval nursery habitat for fishes. In Island in the stream: the ecology and fisheries of the Charleston Bump (G. R. Sedberry. ed.), p. 123-136. Am. Fish. Soc. Symp. 25. Govoni. J. J., D. E. Hoss, and A. J Chester 1983. Comparative feeding of three species of larval fishes in the northern Gulf of Mexico: Breuoortia patronus, Leiosto- mus xanthurus, and Micropogonias undulati/s. Mar. Ecol. Prog. Ser 13:189-199. Govoni, J. J., P. B. Ortner, F. Al-Yamani, and L. Hill. 1986b. Selective feeding of spot, Leiostomus xanturus, and Atlantic croaker, Micropogonias undulatus, larvae in the northern Gulf of Mexico. Mar. Ecol. Prog. Ser 28: 175-183. Govoni, J. J., B. W. Stender, and 0. Pashuk. 2000. Distribution of larval swordfish, Xtphias gla- dius, and probable spawning off the southeastern United States. Fish. Bull. 98:64-74. Grail, C, D. R de Sylva, and E. D. Houde. 1983. Distribution, relative abundance, and seasonality of swordfish larvae. Trans. Am. Fish. Soc. 112:235-246. Hare, J. A., J. H. Churchill, R. K. Cowen, T J. Berger, R C. CorniUon, P Dragos, S. M. Glenn, J. J. Govoni, and T N. Lee. 2002. Routes and rates of larval fish transport from the southeast to the northeast United States continental shelf Limnol. Oceangr. 47: 1774-1789. Hare, J. A., and R. K. Cowen. 1995. Effects of age, growth rate, and ontogeny on the otolith size-fish size relationship in bluefish, Pomatomus sattatrix, and the implications for back-calculation of size in fish early life history stages. Can. J. Fish. Aquat. Sci. 52:1909-1922. Hassler, W. W., and W. T Hogarth. 1977. The growth and culture of dolphin, Coryphacna hip- purus, in North Carolina. Aquaculture 12:115-122. 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. Jones, C. 1986. Determining age of larval fish with the otolith incre- ment technique. Fish. Bull. 84:91-104. Kaji, T, M. Tanaka, M. Oka, H. Takeuchi, S. Ohsumi, K. Teruya, and J. Hirokawa. 1999. Growth and morphological development of laboratory- reared yellowfin tuna Thunnus alhacares larvae and early juveniles, with special emphasis on the digestive system. Fish. Sci. 65:700-707. Kendall, A. W., Jr, E. H. Ahlstrom, and H. G. Moser. 1984. Early life stages of fishes and their characters. In Ontogeny and systematics of fishes (H. G. Moser, W. J. Richards, D. M. Cohen, M. R Fahay, A. W. Kendall, and S. L. Richardson, eds.), p. 11-22. Am. Soc. Ichthyol. Herpetol., Spec.Publ. 1. Kotoulas, G., A. Magoulas, N. Tsimenides, and E. Zouros. 1995. Marked mitochondril DNA differences between Mediterranean and Atlantic populations of the swordfish, Xiphiasgladius. Mol. Ecol. 4:47.3-481. Kraul, S. 1991. Larviculture of the mahimahi, Coryphaena hippurus, in Hawaii, USA. J. World Aquacult. Soc. 24:410-421. LaMonte, F. 1944. Note on breeding grounds of blue marlin and sword- fish o(T Cuba. Copeia 4:258. Lee, T. N., J. A. Voder, and L. P. Atkinson. 1991. Gulf Stream frontal eddy influence on productivity of the southeast U.S. continental shelf J. Geophys. Res. 96: 22,191-22,205. Leis, J. M., B. Goldman, and S. Ueyanagi. 1987. Distribution and abundance of billfish larvae (Pisces: Istiophoridae) in the Great Barrier Reef lagoon and Coral Sea near Lizard Island, Australia. Fish Bull. 85:757-766. Lipskaya, N. Ya., and N. N. Gorbunova. 1977. Feeding of sailfish larvae. Oceanology 17:340-344. Markle, G. E. 1975. Distribution of larval swordfish in the Northwest Atlantic Ocean. In Proceedings of the international billfish symposium Kailua-Kona, Hawaii, 9-12 August 1972, part 1: Report of the symposium (R. S. Shomura and F Williams, eds. ), p. 252-260" NOAA Tech. Rept. NMFS-SSRF-675. Marmorino, G. O., D. R. Lyzenga, and J. A. C. Kaiser 1999. Comparison of airborne synthetic aperture radar imagery with in situ surface-slope measurements across Gulf Stream slicks and a convergent front. J. Geophys. Res. 104:1405-1422. Maul, G. A., and F M. Vukovich. 1993. The relationship between variations in the Gulf of Mexico Loop Current and Straits of Florida volume transport. J. Phys. Oceanogr. 23:785-796. McGowan, C. 1988. Differential development of the rostrum and mandible of the swordfish iXiphias gladius) during ontogeny and its possible functional significance. Can. J. Zool. 66:496-503. McGuillicuddy, D. J., A. R. Robinson, D. A. Siegel, H. J. Jannasch, R. Johnson, T. D. Dickey, J. McNiel, A. F. Michaels, and A. H. Knap. 1998. Influence of mesoscale eddies on new production in the Sargasso Sea. Nature 395:263-266. Megalofonou, P., J. M. Dean, G. DeMetrio, C. Wilson, and S. Berkeley. 1995. Age and growth of juvenile swordfish, Xiphias gladius Linnaeus, from the Meditteranean Sea. J. Exp. Mar Biol. Ecol. 188:79-88. Mooers, C. N. K., and G. A. Maul. 1998. Intra-Americas sea circulation: coastal segment (3,W). In The sea (A. R. Robinson and K. H. Brink, eds.), vol. 11, p. 183-208. John Wiley & Sons, New York, NY. Nishikawa, Y., and S. Ueyanagi. 1975. The distribution of the larvae of swordfish, Xiphias gladius, in the Indian and Pacific oceans. In Proceedings of the international billfish symposium Kailua-Kona, Hawaii, 9-12 August 1972, part 1: Report of the symposium (R. S. Shomura and F Williams, eds.), p. 261-264. NOAA Tech. Rep. NMFS-SSRF-675. Olson, D. B., G. L. Hitchcock, A. J. Mariano, C. J. Ashjian, G. Peng, R. W. Nero, and G. R Podesta. 1994. Life on the edge: marine life and fronts. Ocean- ography 7:52-60. Owre, H. B., and M. Foyo. 1967. Copepods of the Florida Current. Fauna Caribaea 1, Crustacea, part 1: Copepoda, 137 p. Inst. Mar Sci., Univ. of Miami, Miami, FL. 1972. Studies on Caribbean zooplankton. Description of the program and results of the first cruise. Bull. Mar Sci. 22: 483-521. Paffenhofer, G.-A. 1983. Vertical zooplankton distribution on the northeast- ern Florida shelf and its relation to temperature and food abundance. J. Plankton Res. 5:15-33. Govoni et a\ Early life history oi Xiphias gladius in the western North Atlantic 789 1985. The abundance and distribution of zooplankton on the southeastern shelf of the United States. In Oceanography of the southeastern U.S. continental shelf (L. P. Atkinson, D. W. Menzel, andK.A. Bush, eds.), p. 104-117. Am. Geo- physical Union, Washington, DC. Paiko, B. J., G. L. Beardsley and W. J. Richards. 1981. Synopsis of the biology of the swordfish, Xip/iia.s g/a- dius Linnaeus, 21 p. NOAA Tech. Rep. NMFS Circ. 441. 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 ilstiophorus platypterus) and swordfish {Xiphias gladius). Bull. Mar Sci. 60:1026-1034. Potthoff, T, and S. Kelley 1982. Development and structure of the vertebral column, fins and fin supports, branchiostegal rays and squamation in the swordfish A7p/uas g/arfii/s. Fish. Bull. 80:161-186. 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 from otolith microstructure. Fish. Bull. 89:441-459. Reeb, C. A., L. Arcangeli, and B. A. Block. 2000. Structure and migration corridors in Pacific popu- lations of the swordfish Xiphius gladius, as inferred through analyses of mitochondrial DNA. Mar. Biol. 136: 1123-1131. 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. Rogers, J. S., J. A. Hare, and D. G. Lindquist. 2001. Otolith record of age, growth, and ontogeny in larval and pelagic juvenile Stephanotepis hispidus (Pisces: Monacanthidae). Mar. Biol. 138:945-953. Sanzo, L. 1910. Uovoe larva di Pesce-spada (A'lp/iiasg/adii/s L.). Riv. Mens. Pesca Idrobiol. 12:206-209. Scheltema, R. S. 1971. The dispersal of the larvae of shoal-water benthic invertebrate species over long distances by ocean currents. In Fourth European marine biology symposium (D. J. Crisp, ed.), p. 7-28. Cambridge Univ. Press, Cambridge, UK. Schmitz, W. J., J. R. Luyten, and R. W. Schmitt. 1993. On the Florida Current T/S envelope. Bull. Man Sci. 53:1048-1065. Secor, D. H., J. M. Dean, and E. H. Laban. 1991. Manual for otolith removal and preparation for micro- structural examination, 85 p. Electric Power Research Institute and the Belle W. Baruch Institute for Marine Biology and Coastal Research, Univ. South Carolina, Columbia, SC. Shoji, J., T. Maehara, and M. Tanaka. 1999. Short-term occurrence and rapid growth of Spanish mackerel larvae in the central waters of the Seto Inland Sea, Japan. Fish. Sci. 65:68-72. Taning, A. 1955. On the breeding areas of the swordfish iXiphias). Deep-Sea Res. 3(suppl.):438-450. Taylor, R. G., and M. D. Murphy 1992. Reproductive biology of the s-w or Af\s\\ Xiphias gladius in the Straits of Florida and adjacent waters. Fish. Bull. 90:809-816. Theilacker, G. H. 1980. Changes in body measurements of larval northern anchovy, Engraulis mordax, and other fishes due to han- dling and preservation. Fish. Bull 78:685-692. Tibbo, S. N., and L. M. Lauzier. 1969. Larval swordfish (Xiphias gladius) from three locali- ties in the western Atlantic. J. Fish. Res. Board Can. 26: 3248-3251. Tserpes, G., and N. Tsimenides. 1989. Age determination and growth of swordfish Xiphias gladius L., 1958 in the Aegean Sea. Fish Res. 8: 159-168. Voss, G. L. 1953. A contribution to the life history and biology of the sailfish, Istiophorus americanus Cuv. and Val., in Florida waters. Bull. Mar. Sci. Gulf Caribb. 3:206-240. 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, Xiphias gladius 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. 790 Abstract— Although subsampHng is a common method for describing the composition of large and diverse trawl catches, the accuracy of these tech- niques is often unknown. We deter- mined the sampling errors generated from estimating the percentage of the total number of species recorded in catches, as well as the abundance of each species, at each increase in the proportion of the sorted catch. We completely partitioned twenty prawn trawl catches from tropical northern Australia into subsamples of about 10 kg each. All subsamples were then sorted, and species numbers recorded. Catch weights ranged from 71 to 445 kg, and the number offish species in trawls ranged from 60 to 138, and invertebrate species from 18 to 63. Almost 70% of the species recorded in catches were "rare" in subsamples (less than one individual per 10 kg subsample or less than one in every 389 individuals). A matrix was used to show the in- crease in the total number of species that were recorded in each catch as the percentage of the sorted catch increased. Simulation modelling showed that sort- ing small subsamples (about 10% of catch weights) identified about 50% of the total number of species caught in a trawl. Larger subsamples (509c of catch weight on average) identified about 80% of the total species caught in a trawl. The accuracy of estimating the abun- dance of each species also increased with increasing subsample size. For the "rare" species, sampling error was around 80% after sorting 10% of catch weight and was just less than 50% after 40% of catch weight had been sorted. For the "abundant" species (five or more individuals per 10 kg subsample or five or more in every 389 individu- als), sampling error was around 25% after sorting 10% of catch weight, but was reduced to around 10% after 40% of catch weight had been sorted. Does the size of subsamples taken from multispecies trawl catches affect estimates of catch composition and abundance? Donald S. Heales David T. Brewer CSIRO Marine Research 233 Middle St Cleveland, Queensland 4163, Australia E mail address (for D S Heales) don.heales@marine csiro.au You-Gan Wang Dept of Biostatistics Harvard University Boston, Massachusetts 02115 Peter N. Jones CSIRO Mathematical and Information Sciences 233 Middle St., Cleveland, Queensland 4163, Australia Manuscript approved for publication 9 March 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish Bull. 101:790-799 (2003). Concerns are held worldwide regarding the sustainability of bycatch species taken in trawls, particularly prawn trawls. Under the voluntary FAO Code for Responsible Fisheries, managers are required to "take measures to conserve target species, associated or dependent species and nontarget species and their environment" (FAO, 1995). An essen- tial part of this process is the accurate monitoring of population sizes and structures. With large trawl catches, subsam- pHng is often the only cost-effective or feasible way to describe the bycatch composition. How well these subsam- ples represent the total catch depends on how diverse the catch is, how well the catch is mixed before the subsam- ples are taken, and what proportion of the catch is taken as a subsample. There is a large literature on sub- sampling theory for terrestial insect studies (Van Ark and Meiswinkel, 1992), aquatic macroinvcrtcbrate stud- ies (Vinson, 1996; Walsh, 1997), and marine ecological studies (Andrew and Mapstone, 1987). However, in most of these studies, samples of very small animals collected in the field can be resuspended in fluid and mixed evenly in the laboratory before the subsam- ples are taken. In fisheries, in direct contrast, large catches are extremely difficult to manipulate and redistrib- ute evenly before subsampling. A few fisheries studies have examined the impact of subsampling on estimates of the abundance and different size ranges of one or a few dominant species. For ex- ample, in the Crangon trawl fisheries in Belgian waters, sampling strategy had only a minor effect on the reliability of estimates of size selectivity for the tar- geted shrimp (Polet and Redant, 1999). In UK waters, subsampling trawled fish (both target species and discards) from either the sorting conveyor or the pound made no difference to catch composition estimates {Tamsett et al., 1999). However, in tropical trawl fisher- ies, over one hundred species can be recorded in a single catch. Under ESD (ecological sustainable development) guidelines, all these species (both tar- get and bycatch) are equally important but there has been very little research on subsampling techniques applicable to such diverse catches. A recent study in Australia's Northern Prawn Fishery (NPF) examined the accuracy of sub- sampling from large, diverse catches of fish and invertebrates (Heales et al., 2000). For most of the "abundant" Heales et al : Effect of size of subsamples on estimates of catchi composition and abundance 791 species, their position on trawler sorting trays from which the bycatch subsamples were collected, had little effect on the ac- curacy in representing the catches. Although the accuracy of subsampling is a general problem for all multispecies fisheries, few other studies have been pub- lished on the topic. Reliable techniques for subsampling are needed, however, espe- cially with the demands that bycatch spe- cies, as well as the target species, should be ecologically sustainable. We describe here research done in Australia's NPF but the results and methods are applicable to many sampling problems in fisheries. The NPF is a large tropical trawl fishery that extends from Cape York in Queensland to Cape Londonderry in Western Australia. In addition to catch- ing penaeid prawns (e.g. 8531 metric tons (t) in 1997-98, Taylor and Die, 1999), its bycatch component is estimated at over 38,000 t a year (Pender et al., 1992) or more than 80% of the total catch in the tiger prawn fishery (Brewer et al., 1998). The neighboring Torres Straits Prawn Fishery (TSPF) also has an estimated an- nual bycatch of 4800 t (Williams, 1985). The NPF has a management require- ment to assess the impact of trawling on nontarget species. The bycatch of both these prawn trawl fisheries (NPF and TSPF) is very diverse. Ramm et al. (1990) recorded 115 fish taxa in their study of NPF waters, and Brewer et al. ( 1998) recorded over 250 species from one area of the NPF. At least 390 fish species, 234 invertebrate taxa, and 43 elas- mobranch species have been recorded in a current bycatch project in the NPF (Stobutzki et al., 2000). Despite the lack of knowledge on the ability of sub- samples to accurately represent such diverse catches, many studies of trawl communities in Northern Australia have used subsampling techniques to estimate catch rates. In two bycatch studies of the NPF, the smaller catches were entirely sorted and the larger catches subsampled (Poiner and Harris, 1986; Harris and Poiner, 1991). Trawl catches in another bycatch study were spread evenly over the sorting tray and a visually estimated fraction of the catch was subsampled (Ramm et al., 1990). Other work- ers simply subsampled without confirming the accuracy of their subsampling techniques (Blaber et al., 1990 and 1994; Martin et al, 1995; Brewer et al, 1998; Wassenberg etal., 1998). A lack of knowledge of trawl impacts on nontarget spe- cies has led to the present CSIRO study that describes the bycatch from the NPF and provides a framework for any future bycatch monitoring program. As part of that study, we made the first assessment of the accuracy of subsam- pling over a range of subsample sizes as a tool for esti- mating the total catch composition of a large multispecies fishery. 129" 134 139 Stu(jy area \'L J Australia 129" 134 139" 144" Figure 1 Map of the Torres Straits Fishery (area 1 — broken line) and Northern Prawn Fishery (unbroken line), Australia, showing the managed area of the fisheries and the regions where samples were collected. Regional codes are 1 = Torres Straits, 2 = Weipa, 3 = east of Mornington Island, 4 = north of Mornington Island, 5 = west of Mornington Island, 6 = east of Vanderlin Islands, 7 = south of Groote Eylandt, 8 = north of Groote Eylandt, 9 = Melville Island. Materials and methods Data were collected from a series of 14 trawl samples taken during two research cruises of the RV Southern Surveyor. These samples were collected from one region of the Torres Straits Prawn Fishery (TSPF), and from eight of the major tiger prawn (Penaeus esculentus and P. semisulcatus) fish- ing regions of the Northern Prawn Fishery (NPF), (namely Weipa, east of Mornington Island, north of Mornington Island, west of Mornington Island, north of Vanderlin Islands, south of Groote Eylandt, north of Groote Eylandt, and Melville Island) (Fig. 1). All trawls were undertaken in either late summer 1997 (February-March, the end of the wet season) or in mid-spring 1997 (September-October, the dry season). We used a single 14-fathom Florida-Flyer prawn trawl net so that the data would be comparable with data from either of the two nets used by the twin-rigged commercial NPF vessels in the tiger prawn fishery. All trawls were done at night, again to be comparable with the fishery Duration of trawls ranged from 1 to 3 hours (Table 1), and depths ranged from 23 to 42.3 m. A further six trawl catches were sampled by a scientific observer on board commercial NPF vessels fishing for ti- ger prawns north of Mornington Island in late May 1997, and north of Groote Eylandt in late September 1997. Each trawl sample consisted of the entire catch from one of the two 14-fathom Florida-Flyer prawn trawl nets used by these vessels. All trawls were done at night. Their duration 792 Fishery Bulletin 101(4) Table 1 Summary of catch data for 20 entirely sorted trawls froi n the Northern Prawn Fishery and Torres Straits Prawn Fishery. 'Bycatch individuals" refers to the total number of bycatch animals (fish and invertebrates) in each trawl. "Fish species" refers to the total | number offish species recorded in the bycatch of each trawl, (n) is the total number Duration of Start Catch Bycatch Fish Invertebrate All trawl time of weight Subsamples individuals species species species Region (h) trawls (kg) in) in) In) (;i) Ul) Torres Straits 1.2 0400 94 9 4151 74 21 95 Weipa 2.7 0220 166 16 4911 92 36 128 East of Mornington Island 1.0 0415 274 24 7313 90 18 108 East of Mornington Island 2.9 0250 87 8 2019 77 45 122 North of Mornington Island 3.3 1845 182 16 9762 101 36 137 North of Mornington Island 3.0 2215 194 17 11015 94 42 136 North of Mornington Island 3.2 1840 445 36 13,826 114 52 166 North of Mornington Island 2.0 0350 71 7 1771 60 30 90 North of Mornington Island 2.2 0330 100 10 5067 77 34 111 West of Mornington Island 1.6 0415 174 16 5976 100 21 121 West of Mornington Island 1.7 0400 269 26 6967 87 24 111 North of Vanderlin Islands 2.7 0310 156 15 7182 71 19 90 South of Groote Eylandt 2.0 0345 315 27 23,751 105 25 130 South of Groote Eylandt 2.7 0250 165 17 3856 64 25 89 North of Groote Eylandt 2.5 2230 147 12 5558 68 28 96 North of Groote Eylandt 3.0 0215 85 9 2792 60 34 94 North of Groote Eylandt 3.5 1830 158 13 5664 89 43 132 North of Groote Eylandt 3.5 2215 169 15 8289 96 63 159 North of Groote Eylandt 3.5 2215 189 16 5748 96 43 139 Melville Island 3.0 0130 170 14 4635 60 30 90 Total 3610 323 140,253 ranged from 3 to 3.5 hours (Table 1), and depths ranged from 29 to 41 m. Sample collection On the research vessel, catches were spilled from the codend onto the flat deck (equivalent to the sorting tray on commercial vessels). The entire catch of each trawl was progressively partitioned by shovelling the catch into con- secutively numbered boxes (subsample replicates), each of about 10 kg (according to the methods described by Heales et al., 2000). Partitioning of the catch began at the outer edges and continued in a clockwise direction, and sub- samples were taken from each of four major compass direc- tions: north, east, south, and west until the entire catch had been collected in successively numbered boxes. The direc- tion of the ship's bow was always designated as north. Samples on the research vessel were sorted immediately. Fish and invertebrates were identified to the lowest taxo- nomic level possible (mostly species). Where this was not possible, the data were grouped to genus and in a few cases to family. Total numbers and weights were recorded for each species in each subsample and entered directly into a relational database. On the commercial vessels, the catches were spilled onto the sorting tray and the commercial-size prawns were removed. The bycatch would then normally move down a trash chute and spill overboard. However, to sample a catch, the trash chute was diverted so that all the bycatch was collected in consecutively numbered boxes (subsample replicates) each of approximately 10 kg. All samples collected from commercial vessels were frozen on board and transported to the laboratory for sub- sequent sorting, identification, and data entry according to the methods described above. Although most bycatch species were identified to spe- cies level, some could only be identified to genus and in a few cases to family. In order to be consistent in terms used throughout this study, we use the term species (plural form) even when referring to multispecies groups. The methods used to collect subsamples on both the re- search and commercial vessels differed only in the position from which subsamples were taken (see earlier "Material and methods" section). However, a previously published study (Heales et al., 2000) showed that the majority of the "abundant" bycatch species were evenly distributed throughout the catch. Consequently, for all analyses we combined the 14 catches collected from the research vessel with the six catches from commercial vessels. Data analysis Abundance groupings Within each catch, there was a large range of species (both fish and invertebrates) and they Heales et al : Effect of size of subsamples on estimates of catcfi composition and abundance. 793 occurred at many different levels of relative abundance. To obtain an overview of how rarely, or how frequently, differ- ent species occurred in catches, we reduced each occurrence of a species in a catch to an index of relative abundance. We concentrated solely on determining the accuracy of taking different size subsamples in representing the range of rela- tive abundances (from very low to very high) of the species in these catches. The relative abundance indices were based on the av- erage number of individuals of a given species that were recorded in a standardized 10-kg subsaniple taken from that catch. To generate this index, we used the following equation; n = 10 X (TotNum I Weight). (1) where n = the mean number of individuals of a given spe- cies per 10-kg subsample; TotNum = the total number of individuals of that species in the whole catch; and Weight = the total weight of the catch in kg. We derived a separate index of abundance for each species in every catch where it was recorded. Thus, a species that occurred in all 20 catches would have 20 different abun- dance indices in the analysis. To highlight the differences in distribution between the two extremes of the "rare" species and the "abundant" species when estimating catch composition, we grouped the indices of abundance into 11 categories, ranging from less than one individual per 10-kg subsample, up to 10 or more individuals per subsample. Species with abundance indices of less than one individual per 10-kg subsample were classed as "rare"; those with one to less than five individuals were classed as "common"; and those with five or more individuals per 10-kg subsample were classed as "abundant." For example, the common ponyfish (Leiognathus moreto- niensi), was classed as "abundant" in 11 of the 20 catches, as "common" in eight catches, and "rare" in one catch. Because a species could have different abundance indices in each catch, individual species are not referred to by name in the results. Instead, we refer to the occurrence of each species in a catch, as one case of some relative abundance index that was recorded in that catch (i.e. one "case" of species by trawl abundance). The relativefrequency of all the cases (i.e. abun- dance indices) in each of the three abundance categories (throughout the combined 20 catches) was then calculated. To calculate the average number of bycatch individuals per 10-kg subsample (over all the catches), the following equation was used: X = 10 X (Total number /Total weight) (2) where X = the mean number of bycatch individuals (per 10-kg subsample); Total number = the total number of all bycatch individu- als (summed over all 20 catches); and Total weight = the total weight (kg) of all bycatch indi- viduals (summed over all 20 catches). We then examined the average occurrence ratios within 10-kg subsamples for the "rare," "common ," and "abundant" species. Catch composition To examine the relationship between the number of recorded species and the weight of sorted catch, the subsamples were first analyzed in the order that they were collected. The cumulative number of spe- cies (both fish and invertebrates) was plotted against the cumulative weight of sorted catch, for each of the 20 catches. Each catch was also summarized in terms of the percentage of species recorded for each 10% increment of weight of sorted catch. The order (position on the sorting tray) where the sub- samples were collected on both the research and commercial vessels was just one of the many possible ways that a catch can be divided into 10-kg subsamples. To determine the level of accuracy in recording the number of species in a catch, we examined 200 combinations of subsample selection (with no replacement), by randomly reordering the subsamples using Monte Carlo simulations for each catch. We also cal- culated the cumulative number and percentage of species recorded, as well as the cumulative weight and percentage of the sorted catch, for each catch. The proportion of species recorded was fitted as a power function of the proportion of the weight of sorted catch, as described by the following asymptotic equation (Snedecor and Cochran, 1980): y=p'' + e. (3) where y = the proportion of species recorded; p = the proportion of the weight of catch sorted; k = the mean exponential parameter; and e = the random normal error term, with unequal variance. The variance of £ is assumed to be p (1-p) a^ to ensure that the variance of y is fixed at zero when p = 0 and 1. This formulation has the property that, when none of the catch has been sorted, no species will have been recorded. It also ensures thaty = 1 whenp =1, i.e. when all the catch has been sorted, all of the species have been recorded. The estimate of ct- was obtained by fitting the following model according to the SAS procedure NLIN (version 7, SAS Inst., Cary NC): y * = p' I ^{pO-p)) + £' (4) where y* = y H(p(l-p)) and £* = E/^(p(l-p)); and £* now has homogeneous variance structure. Different k values were estimated for each catch to re- flect the variation in the relationship. The mean ki value for a given catch (! = l-20) was obtained from 200 analyses for that catch. The predicted y values i.e. y^ (atp=0.1, 0.2 etc. to 1.0) were obtained by averaging p*' values across_the 20 catches (note that this is different from p'- where k is the mean k value for the 20 catches). We defined the y^ values as the predicted expected proportion of species recorded after p proportion of catches had been sorted. 794 Fishery Bulletin 101(4) The corresponding 95% confidence interval for the pre- dicted mean values (vp) was evaluated by using the width 1.96 a„ where 10 to <50, and >50 per subsample) had mean sampling error rates below 25%. For the "rare" species ( ^ ^^ ^"^ Cumulative weight of catch sorted (l0.05) between the southern CPFV and northern speared samples (Fig. 6). The models representing males were virtually identical, with param- eters for the CPFV model of t^ = -0.94 years, k = 0. 195/year, and L„ = 331.66 mm FL, and for the speared model of t^ = -0.99 years, k = 0.194/year, and L^ = 323.14 mm FL. There was no significant difference between females; <„ = -1.94 years, k = 0.107/year, and L„ = 430.74 mm FL for the CPFV samples and t^ = -1.14 years, k = 0.166/year and L„ = 393.34 mm FL for the speared. Females from CPFV were larger at ages after 15 years than those that had been speared. Discussion 100 200 300 400 500 Julian date of capture (day) 600 700 Figure 2 Growth of blue rockfish (Sebastes mystinus) collected nearshore during their first 1.5 years. Solid line is the fitted linear growth model l/i=99; r-=0.91). Vertical line represents 1 May, one year after the parturition date of 1 January (i.e. 365 +120=485 days). We estimated the age (using the break-and- burn technique) of blue rockfish to be greater than that reported in earlier studies. Aging the scales of blue rockfish. Miller and Geibel ( 1973 ) reported maximum ages of 24 and 17 years for females and males, respectively, whereas the oldest of either sex reported by MacGregor (1983) was only 13 years. Based on modal pro- gression of length distributions, the estimate of the oldest individuals of either sex calculated by Karpov et al. (1995) was 17 years. In a study of blue rockfish off Newport, Oregon, McClure (1982) examined otolith surfaces and deter- mined that the oldest female was 16 years, and the oldest male was only 12 years. In aging males to 44 years and females to 41 years, our study more than doubled the recorded maxi- mum ages for blue rockfish, demonstrating the value of the break-and-burn section method for accurate age determination. Age data were validated by using an edge analysis and the first translucent zone was validated as corresponding to the first annual growth increment. Campana (2001) pointed out that there are problems in using edge analysis as a validation tool. Specifically the extension of younger, validated ages to older, nonvalidated ages. In our study, we validated ages up to 23 years for females and up to 25 30 2.5 2.0 g 1 5 05 0,0 Diameter = 0.02(FL) - 0.02 60 Fork length (mm) 120 Figure 3 Change in otolith diameter with fish length for young-of-the-year and one-year-old blue rockfish (Sebastes mystinus). Solid line is the fitted linear model ( n = 198; r2=0.95 ). Dashed line represents an estimated total otolith diameter of 2.19 mm for a fish of 108.5 mm FL (i.e. fish length at time of translucent zone completion). years for males; ages of older fish could not be positively validated. Therefore, caution must be taken when using the older ages. The growth rates of blue rockfish in our study were similar to those estimated by others in California, but slower than conspecifics off Oregon (Fig. 7). MacGregor (1983) examined blue rockfish from southern California and determined the combined male and female growth 804 Fishery Bulletin 101(4) 4.5 1.5 *,(« S» - » . - . ^i^^^ y» 4t - - , - §^ -S**^ 100 200 300 Fork length (mm) 400 500 Figure 4 Diameter of the first, second, and third translucent zones in otoliths from blue rockfish iSebastes mystinus) at various lengths. Dashed lines represent the mean diameters. 500 450 400 F 350 300 0) 250 o 200 150 100 20 25 30 Age (yr) Figure S Von Bertalanffy growth models for all male (solid line) and female (dashed line) blue rockfish {Sehastes mystinus). (;i=348 for males and /i = 1348 for females.) rate and calculated k (instantaneous growth rate) to range from 0.13-0.16/years, which was comparable to k in our study (0.2/year for males and 0.15/years for females; mean ^=0.17/year). Karpov et al. (1995) calculated k for the combined male and female growth rate from modal progressions studies to be 0.12/year. This also was similar but less than the k from our present study. On the other hand, McClure ( 1982) estimated a much faster growth rate for blue rockfish off Oregon, with a k for males of 0.23/years and for females of 0.31/year. Although the Oregon fish were larger at age (Fig. 7), maximum sizes from Oregon and California were similar; the largest specimen from Oregon was 460 mm FL (McClure, 1982) and the largest individual from our study was 444 mm FL. The difference in growth between studies may be attrib- uted to a temporal difference in the collection offish. Two thermal regime shifts have occurred in the Pacific Ocean over the past 25 years; one in 1977 and the other in 1989 (Hare and Mantua, 2000). The samples from our study came from two different thermal regimes, but the growth Laidig et al.: Age and growth of Sebastes mystinus 805 curves were not statistically different (Table 1). Therefore, these regimes did not appear to effect the growth of adult fish. Out of the four other surveys mentioned above, three came from one of these two regimes, and the fourth, Miller and Geibel (1973), came from an earlier regime. If there were any effects from the three different thermal regimes, it would seem clear that these differences would show up between samples from such varied regimes. However, the only study with different measures of growth was that from Oregon (McClure, 1982), with samples that were collected during the same regime as two of the other studies (Mac- Gregor, 1983; and the present study). Therefore, thermal regime alone does not seem to have a major impact on the growth of blue rockfish, although further analysis is needed to confirm this point. These differences in growth parameters between fish from California and Oregon may be attributed to differ- ences in aging methods. Wilson and Boehlert (1990) found that estimates of growth based on aging of otolith surfaces were higher for Sebastes pinniger, but were similar to growth rates estimated from otolith sections for S. diplo- proa. The ages of S. alutus determined from otolith surfaces had poor correlation with ages from otolith cross-sections for fish older than 17 years, but there was close agreement for younger fish (Stanley and Melteff 1987). Reading ages from the surface of an otolith may underestimate the age of a rockfish (Munk, 2001) and thus result in greater size- at-age and growth rate estimates. However, aging methods may not be the only factor influencing the growth discrep- ancies. Miller and Geibel (1973) and MacGregor (1983) both used scales to age blue rockfish (which also can under- estimate the age offish [Beamish and MacFarlane, 1987]), and, yet, their growth models more closely approximated the model produced by our study. Faster growth estimated for blue rockfish off Oregon may reflect a latitudinal difference in growth. Fraidenburg (1980) examined length and age composition of Sebastes flavidus and reported evidence of a north-to-south cline of decreasing size-at-age. Pearson and Hightower (1991) studied S. entomelas and noted smaller k values and larger average maximum lengths with increasing latitude. Boe- hlert and Kappenman ( 1980) reported faster growth in the north for S. diploproa and no difference in growth with latitude for S. pinniger. They postulated that because the fish live demersally on the continental shelf latitudinal variation in environmental factors may be insufficient to explain the difference in growth rates and that differential exploitation by the fishery may be a possible influence on growth. Blue rockfish live at relatively shallow depths where environmental and biological factors may have a greater influence on their populations. Although blue rockfish display a possible latitudinal trend in growth rate between California and Oregon, within California no latitudinal trend in growth rates was observed. Specimens from both the southern CPFV sample and the northern speared sample areas had translucent zone completion by 1 May, which was consistent with the findings of Miller and Geibel (1973) using ages from scales. Individual fish in our study also had similar maximum ag- es and maximum fish lengths in the north and south areas. 4bU Males 400 350 300 /"""^ bpearea 250 200 / 150 ^- 100 50 10 20 30 40 50 1° '*^° Females 400 350 300 250 200 150 100 50 10 20 30 Age (yr) 40 50 Figure 6 Von Bertalanffy growth models for male and female blue rockfi.sh ^Sebastes mystinus) from the CPFV (thick Hne) and speared samples (narrow line). Individual data points are plotted for blue rockfish from CPFV (open circle) and speared samples (open triangles). No latitudinal trend in growth rates was observed over the 280 km between the centers of the two sampling ar- eas. Although growth rates varied throughout their study area from Half Moon Bay in the north to Morro Bay in the south. Miller and Geibel (1973) likewise observed no latitudinal trend in growth for blue rockfish. Blue rockfish have average maximum ages and growth rates when compared to other rockfish species. Maximum ages for rockfishes {Sebastes spp.) range from 12 years for the relatively small calico rockfish to 205 years for rough- eye rockfish, one of the largest species (Cailliet et al., 2002; Love et al., 2002). According to Love and Johnson (1998), of the 38 species most accurately aged, most lived to more than 40 years. Love et al. (1990) found growth rates for three species that share the blue rockfish habitat (black, /fe=0.12-0.21/year;yellowtail,^=0.16-0.20/year; and widow rockfish, /j=G.14-0.22/year) to be similar to that for blue rockfish (k=0.n years). Mean k values for rockfish varied 806 Fishery Bulletin 101(4) 500 400 ^ 300 200 100 Females (OR, McClure, 1982) Males (OR, McClure, 1982) ■Miller and Geibel (1973) -MacGregor (1983) 9 Age (yr) 12 15 Figure 7 Growth models from five published studies of blue rockfish (Sebastes mystinus). from 0.04/year for female silvergrey rockfish to 0.62/year for the shorter-lived dwarf Puget Sound rockfish, with the average range of ^ values occurring from 0.1 to 0.3/years (Love et al,, 1990; Beckman et al., 1998). This considerable longevity and relatively slow growth rate have significant effects on the ability of many rockfish stocks to withstand exploitation. The age and growth relationships described in this study indicate that both recruitment of blue rockfish to the fishery and their maturity occur at younger ages than previously reported. Blue rockfish enter the fishery at a size of approximately 200 mm (Miller et al., 1967; Miller and Geibel, 1973). This length equates to ages of 2-4 years as determined in our study compared to 3-5 years as es- timated by Miller et al. (1967). The new estimates for age at which 50% of individuals are mature (using fish lengths from Miller and Geibel, 1973) are even more striking: our estimated age at 50% maturity is 5-6 years for males and 5 years for females, whereas estimates from Miller and Geibel ( 1973) and Echeverria ( 1986) were 7 years for males and 7-8 years for females. Similarly, the youngest mature males and females in these early studies were 4-5 years, whereas we estimated the age to be 3 years. The changes observed in our study in age-at-length, maximum age, recruitment age, and age at 50% matu- rity have important implications for stock assessments. Accurate information on age composition, weight-at-age, age specific availability to the fishery, and maturity-at-age is crucial to the proper functioning of the stock synthe- sis model (Methot, 1990), which is used for Pacific coast groundfish. If incorrect age data are used, it could lead to erroneous estimates of population size, and subsequently to either overfishing or an unnecessary reduction in allow- able catch. Acknowledgments We would like to thank all the port samplers who collected the CPFV catch data. We also thank the crew and scientific personnel aboard the RV David Starr Jordan for collecting samples. We thank James Chess, Edmund Hobson, Dan Howard, and Kelly Silberberg for braving the cold waters of the Pacific Ocean to collect the nearshore specimens. We also thank Churchill Grimes and Mary Yoklavich for their many constructive comments and the reviewers of this manuscript. Literature cited Beamish, R. J., and D. A. Fournier. 1981. A method of comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Beamish, R. J., and G. A. MacFarlane. 1987, Current trends in age determination methodology. In Age and growth of fish (R. C. Summerfelt and G. E. Hall, eds), p. 15-42. Iowa State Univ, Press, Ames, lA, Beckmann, A. T., D, R. Gunderson, B. S. Miller, R. M. Buckley, and B. Goetz. 1998. Reproductive biology, growth, and natural mortality of Puget Sound rockfish, Sebastes emphaeus (Starks, 1911). Fish, Bull. 96:352-356. Laidig et al : Age and growth of Sebastes mystinus 807 Boehlert, G. W., and R. F. Kappenman. 1980. Variation of growth with latitude in two species of rockfish (Sebastes pinniger and S. diploproa) from the northeast Pacific Ocean. Mar Ecol. Prog. Ser 3:1-10. Brown, I. W., and W. D. Sumpton. 1998. Age, growth and mortality of redthroat emperor Leth- rinus mintatus (Pisces: Lethrinidae) from the southern Great Barrier Reef, Queensland, Australia. Bull. Mar Sci. 62:905-917. Cailliet, G. M., A. H. Andrews, E. J. Burton, D. L. Watters, D. E. Kline, and L. A. Ferry-Graham. 2002. Age determination and validation studies of marine fishes: do deep-dwellers live longer? Exp. Gerontology 36: 739-764. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Chilton, D. E., and R. J. Beamish. 1982. Age determination methods for fishes studied by the groundfish program at the Pacific Biological Station, 102 p. Can. Spec. Publ. Fish. Aquat. Sci. 60. 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. Draper, N., and H. Smith. 1981. Applied regression analysis, 2d ed., 709 p. John Wiley and Sons, New York, NY. Echeverria, T. 1986. Sexual dimorphism in four species of rockfish genus Sebastes (Scorpaenidae). Environ. Biol. Fishes 15: 181-190. Echeverria, T., and W H. Lenarz. 1984. Conversions between total, fork, and standard lengths in 35 species of Sebastes from California. Fish. Bull. 82: 249-251. Fraidenburg, M. E. 1980. Yellowtail rockfish, Sebastes flavidus. length and age composition ofTCalifornia, Oregon, and Washington in 1977. Mar Fish. Rev 42:54-56. Hare, S. R., and N. J. Mantua. 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr. 47:103-145 Karpov, K. A., D. P Albin, and W. H. Van Buskirk. 1995. The marine recreational fishery in northern and cen- tral California, 192 p. Calif Dep. Fish Game Fish Bull. 176. Kimura, D. K., R. R. Mandapat, and S. L. Oxford. 1979. Method, validity, and variability in the age determi- nation of the yellowtail rockfish (Sebastes flavidus) using otoliths. J, Fish. Res. Board Can. 36:377-383. Lea, R. N., R. D. McAllister, and D. A. VenTresca. 1999. Biological aspects of nearshore rockfishes of the genus Sebastes from central Cahfomia, 109 p. Calif Dep. Fish Game Fish Bull. 177. Love, M. S., and K. Johnson. 1998. Aspects of the life histories of grass rockfish, Sebastes rastrelliger . and brown rockfish. S. auriculatus, from south- ern California. Fish. Bull. 87:100-109. Love, M. S., P. Morris, M. McCrae, and R. Collins. 1990. Life history aspects of 19 rockfish species (Scorpae- nidae: Sebastes) from the Southern California Bight, 38 p. NOAATech. Rep. NMFS 87 Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the northeast Pacific, 405 p. Univ California Press, Berkeley, CA. MacGregor, J. S. 1983. Growth of the blue rockfish (Sebastes mystinus). CALCOFI (Calif Coop. Ocean. Fish. Investig.) Rep. XXIV: 216-225. Mayo, R. K., V. M. Gifford, and A. Jearid Jr 1981. Age validation of redfish, Sebastes marinus (L.), from the Gulf of Maine-Georges Bank region. J. Northwest Atl. Fish. Sci. 2:13-19. McClure, R. E. 1982. Neritic reef fishes off central Oregon: aspects of life histories and recreational fishery. M.S. thesis, 94 p. Oregon State Univ., Corvallis, OR. Methot, R. D. 1990. Synthesis model: an adaptable framework for analy- sis of diverse stock assessment data. Bull. Int. North Pac. Fish. Coram. 50:259-277. 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, 137 p. Calif Dep. Fish Game Fish Bull. 158. Miller, D. J., M. W. Odemar, and D. W. Gotshall. 1967. Life history and catch analysis of the blue rockfish (.Sebastodes mystinus) off central California, 1961-1965. Calif Dep. Fish Game Mar Res. Operations Ref 67-14: 1-130. Munk, K. M. 2001. Maximum ages of groundfish in waters off Alaska and British Columbia and considerations of age deter- mination. Alaska Fish. Res. Bull. 8:12-21. Pearson, D. E. 1996. Timing of hyaline-zone formation as related to sex, location, and year of capture in otoliths of the widow rock- fish, Sebastes entomelas. Fish. Bull. 94:190-197. Pearson, D. E., and J. E. Hightower 1991. Spatial and temporal variability in growth of widow rockfish (Sebastes entomelas), 47 p. NOAA Tech. Memo., NOAA-TM-NMFS-SWFSC 167. Pearson, D. E., J. E. Hightower, and J. T. H. Chan. 1991. Age, growth, and potential yield for shortbelly rockfish, Sebastes jordani. Fish. Bull. 89:403-409. PFMC (Pacific Fishery Management Council). 2001. Status of the Pacific coast groundfish fishery through 2001 and recommended acceptable biological catches for 2002. Pacific Fishery Management Council, Portland, OR. Ratkowsky, D. A. 1983. Nonlinear regression modeling, 276 p. Marcel Dekker, New York, NY. Rogers, J. B., M. Wilkins, D. Kamikawa, F. Wallace, T. Builder, M. Zimmerman, M. Kander, and B. Culver 1996. Status of the remaining rockfish in the Sebastes complex in 1996 and recommendations for management in 1997. Appendix E: Status of the Pacific coast groundfish fishery through 1996 and recommended biological catches for 1997: stock assessment and fishery evaluation, 59 p. Pacific Fishery Management Council, Portland, OR. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. Six, L. D., and H. F. Horton. 1977. Analysis of age determination methods for yellowtail rockfish, canary rockfish, and black rockfish off Oregon. Fish. Bull. 75:405-414. 808 Fishery Bulletin 101(4) Stanley, R. D., and B. R. Melteff. 1987. A comparison of age estimates derived from the sur- face and cross-section method of otoUth reading for Pacific ocean perch (Sebastes alutus). Lowell Wakefield fisheries symposium: proceedings of the international rockfish sym- posium, Anchorage, Alaska, USA, Oct. 20-22, 1986 Alaska Sea Grant Rep. 87-2, p. 187-196. Univ. Alaska, Anchorage. AK. Wales. J. H. 1952. Life history of the blue rockfish, Sebastodes mystinus. Calif Fish Game 38(4):485-498. Wilson, C. D., and G. W. Boehlert. 1990. The effects of different otolith ageing techniques on estimates of growth and mortality for the splitnose rockfish, Sebastes diploproa, and canary rockfish, S. Pinniger. Calif Fish Game 76:146-160. 809 Abstract— The reproductive activity and recruitment of white mullet (Mugil curema) was determined by observa- tions of gonad development and coastal juvenile abundance from March 1992 to July 1993. Adults were collected from commercial catches at three sites in northeastern Venezuelan waters. Spawning time was determined from the observation of macroscopic gonadal stages. Coastal recruitment was deter- mined from fish samples collected biweekly by seining in La Restinga Lagoon. Margarita Island, Venezuela. The examination of daily growth rings on the otoliths of coastal recruits was used to determine their birth date and estimate the period of successful spawning. Fish with mature gonads were present throughout the year but were less frequent between September and January when spawning individu- als migrated offshore. In both years, juvenile recruitment to the lagoon was highest between March and June when high densities of 25-35 mm juve- niles were observed. Back-calculated hatching-date frequency distributions revealed maximum levels of successful spawning in December-January that were significantly correlated with peri- ods of enhanced upwelling. The rela- tion between the timing of successful spawning and the intensity of coastal recruitment in white mullet was likely due to variations in food availability for first-feeding larvae as well as to varia- tions in the duration of the transport of larvae shoreward as a result of vary- ing current conditions associated with upwelling. Reproduction and recruitment of white mullet iMugil curema) to a tropical lagoon (Margarita Island, Venezuela) as revealed by otolith microstructure* Baumar J. Marin E. Antonio Quintero Institute Oceanografico de Venezuela Universidad de Oriente Cumana 6101 Edado Sucre, Venezuela E-mail address (for B. J Mann E.): bmann@sucre. udo.edu. ve Dany Bussiere Julian J. Dodson Departement de biologle Unlverslte Laval, Ste-Foy Quebec, Canada GIK 7P4 Manuscript approved for publication 10 June 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:809-821 White mullet (Mugil curema) is a wide- spread coastal pelagic fish occurring from Massachussetts to southern Brazil. Considered to be catadromous, the juve- nile fish recruit to lagoons and estuaries following a period of offshore spawning (Blaber, 1987; Ibahez-Aguirre, 1993; Ditty and Shaw, 1996). White mullet is an important economic resource supporting many small communities through both fishing and aquaculture (AIvarez-Lajonchere, 1982; Gomez and Cervigon, 1987). Small schools of mullet are captured with gill and "atarraya" nets near the coast and in neritic waters and between 300 and 400 metric tons are sold annually on Margarita Island, Venezuela. Reproductive periodicity in white mullet varies over its geographic distri- bution. Several authors have reported protracted or continuous reproduction in tropical waters and generally two spawning peaks per year (Jacot, 1920; Anderson, 1957; Angell, 1973; Moore, 1974; AIvarez-Lajonchere, 1976, 1980; Yaiiez-Arancibia, 1976; Rodriguez and Nascimento, 1980; Garcia and Bus- tamente, 1981; Franco, 1986; Ibanez- Aguirre, 1993). Figure 1 summarizes previous work describing the spawning periods of M. curema based on gonad development and estimated according to the arrival of juveniles in the coastal zones. The spawning period is quite variable. Angell (1973) suggested that schooling occurs in coastal areas just prior to the offshore spawning migra- tion and that the departure of individu- als for the spawning grounds causes a reduction of the gonadosomatic index in the nearshore populations. Moore ( 1974) also reported that during the spawning period fully ripe fish are rare in coastal collections. Despite these studies, little is known of the factors influencing re- productive patterns of the white mullet. Ibanez-Aguirre (1993) suggested that the timing of reproduction in M. curema in Tamiahua Lagoon, Mexico, is an ad- aptation to avoid competition with juve- niles of the conspecific Mugil cephalus. In areas of favorable thermal regimes, M. curema may penetrate a wider range of salinities and competitively exclude M. cephalus (Moore, 1974). The periodicity of white mullet re- production may be related to environ- mental variability that signals periods of optimal early growth and survival. Stability of the water column and suit- ' Contribution of Quebec-Ocean, Pavilion Alexandre-Vachon, Local 2078, Universite Laval, Quebec, Qc. GIK 7P4. 810 Fishery Bulletin 101(4) Location ^ FMAMJJ ASOND I I I I I I I I I I I I Cariaco Gulf, Venezuela (Franco, 1986) 1.2 La Restinga Lagoon, Venezuela (Angell, 1973) 1,2 Patanemo Lagoon, Venezuela (Blanco, 1985) 2 Tunas de Zaza. Cuba (Alvarez-Lajonchere, 1980)2 Cuba (Garcia and Bustamante, 1981) 3 Guerrero Stale Lagoons, Mexico (Yanez-Arancibia, 1976) 3 Gulf of fdlexico (Ibanez-Aguirre, 1993) 3 Southern Texas, USA (Moore, 1974) 3 South Florida, USA (Anderson, 1957) 3 North Carolina, USA (Jacol, 1920) 3 Figure 1 Spawning periods for white mullet (Mugil curema) in tropical waters, based on the histological observations of gonads (1), macroscopic observations of gonads (2), and estimated from the arrival of juveniles in coastal areas (3). Thin horizontal lines indi- cate periods of continuous but minor spawning between periods of peak spawning. able food in coastal lagoons, river deltas, and estuarine mangrove areas have been identified as important factors influencing the recruitment of juvenile Mugilidae (Yafiez- Arancibia, 1976; Blaber and Blaber, 1980; Blaber, 1987; Vieira, 1991), Based on macroscopic gonad observations of schools of white mullet captured offshore, Etchevers (1974) proposed that the spawning of white mullet recruiting along the southern coast of Margarita Island, Venezuela, occurs between La Tortuga Island and Margarita Island in the vicinity of the 1000-m deep Cariaco trench (Fig. 2). Seasonal environmental variability in this area is mainly generated by the alternation between upwelling during the dry season and freshwater discharge during the wet season (Gomez, 1983;Muller-Kargeret al., 1989). The rainy season strongly influences the eastern Caribbean as freshwater plumes from the Amazon and Orinoco Rivers lower salini- ties throughout the region. Both upwelling and freshwater runoff produce intense peaks in coastal primary production (Gines 1972; Ferraz-Reyes et al,, 1987; Miiller-Karger et al,, 1989), which could influence spawning periodicity and recruitment success. The purpose of this study was to docu- ment the periodicity of reproduction and recruitment of A/, curema along the southern coast of Margarita Island and to examine their relationship with environmental signals, particularly those associated with upwelling. Methodological advances in counting daily growth increments in otoliths of marine fishes (Pannella, 1971; Campana and Nielsen, 1985) have greatly aided studies of the age, growth, and recruitment of larval and juvenile fishes (Wilson and Larkin, 1980; McBride and Conover, 1991; Jenkins and May, 1994; Sirois and Dodson, 2000). For the striped mullet (M, cephalus), a close relative of the white mullet, Radtke (1984) showed that the first incre- ment is formed one day after hatching and that additional increments are formed daily thereafter. Daily growth rings have also been demonstrated in laboratory studies for M. so-iiiy by Li et al, (1993). In the present study, we examined the microstructure of the otoliths of juveniles recruiting to a coastal lagoon in order to back-calculate the date of hatching and hence the time of successful spawning. We first validated that otolith growth increments of juvenile M. curema were formed daily. Material and methods Reproductive periodicity was documented from samples of adult fish taken monthly from commercial catches in three fishery zones in Venezuela: 1) the Chacopata zone, located between Chacopata lagoon and Coche and Cubagua Islands; 2) the Cariaco Gulf zone; and 3) the Margarita zone located along the southern coast of Margarita Island and the northern coast of Cubagua Island (Fig, 2), Measure- ments of water temperature, salinity, and rainfall were Marin et al,: Reproduction and recruitment of Mugil curema 811 n Sea La Tonuga > — ■ / — 150 m \ \ 500 s. lOOOm lOOOm \ Cariaco Trench ^ ^ (western part) — 1 — 65° Santa Fe Gulf Figure 2 Map of the northeastern coast of Venezuela showing locations mentioned in the text. collected periodically throughout the entire study period at the La Salle meteorological station, next to La Restinga Lagoon on Margarita Island. Total (TL) and standard (SL) lengths of adult mullet were measured to the nearest 0.5 cm and total and gut- ted body mass were recorded to the nearest 0.1 g. Sexual maturity was determined by observation of the gonads and gonadal stages were classified as follows: Stage I Ovaries transparent and inconspicuous, whit- ish-yellow in color and rounded with a small diameter. Testes longer than ovaries and rib- bonlike in form. Stage II Ovaries rounder and wider than in stage I, and yellow in color. Testes thinner, and wider than stage I, but still with thin edges and a ribbon- Uke form; white in color. Stage III Ovaries large, pale yellow, smooth in appear- ance, turgid, and round. Ovocytes easily distin- guished macroscopically (as granular). Testes milky-white in color (bright), turgid, and wider in appearance and having thicker edges than in stage II. Stage rV Spawned (spent) ovaries purple and wrinkled in appearance. Testes whitish, or transparent with white patches, and wrinkled in appearance. Recruitment periodicity was documented from samples of juveniles seined at semimonthly intervals at the mouth of La Restinga Lagoon (Fig. 2). The 2-cm mesh beach seine measured 1.5 m deep and 50 m long. Juvenile white mullet were distinguished from other juvenile mullets according to the descriptions of Alvarez-Lajonchere et al. ( 1976). White mullet juveniles were characterized by a scaly gray appear- ance as opposed to the shiny metallic gray appearance of a sympatric mullet species (Mugil incilis). For white mullet, recruitment is defined as the appearance of juveniles in coastal areas (Vieira, 1991). We calculated catch per unit of effort (CPUE) as the number of juveniles per seine haul. For all samples, standard length of fish was measured to the nearest 1 mm. Otolith analysis was restricted to one sampling period per month. After examining size-fre- quency distributions of juveniles captured in the lagoon, the otoliths of approximately 20 individuals representing all cohorts collected on a given sampling date were ana- lyzed. The otoliths (sagittae) were removed with needles, rinsed in water, and then attached to strips of masking tape. The otolith was then sanded to obtain a transversal section (Fig. 3) with a thickness of approximately 20 ^m by using the technique described by Secor et al. ( 1992) and a metallurgic jig adapted from Neilson and Geen (1986). Readings of the number of increments were made along the curvilinear surface running from the nucleus to the edge of the otolith (Fig. 3). Because daily growth increments were 812 Fishery Bulletin 101(4) Regular rings (141) Central rings (14) Nucleus B Central Regular nngs 1 1 null I I I I 1 1 mill I I I I 1 1 iiiiii I I I I 1 1 mill I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I II I I I I I I I I I I I I II I I I I I I I I I I I I Nucleus «r Cut plane Figure 3 (A) Photomicrograph in optical microscopy of the posterior radius of a polished otolith from a 180-day-old juvenile white mullet iMugil curema). Different sequences of rings are indicated which result from a horizontal plane not aligned with the changing growth plane of the otolith. Numbers of rings in each sequence are shown in parentheses. (Bi Schematic presentation of the changing growth plane of the otolith, the cut plane, and the resultant increment distortion (black rectangles represent the distance between adjacent rings). less consistent in the anterior field, otolith counts were always made along the posterior radius of the sagittae. We also measured the area of the nucleus, which represents the prehatch zone. All measurements were made under a microscope which was connected to an image analyzer and computer. To evaluate the error in counting growth increments on otoliths, one reader made 7 independent counts of the number of growth increments on otoliths obtained from 11 juveniles representing the size-range of sampled fish. The number of growth increments ranged from 63 to 289 (mean=147) and the mean coefficient of variation was 8.71% (SD=2.13%) (Fig.4). We therefore considered an error of approximately 10% for the counts of growth incre- ments. In applying the technique to the subsamples of the different cohorts sampled in the lagoon, at least two counts were made for each otolith. All counts were made by the same person. To evaluate if otolith growth increments were formed daily, we read the otoliths of juveniles sampled on succes- sive sampling dates and compared the average increase in the number of otolith increments to the number of days between samplings (Struhsaker and Uchiyama, 1976; Jor- dan, 1993; Jenkins and May, 1994). Birthdate was obtained by subtracting the number of daily growth rings on otoliths from the date of capture. We used the hatching mark as defined for M. cephalus (Radtke, 1984) and M. iso-iuy {Li et al., 1993) to locate the hatching mark on the otoliths of the white mullet. Knowing that white mullet embryos hatch from 24 to 40 hours after fertilization (Anderson, 1957; Houde et al., 1976), we back-calculated hatching dates of recruits to es- timate when successful spawning occurred. We examined the relationship between the spawning dates of recruits and an index of the intensity of upwelling. In calculat- ing this index, we determined wind stress based on data fnmi Fundacion La Salle, Margarita Island, and Cumana Airport meteorological stations. The upwelling index (UI) was based on Bowden's (.1983) theoretical calculations as follows: Mann et al ■ Reproduction and recruitment oi Mugil curema 813 Figure 4 Number of growth increments (counted on seven independent occasions) on otoliths from 1 1 juvenile white mullet of varying size. Means, standard errors, and standard deviations are illustrated. Ul=- 100 where f = Coriolis parameter; Tjj. = surface wind stress; and p„, = average density of the water (1025 kg/m'^). The term /'was calculated as /■=2ojsin(/;), where co = angular velocity of rotation of the earth (7.29xl0-5s [seconds]): and /j = latitudinal position at the place i. The term t^^ represents surface wind stress measured in the .r-axis perpendicular to the coast (Bowden, 1983), often considered in terms of the empirical equation Tj^ = ^ X p_j X W^, where k = empirical drag coefficient (1.11 to 3.25, as a function of wind velocity; Bowden 1983); p^ = mean density of the air ( 1.25 kg/m^); and W = wind velocity. pling period during each of the 18 months of the study by applying the frequency distribution of birth dates of aged juveniles to the total catch for that date. A total of 398 juveniles were aged by otolith analysis. If x% of aged fish captured on a given date were hatched on Julian Day y, this percentage was applied to the total catch of juveniles for that sampling date. Secondly, all fish hatched on a given day were summed across the 18 monthly sampling dates. This frequency distribution was then correlated with the distribution of UI estimates over the same period of time as that of the birth dates. Before proceeding with correlation, trends in birth date and UI data series were described by using a smooth- ing spline. The spline fit uses a set of smoothly spliced 3'''' degree polynomial segments (Simonoff, 1996; JMP® software, version 3.2.1, SAS Institute, Gary, North Caro- lina). Predicted values were correlated with the raw data points in order to optimize the value of lambda used to fit the smoothing spline. Increasing the value of lambda increases the degree of smoothing but weakens the correla- tion between predicted and raw data. Pearson correlation and cross-correlation functions were used to describe the temporal relationship between upwelling and the date of hatching offish recruited to the coastal lagoon. The drag coefficient, k, changes as a function of wind velocity and gives values equivalent to those of Bakun et al.(1974). The relationship between upwelling and the birth dates of successful recruits (captured in La Restinga Lagoon) was determined in two steps. First, we calculated the birth dates of juveniles captured during one monthly sam- Results The white mullet surveyed in the commercial catches measured from 4 to 36 cm SL. The largest fish were from the Chacopata zone where the most abundant sizes classes were those from 22 to 30 cm. The most abundant sizes in the Cariaco Gulf and Margarita Island zones were 20 to 814 Fishery Bulletin 101(4) 26 cm and 18 to 26 cm, respectively. The mullet from the Margarita zone were mainly juveniles and small adults (Fig. 5). Maturity and reproductive periodicity An examination of gonad maturity revealed that 90% of the male and female mullet in the Margarita zone were immature or at developing stages (I and II) and only 10% in developed and spawning stages (III and IV). In contrast, 53.8% of females and 42.8% of males of the population of generally larger mullet sampled at Chacopata were in developed and spawning stages. Finally in the Cariaco Gulf zone, where the fish were of intermediate size, compared to fish at Margarita and Chacopata, 41.2% of females and 32.7% of males were in developed and spawning stages. Throughout our study, sexually mature (stage-Ill) fish were present in the samples from the Chacopata fishery (Fig. 6), and their abundance showed a marked seasonal pattern. Mature and spent fish were least abundant (<25% of the population) between September and January and most abundant from April to August 1992 and May to June 1993. In contrast, in the Margarita fishery, immature (stage-I) fish dominated the samples and mature fish only occurred sporadically (Fig. 6). Finally, in the Cariaco Gulf zone, immature and maturing fish (stage-I, and stage-ID generally dominated the population, except in July when mature fish became abundant. Otolith microstructure The otolith of M. curema had a round nucleus with a mean radius of 9.26 ^m (95% confidence interval (CI)=0.54, n=8), and the dark area in the center had a mean diameter of 4.97 jjm (CI=0.82, n=8). The otolith was round during the larval period and became ovoid when mullet reached 10-12 mm (SL). In the early juvenile stage (18-20 mm SL), the otolith was strongly elongated and the anterior end was arrow-like. In juveniles (>20 mm SL), the otolith was always thin, concave, and umbrella shaped in form (Fig. 3). Validation of otolith increment lines The otolith increments counted for the first strong cohort present in the lagoon during March and April 1992 dem- onstrated that the average number of increments added during the 14-day interval between sampling collections was close to 14 days (Table 1). This indicated that otolith increments were formed daily, as observed in other species of mullet. Chacopata _LL Margarita 160 140 120 100 80 60 40 20 0 160 140 120 - 100 I 80 C I 60 40 20 0 180 160 140 120 100 80 60 40 20 0 12 14 12 16 18 20 22 24 26 28 30 32 34 36 38 Standard length (cm) Figure 5 Length-frequency distributions of white mullet surveyed in the commercial catches from the three fishery zones of northeast Venezuela. Cariaco Gulf Juvenile recruitment The catch-per-unit-of-effort measurements for juveniles captured in La Restinga Lagoon demonstrated a seasonal pattern and high recruitment from March to early July 1992 and from late March to May 1993, and low recruit- ment during the remaining months (Fig. 7). The recruit- ment peak in 1992 was more than twice that in 1993. The periods of strong recruitment were associated with the rainy season in northeastern Venezuela (Fig. 7), as previ- ously reported by Okuda et al. (1978), Gomez (1983), and Ferraz-Reyes(1989). The discontinuous length-frequency distributions of juveniles sampled biweekly suggested the presence of four cohorts in the lagoon during the study period (Fig. 8). Two cohorts were present on 5 March 1992, the first sampling date. The cohort of smaller juveniles, referred to as the first cohort, had a mean length of 29.8 mm (range of 18 to Mann et a\ Reproduction and recruitment of Mugil curema 815 100 g 80 o. 60 B 40 I 20 100 80 60 40 20 0 Chacopata Cariaco Gulf Margarita MAMJ JA sond;j FMAMJ J 1992 ■ 1993 Figure 6 Percent composition of four maturity stages (I to IV, see text for definitions) of Mwg;/ curema (males and females combined) captured from Margarita, the Cariaco Gulf, and the Chacopata fisher)' zones of northeast Venezuela. 36 mm) and varied in age from 50 to 70 days (Fig. 9). The cohort oflarger juveniles, referred to as the second cohort, had a mean length of 105 mm (range of 90 to 130 mm), and otolith analysis indicated an age of 160 to 240 days (Fig. 8). This cohort was present until May 1992 and was largely absent thereafter (Fig. 7). A third cohort was first observed in mid October 1992. It measured from 50 to 100 mm in length, overlapping the length distribution of cohort 1. As such, no clear distinction could be made between these two cohorts on these dates. The third cohort became more dis- tinct in December 1992 and January 1993 as individuals from the first cohort left the lagoon. These individuals were aged from 95 to 200 days old in December. This cohort was present until approximately April 1993 and disappeared thereafter (Fig. 9). Finally, a fourth cohort first appeared in March 1993 and consisted offish of similar size and age as the first cohort observed in March of the previous year. The back-calculated hatching dates showed that the sec- ond, older cohort observed in the lagoon in March 1992 was composed of mullet that had hatched between August and October 1991 (Fig. 10). The younger cohort in the March 1992 sample originated from continuous hatching from Table 1 Validation of daily increment formation in the otoliths of juvenile white mullet (Mugil curema) sampled at 14-day intervals in March and April 1992. /i= number of otoliths in sample, mean age (standard deviation) is given in days on date of capture, and "difference" is the difference in mean age between successive sampling dates. Sampling dates 5 March 1992 19 March 1992 2 April 2 1992 15 20 19 56.46(4.81) 70.85(9.88) 84.3(11.24) 14.39 13.45 Mean age (SD) Difference late December 1991 to late March 1992. The back-calculated hatching dates indicated that larval production of successful recruits was almost absent during April and May 1992 but small increases were observed in June and July 1992. The third cohort, which first appeared in September 1992, was 816 Fishery Bulletin 101(4) jLil 50 ST MAMJJASOMDJFMAMJJ 1992 1993 Figure 7 Histogram of monthly abundance of juvenile white mullet as indicated by catch per unit of effort (CPUE) from semimonthly sampling in La Restinga lagoon from March 1992 to July 1993. The continuous black line presented in the first annual cycle represents the mean monthly rainfall averaged over the period 1985-93. Table 2 Correlation analyses of hatching weakens the r^ values between contrast, increasing lambda va' BD = birth date, UI = upwelling date and upwelling index data series. Increasing lambda values abserved and predicted values of birth date and upwelling index ues strengthen the correlation (Pearson's correlation) between index used to fit the over time (col the two data smoothing spline umns 2 and 3). In series (column 4). Data series Hatching dates (r2, P) Upwel! (r ing index 2,P) BD versus UI (r,P) Raw data 0.28, <0.05 Smoothing spline A=l 0.82 <0.000 0.86, <0.000 0.36, <0.000 Smoothing spline A= 10 0.73, <0.000 0.74, <0.000 0.41, <0.000 Smoothing spline A= 100 0.67, <0.000 0.63, <0.000 0.45, <0.000 Smoothing spline A = 1000 0.64, <0,000 0.54, <0.000 0.52, <0.000 Smoothing spline A = 10,000 0.56, <0.000 0.45, <0.000 0.57, <0.000 Smoothing spline A= 100,000 0.48, <0.000 0.40, <0.000 0.64, <0.000 composed of individuals that had hatched between June and August 1992. Finally, individuals in the fourth cohort, which first appeared in the lagoon in March 1993, were fish that had hatched in January and February 1993. The hatching dates of recruits coincided with periods of increasing upwelling, particularly during January and February of 1992 and 1993 (Fig. 10). The use of increa.sing levels of lambda to fit the smoothing spline increasingly weakened the correlations between predicted and observed values of birth dates and UI index and strengthened the correlations between birth dates and the upwelling index (Table 2). Choosing a lambda value of 10 resulted in r'^ values greater than 0.70 (/'<0.000) between the raw and predicted data series and in an rvalue of 0.41 (P<0.000) between birth date and the upwelling index. Cross-corre- lation analysis between these two series revealed that the strongest correlation (r=0.52, P<0.000) occurred when the upwelling index lagged behind birth dates by 8 days and by 46 days. These lag periods reflect the coincidence of the major peak of upwelling with the two peaks of birth dates that are separated by approximately 35 days. Given the estimated 10-day error associated with aging otoliths, the 8-day lag cannot be interpreted. Discussion Our sampling of white mullet in the coastal waters of northeastern Venezuela revealed the presence of mature fish throughout the year, but abundance was lowest Mann et al,: Reproduction and recruitment of Mugil curema 817 between August and January. Mullet from the Margarita zone were small (4 to 36 cm in SL) and mostly immature (>80%). Because size at maturity of white mullet is 24 cm (Marin and Dodson, unpubl. data), most of the adults in the lagoon were probably in their first spawn- ing cycle. Similarly, mullet from the Cariaco Gulf zone also appeared to be young adults in their first spawning cycle. In contrast, mullet from the Chacopata zone were larger and generally in more advanced stages of gonadal development. This finding suggests that the Chacopata mullet were part of a prespawning aggregation, and the location of the aggrega- tion agrees with the more offshore location of the Chacopata fishery. Because white mullet spawn offshore (Ja- cot, 1920; Anderson, 1957; Ditty and Shaw, 1996), the small proportion of mature fish in the coastal fisheries from July to April is likely explained by the migration of adults to the offshore spawning grounds. If this is so, reproduction is not indicated by an increase in the frequency of fish in advanced stages, but rather is associated with the disappearance of maturing and mature fish from coastal areas. The disappearance of fish in advanced stages from coastal areas as the spawning season ap- proaches was also reported by Angell (1973) and Moore (1974). The analysis of birth dates of juveniles sampled in the La Restinga La- goon indicates that successful spawnings are concentrated in the periods of increased up- welling and also coincide with the end of the rainy season. The spawning season may or may not be concentrated at these times but larvae that hatch during upwelling events are most likely to successfully recruit to the lagoon. Although reproduction in tropical fishes is often protracted, peaks in successful spawning may nevertheless be initiated by environmental cues (Redding and Patino, 1993). The white mullet possibly uses tem- perature or other signals associated with upwelling to synchronize its spawning with upwelling events. The variations in the tim- ing of recruitment of white mullet in differ- ent geographical regions may be the result of variation in the timing of favorable condi- tions that enhance survival. Such conditions may include increased primary production (Ferraz-Reyes, 1983; Miiller-Karger et al., 1989) or hydrographic mechanisms likely to facilitate transport of larvae to coastal nurs- ery areas (Blaber, 1987), so that survival is increased. Populations likely have adequate time to adapt to environmental conditions in particular areas because local hydrographic patterns develop over geological time scales (Bakun, 1986; Sinclair, 1988; Heath, 1992). Uj a. o jlllu. __1M. miiL noc n = 416 iDnDnc JlDnnnnnn n n n _dQ jmd jaa n = 103 jm nn_ nnnr,n nnn^nnnnnOnn r. n = 88 ,nnnn) nflnn Innr _la_ DnJl n = S7 -a □ □. Dnn nHn- .. JUMbIjuu^ wn-.-n sx^M ■■■■■■111. ■■■■■Illll- ■ ■■ III Uihi. D n JUlL II 1 n = 4 ill n = 104 Uu n = 415 JlIlL n = 252 Jim iiliiiii. ilii ..III lUlllliIh^ ^■lIUImlBllkMlL JUI.Ulill. -■Illl n = 66 Mar-05 Mar-19 Apr-02 Apr-14 Apr-30 May-14 May-28 Jun-11 Jun-25 Jljl-08 Jul-23 Aug-05 Aug-20 Sep-03 Sep-17 Oct-02 OcM5 Ocl-29 Nov-10 Nov-26 Dec- 10 Dec-28 Jan-08 Jan-21 Feb-04 Mar-04 Mar-18 Apr-01 Apr-15 Apr-29 May-13 May-25 Jun-10 Jun-25 Jul-08 Jul-22 1992 1993 10 30 50 70 90 110 130 150 170 190 210 Standard length (mm) Figure 8 Semimonthly size distributions of juvenile white mullet from March 1992 to July 1993 in La Restinga Lagoon. The abundance for each size class is presented as the Log CPUE + 1. From top to bottom, open bars represent cohort 1, black bars represent cohort 2, gray bars represent cohort 3, and black bars represent cohort 4. Cohorts were identified by discontinuities in size distributions of juveniles, n = number offish sampled. 818 Fishery Bulletin 101(4) Strong offshore transport of surface waters occurs during upwelling events, so that a rapid metamorphosis to the demersal stage may be critical for the coastal recruitment of white mullet. This rapid metamorphosis is suggested for several offshore spawning fishes with pelagic larvae that subsequently recruit to estuaries (Creutzberg et al., 1978; Heath, 1992) or that remain near the bottom during ebb flow, once close to the coast, thereby reducing offshore transport (Bartsch and Knust, 1994). White mullet undergo metamorpho- sis to the demersal stage 14 days after hatching (Houde et al., 1976) at which time they would be entrained in the inshore transported water that occurs at depths greater than 50 m in the coastal zone of northeastern Venezuela (Quintero, unpubl. data). Several studies suggest that increased mor- tality is caused by increased predation associated with the change to bottom habitat (Johannes, 1978; Bakun, 1986). Heath (1992) suggested that mortality from predation is particularly high dur- ing migration to nursery areas. Given the time to metamorphosis (14 days) and the age of white mullet when they enter the lagoon (50 to 70 days for the first cohort), metamorphosis to the demer- sal stage most probably occurs at least one month before entry into the lagoon (Anderson, 1957; Caldwell and Anderson, 1959; Yanez-Arancibia, 1976;Vieira, 1991). During the demersal period at sea, white mullet may be exposed to considerable mortality due to benthic predators. Variation in the abundance of recruitment pulses into La Restinga Lagoon may reflect the interplay between spawning time and the mortality during transport to the coastal area. At some point between metamorphosis and lagoon entry, juvenile mullet also develop active swim- ming behavior to facilitate passive transport. We observed intensive recruitment of small mullet into the lagoon between March and June by indi- viduals that had hatched the previous December to February. The timing of their hatching means that their return to the lagoon was likely facilitat- ed by prevailing currents. In contrast, recruitment of mullet to the lagoon over the remainder of the year was weak and sporadic, and fish were much larger and older At its first appearance in the lagoon, the third cohort was twice the age of the first cohort. These fish were not produced during a period when currents would likely have facilitated lar\'al transport to the lagoon (little upwelling) and their lower densities may partially reflect increased mortality during the more prolonged return to the lagoon. We propose that spawning during periods of weak upwelling causes a delay in transport to coastal nursery areas and conse- quently decreased survival. Periods of hatching leading to successful recruitment, from late December to March, coincided with moderate peaks in the upwelling index. This successful recruitment JL I ■ ii I n nnlllLnn . I. . _oDoolUnOlk_ Mn .ndnnn Jl I —O a II nn n nHI n I ru-n .,~,n^ ntlnnnn UnO n W nnrlln, jJlu . on n rinnii n„ I .1.1 Jllll_ _llllll„ juL n = 32 Mar-05 n = 27 Apr-02 n = 22 Apr-30 n = 20 May-28 n = 23 Jun-25 n = 22 Jul-23 n= 16 Aug-20 n = 20 Sep-17 n = 20 Oct-15 n=19 Nov-10 n = 25 Dec-10 n = 27 Jan-08 n=14 Feb-04 n = 6 Mar-18 n= 19 Apr-15 n = 27 May-13 n= 17 Jun-10 n= 16 Jul-08 40 80 120 160 200 240 280 320 360 400 Age (days) Figure 9 Monthly age distributions of juvenik' white mullet from March 1992 to July 199.3. Age wa.s calculated by counting daily growth ring.s start- ing with the hatch mark. From top to bottom, open bars represent cohort 1, black bars represent cohort 2, gray bars represent cohort 3 and black bars represent cohort 4. n = number of otoliths examined. may be the result of moderate levels of wind speed (<6 m/s-), that promote moderate upwelling and yield optimal trophic conditions for fish larvae (Cury and Roy, 1989). Coastal upwellings in northeast Venezuelan waters are caused by Mann et a\ ; Reproduction and recruitment oi Mugil curema 819 Hatching date ooooooooooooooooooo Date Figure 10 (Upper panel) Frequency distribution of hatching dates for juvenile white mullet recruits caught in 1992 and 1993 in La Restinga lagoon. Points represent raw data. Continuous black line represents the trend line generated after application of a smoothing spline (lambda=10). (Lower panel) Upwelling index (UI) for northeastern Venezuela calculated from August 1991 to March 1993. See text for description of cal- culation. Points represent raw data. Continuous black line represents the trend line generated following application of a smoothing spline (lambda=10). moderate levels of wind stress and this differs from the strong upwellings observed in such places as Peru and Sen- egal. The relation between the timing of successful spawn- ing and the intensity of coastal recruitment in white mullet is likely due to variations in the duration of the transport of larvae and juveniles to the shore as a result of varying current conditions as well as variations in food availability for first-feeding larvae. Acknowledgments This work is part of a Ph.D. thesis submitted to Laval Uni- versity by the senior author who was financially supported by the Fundayacucho Program of Scholarships. We thank the technical and analytical assistance of the zooplankton staff of lOV-UDO (Institute Oceanografico de Venezuela, Universidad de Oriente), Domingo Figueroa and Rafael Briceno, and Caroline Berger for her work in the field survey, figure preparation and the processing of otoliths with Julie Paquet. Martin LLano, of the Metereological Station of Fundacion La Salle, kindly supplied the enviro- mental data. We thank Luis Trocolis, Jose Luis Fuentes, and Alfredo Gomez from de Edimar-UDO, Nueva Esparta, for laboratory support, Jose Bechara, Casimiro Quinones, and Jean Paul Boulianne for advice and statistical support and Idelfonso Liheros and Jesus Marcano for support and com- plementary information. This work was funded by grants from NSERC (Natural Sciences and Engineering Research Council of Canada) and FCAR (Fonds pour la Formation des Chercheurs et L'Aide a la Recherche, Quebec, Canada) to Julian J. Dodson and GIROQ (Groupe Interuniversitaire de Recherches en Oceanographie du Quebec) and by the Consejo de Investigacion-Universidad de Oriente, proyecto LISA-92 (CI-5-019-00554/92). Literature cited Alvarez-Lajonchere, L. 1976. Contribucion al estudio del ciclo de vida de Mugil curema Valenciennes (Pisces: Mugilidae). Cien. Ser. 8 Investig. Mar (Havana) 28:3-130. 1980. Composicidn per especies y distribucion de las post- larvas y juveniles de Lisas (Pisces, Mugilidae) en tunas de Zaza, Cuba. Rev. Investig. Mar. Cuba l(2-3):28-60. 1982. The fecundity of mullet (Pisces, Mugilidae) from Cuban waters. J. Fish. Biol. 21:607-613. Anderson, W. W. 1957. Early development, spawning, growth and occurrence of the white mullet (Mugil curema) along the south Atlantic 820 Fishery Bulletin 101(4) Coast of the United States. Fish. Bull.Wildl. Serv. 57(120): 415-425. Angell, C. 1973. Algunos aspectos de la biologia de la lisa, Mugil curema Valenciennes, en aguas hipersalinas del nororiente de Venezuela. Contrib. Fund. La Salle. Cienc. Nat. 51: 223-238. Bakun, A. 1986. Local retention of planktonic early life stages in tropi- cal reef bank demersal systems: the role of vertically-struc- tured hydrodynamic processes. In lOC/FAO workshop on recruitment in tropical coastal demersal communities, Ciudad del Carmen, Campeche, Mexico. Workshop Rep. 44(suppl), p. 16-32. Bakun, A., D. R. McLain, and F. V. Mayo. 1974. The mean annual cycle of coastal upwelling off west- ern North America as observed from surface measurements. Fish. Bull. 72:843-844. Bartsch, J., and R. Knust. 1994. Simulating the dispersion of vertically migrating sprat larvae [Spratlus sprattus) in the German Basin with a circulation and transport model system. Fish. Oceanogr. 3:92-105. Blaber, S. J. M. 1987. Factor influencing recruitment and survival of mugilids in estuaries and coastal waters of Southeastern Africa. In Common strategies of anadromous and catadromous fishes (M. Dadswell, R. Klauda, C. Saunders. R. Rulifson, and J. Cooper, eds.), p. 507-518. Am. Fish. Soc.Symp. 1? Blaber, S. J. M., and T. G. Blaber 1980. Factors affecting the distribution of juvenile estua- rine and inshore fish. J. Fish. Biol. 17:143-162. Bowden, K. F. 1983. Physicaloceanography of coastal waters, 302 p. Ellis Horwood Ltd., Chichester, UK. Caldwell, D. K., and W. W. Anderson. 1959. Offshore occurrence of larval white mullet, Mugil curema, in the western Gulf of Mexico. Copeia 1959:252. Campana, S. E., and J. D. Nielsen. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Creutzberg, F, A. T. G. W. Eltink, and G. J. van Noort. 1978. The migration of plaice larvae Pleuronectes platessa in the western Wadden sea. In Physiology and behaviour of marine organisms (D. S. Lunsky and A. J. Berry, eds.), p. 243-251. Pergamon Press, New York, NY. Cury, P, and C. Roy. 1989. Optimal environmental window and pelagic fish recruitment success in upwelling areas. Can. J. Fish. Aquat. Sci. 46:670-680, Ditty, J. G., and R. F Shaw. 1996. Spatial and temporal distribution of larval striped mullet {Mugil cephalus) and white mullet (A/, curema. family: Mugilidae) in the northern Gulf of Mexico, with notes on mountain muWet, Agonostomus monticola. Bull. Mar. Sci. 59:271-288. Etchevers, S. L. 1974. Fecundidad de la lisa {Mugil curema Valenciennes) en el Oriente de Venezuela. Bol. CientiT Tdc. Ser. Recur. Mar. CIC UDO (Centre de Ingenieria y Computacidn, Universi- dadde Oriente) 1(1), 19 p. Ferraz-Reyes, E. 1983. Estudio del fitoplancton de la Cuenca Tuy-Cariaco, Venezuela. Bol. Inst. Oceanogr. Venez.Univ.Oriente. 22(1/2): 111-124. 1989. Influencia de los factores fisicos en la distribucion ver- tical de la biomasa fitoplanctonica en el Golfo de Cariaco (Venezuela). Bol. Inst. Oceanogr. Venez. Univ. Oriente. 28(l/2):47-56. Ferraz-Reyes, E., E. Mandelli, and G. Reyes-Vasquez. 1987. Fitoplancton de la Laguna Grande del Obispo, Ven- ezuela. Bol. Inst. Oceanogr. Venez. Univ. Oriente. 26(1/2): 111-124. Franco, L. 1986. Biologia y reproduccion de la lisa, Mugil curema Valenciennes (Pisces: Mugilidae) en el Golfo de Cariaco, Venezuela. M.S. thesis, 103 p. Instituto Oceanografico de Venezuela, Univ. Oriente, Cumana, Venezuela. Garcia, A., and G. Bustamante. 1981. Resultados preliminares del desove inducido de lisa {Mugil curema Valenciennes) en Cuba. Acad. Cienc. Cuba Inf Cient.-Tec. 158:7-26. Gines, H. 1972. Cartas pesqueras de Venezuela. Caracas, 328 p. Fundacion La Salle de Ciencias Naturales. M.E.L.S.A., Madrid, Venezuela. Gomez, A. 1983. Pigmentos clorofilicos, produccion primaria y abun- dancia plantonica en el canal de entrada de la Laguna de la Restinga, Venezuela. Bol. Inst. Oceanogr. Univ. Oriente 22:43-64. Gomez, A., and F Cervigon. 1987. Perspectivas del cultivo de peces marines en el Caribe Sur y noreste de Suramerica. Rev. Latinoam. Acuicult. 34: 40-50. Heath, M. R. 1992. Field investigations of the early life stages of marine fish. Adv. Mar. Biol. 28:1-173. Houde, E. D., S. A. Berkley, J. J. Klinovsky, and R. C. Schekter. 1976. Culture of larvae of the white mullet, Mugil curema Valenciennes. Aquaculture 8:36.5-370. Ibaiiez-Aguirre, A. L. 1993. Coexistence de Mugil cephalus and M. curema in a coastal lagoon in the Gulf of Mexico. Mar. Biol. 42:959-961. Jacot, A. R 1920. Age, growth and scale characters of the mullets, Mugil cephalus and Mugil curema. Trans. Am. Microsc. Soc. 39: 199-229. Jenkins, G. P., and H. M. A. May. 1994. Variation in settlement and lai-val duration of King George whiting, Silloginodes punctata (Sillaginidae), in Swan Bay, Victoria, Australia. Bull. Mar. Sci. 54(1): 281-296. Johannes, R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fishes 3:65-84. Jordan, A. R. 1993. Age, growth and back-calculated birthdate distribution of lan'al jack mackerel, TYachurus declives (Pisces: Caran- gidae) from eastern Tasmanian coastal waters. Aust. J. Mar Freshw. Res. 45:19-33. Li, C, S. Xueshen, Y Feng, Y Chunwu, and H. Ruidong. 1993. Daily growth increments in otoliths of mullet larva, Mugil so-iuy Basilewsky, and determination from field-col- lected ones. Oceanol. Limnol. Sin. 24(4):345-349. McBride, R. S., and D. O. Conover 1991. Recruitment of young-of-the-year blucfish Pomatus saltatrix to the New York Bight: variation in abundance and growth of spring- and summer-spawned cohorts. Mar Ecol. Prog. Ser. 78:205-216. Moore, R. H. 1974. General ecology, distribution and relative abundance Mann et a\ : Reproduction and recruitment of Mugil curema 821 of Mugil cephalus and Mugil curema on the south Texas coast. Contr. Mar. Sci. 18:241-255. Miiller-Karger, F. E., C. R. McClain, T. R. Fisher, W. E. Esaias, and R. Varela. 1989. Pigment distribution in the Caribbean Sea: observa- tions from space. Prog. Oceanogr. 23:23-64. Neilson, J. D., and G. H. Geen. 1986. First-year growth rate of Sixes R. chinook salmon as infered from otohths: effects on mortality and age at maturity. Trans. Am. Fish. Soc. 115:28-33. Okuda, T., J. Benitez, and G. Cedeno. 1978. Caracteristicas hidrograficas del Golfo de Cariaco, Venezuela. Bol. Inst. Oceanogr. Venez. Univ. Oriente 17: 69-87. Panella, G. 1971. Fish otoliths: daily growths layers and periodic patterns. Science. 173:1124-1127. Radtke, R. L. 1984. Formation and structural composition of larval striped mullet otoliths. Trans. Am. Fish. Soc. 113:186-191. Redding, M., and R, Patifio. 1993. Reproductive physiology. In Fish physiology (E. Evans, ed. ), p. 503-534. ORG Press, Boca Raton, FL. Rodriguez, C. L., and I. V. do Nacimiento. 1980. Estudio microscbpico dos ovarios de Mugil curema Valenciennes, Brasil. In I simposio Brasileiro de aqua- cultura, Recife, 1978, p. 213-219. Academia Brasileira de Ciencias, Rio de Janeiro, Brazil. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstructural examination. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds. ), p. 9-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Simonoff J. S. 1996. Smoothing methods in statistics. Springer-Verlag, New York, NY. Sinclair, M. 1988. Marine populations: an essay on population regulation and speciation, p. 252. Univ. Washington Press, Seattle, WA. Sirois, P., and J. J. Dodson. 2000. Critical periods and growth-dependant survival of larvae of an estuarine fish, the rainbow smelt Osnierus mordax. Mar Ecol. Prog. Ser 203:233-245. Struhsaker, P., and J. H. Uchiyama. 1976. Age and growth of the nehu, Stolephorus purpureus (Engraulidae), from the Hawaiian islands as indicated by daily growth increments in sagittae. Fish. Bull. 74:9-17. Vieira, J. P. 1991. Juvenile mullets (Pisces: Mugilidae) in the estuary Lagoa dos Patos, RS, Brazil. Copeia 1991:409-418. Wilson, K. H., and R A. Larkin. 1980. Daily growth rings in the otoliths of juvenile sockeye salmon Oncorhynchus nerka. Can. J. Fish. Aquat. Sci. 37: 1495-1498. Yanez-Arancibia, L. A. 1976. Observaciones sobre Mugil curema Valenciennes, en areas naturales de crianza, crianza, maduracibn, creci- miento, madurez y relaciones ecologicas. Ann. Cent. Cienc. Mar Limnol Univ Nac. Auton. Mexico 2:211-243. 822 Abstract— Fecundity in striped mullet (Mugil cephalus) from South Carolina correlated highly with length and weight, but not with age. Oocyte counts ranged from 4.47 x 10'^ to 2.52 x 10^ in 1998 for fish ranging in size from 331 mm to 600 mm total length, 2.13 x 105 to 3.89 X 10« in 1999 for fish ranging in size from 332 mm to 588 mm total length, and 3.89 x 10= to 3.01 x lO'' in 2000 for fish ranging in size from 325 mm to 592 mm total length. The striped mullet in this study had a high degree of variability in the size-at-age relation- ship; this variability was indicative of varied growth rates and compounded the errors in estimating fecundity at age. The stronger relationship of fecun- dity to fish size allowed a much better predictive model for potential fecundity in striped mullet. By comparing fecun- dity with other measures of reproduc- tive activity, such as the gonadosomatic index, histological examination, and the measurement of mean oocyte diam- eters, we determined that none of these methods by themselves were adequate to determine the extent of reproductive development. Histological examinations and oocyte diameter measurements revealed that fecundity counts could be made once developing oocytes reached 0.400 nm or larger. Striped mullet are isochronal spawners; therefore fecundity estimates for this species are easier to determine because oocytes develop at approximately the same rate upon reaching 400 idm. This uniform development made oocytes that were to be spawned easier to count. When fecundity counts were used in conjunc- tion with histological examination, oocyte diameter measurements, and gonadosomatic index, a more complete measure of reproductive potential and the timing of the spawning season was possible. In addition, it was determined that striped mullet that recruit into South Carolina estuaries spawn from October through April. Fecundity and spawning season of striped mullet {Mugil cephalus L.) in South Carolina estuaries* Christopher J. McDonough William A. Roumillat Charles A. Wenner Marine Resources Research Institute South Carolina Department of Natural Resources 217 Fort Johnson Road Charleston, South Carolina 29422-2559 E-mail address (for C J McDonougfi): mcdonoughc@mrd.dnrstate.sc,us Manuscript approved for publication 19 June 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish Bull. 101:822-834 The striped mullet iMugil cephalus L.) is distributed circumglobally in tropi- cal and semitropical waters between latitudes 42°N and 42°S (Thompson, 1963; Rossi et al., 1998). Even though considered a marine species, striped mullet are euryhaline and can be found year round throughout the full range of estuarine salinities in the southeastern United States (Jacot, 1920; Anderson, 1958). Striped mullet are a commer- cially important fish throughout the world sustaining both fisheries and aquaculture industries. In the south- eastern United States (North Carolina and Florida) there are significant com- mercial fisheries for striped mullet, whereas in South Carolina and Georgia landings are more limited (NMFS'). The primary fishery in most of these states is for "roe" mullet during the fall spawning migration. Throughout the rest of the year mullet are fished commercially for bait, if they are fished at all (Anderson, 1958). Striped mullet have a significant economic impact in the areas where they are more heavily fished by the commercial fisheries and the landings of this species from 1994 to 1998 yielded a landings (wholesale) value of 38.2 million dollars. Striped mullet are also one of the most impor- tant forage fishes that are found in the estuaries of the southeastern United States and represent a significant food source for upper-level piscivores (Wenner et al.^). The biological features of striped mullet has been well documented (Jacot, 1920; Anderson, 1958; Thomson, 1963, 1966; Chubb et al., 1981), but much less information is available on the biological aspects of reproduction in the wild (Anderson, 1958; Stenger, 1959; Greeley et al., 1987; Render et al, 1995). There is a large body of work con- cerning striped mullet reproduction in aquaculture but many of these studies have used artificial manipulation of the reproductive cycle. Although the matu- ration process of oocytes may have been the same as that in wild striped mullet, the environment and conditions under which maturation occurs were artifi- cial (Shehadeh et al., 1973a; Kuo et al., 1974; Pien and Liao, 1975, Kelly, 1990; Tamaru et al., 1994; Kuo, 1995). In ad- dtion, despite the demonstrated ability to initiate reproduction (both in and out of season) for striped mullet, the majority of aquaculture studies have had to rely on wild fish for broodstock (Kuo et al., 1974; Pien and Liao, 1975; Kuo, 1995). * Contribution 522 of the Marine Resources Institute, South Carolina Dept. of Natural Resources, Charleston, SC 29422-2559. ' National Marine Fisheries Service. 2001. Personal commun. Statistics and Eco- nomic Division, 1315 East-West Highway, Silver Spring, Md. 20910. http://www. St. nmfs.gov/stl/index. html. 2 Wenner, C. A., W. A. Roumillat, J. E. Moran, M. B. Maddox, L. B. Daniel, and J. W. Smith. 1990. Investigations on the life history and population dynamics of ma- rine recreational fishes in South Carolina, part 1. Completion report, project F-37, p. 3-13 and project F-31, p. 6-35. South Carolina Marine Resources Research Institute, P.O. Box 12559, Charleston, SC 29422 McDonough et al : Fecundity and spawning season of Mugil cepha/us 823 In the southeastern United States the spawning season lasts from two to five months depending on the coastal area involved (Jacot, 1920; Broadhead, 1956; Anderson, 1958; Arnold and Thompson, 1958; Stenger, 1959; Dindo and MacGregor, 1981; Greeley et al., 1987; Render et al., 1995; Hettler et al., 1997). Striped mullet are considered isochronal spawning fishes (Greeley et al., 1987; Render et al., 1995), i.e. they have synchronous gamete development and individuals spawn all their reproductive material at once or in batches over a very short time period (days, as opposed to weeks). There have been limited observations of offshore spawning activity (Arnold and Thompson, 1958), and few examples of eggs and larvae collected offshore (An- derson, 1958; Finucane et al., 1978; Collins and Stender, 1989). Collins and Stender (1989) concluded that striped mullet spawn in and around the edge of the continental shelf off the coasts of North Carolina, South Carolina, Georgia, and the east coast of Florida (an area often re- ferred to as the South Atlantic Bight) and have a protracted spawning season from October to April. This spawning season contrasts with that estimated by most other stud- ies (Jacot, 1920; Broadhead, 1956; Anderson, 1958; Arnold and Thompson, 1958; Stenger, 1959; Dindo and MacGregor, 1981; Greeley et al., 1987; Render et al., 1995; Hettler et al, 1997 ). These studies based their estimates on the reproduc- tive condition of migrating adults and the subsequent re- cruitment of juvenile fish back into coastal estuaries — not on actual data on the offshore presence of striped mullet larvae. Female mullet have been shown to mature at two to three years of age at a size range from 230 mm to 350 mm standard length (Thomson, 1951, 1963; Greeley et al., 1987). Determination of spawning activity in mullet has been estimated by using gonadosomatic indices (Dindo and MacGregor, 1981; Render et al., 1995), examination of oo- cyte size and maturity stage (Stenger, 1959; Greeley et al., 1987 ), and by the presence of enlarged, developing ovaries in migrating fish (Jacot, 1920; Anderson, 1958). The purpose of the present study was to develop size- or age-related estimates of fecundity in striped mullet from South Carolina estuaries. These fecundity estimates were determined in order to develop models for estimating potential fecundity from catch data, such as length and weight data. In addition, other indicators of reproductive activity such as gonadosomatic indices and oocyte size were examined to provide information on the duration of the spawning season. Potential fecundity estimates can give a barometer of reproductive potential based on morphologi- cal information from catch curve data and size-frequency distributions. Materials and methods Striped mullet were collected monthly from January 1998 through December 2000 by using a randomly stratified sampling regime within three different estuarine systems along the South Carolina coast: (Ashepoo-Combahee- Edisto (ACE) Basin, Charleston Harbor, and the Cape Remain estuary, Fig. D.The female striped mullet used for fecundity estimates were collected from October through February each year because these were the only months when fecund females were present. Fecund female striped mullet were defined as specimens in the tertiary stage of vitellogenesis with oocyte diameters greater than 400 ^m and that had gonadosomatic indices greater than 5.0. The vitellogenic stage for each specimen used in the fecundity estimates was confirmed histologically. Fish were captured during daylight ebbing tides with water levels ranging from 0.3 to 2.0 meters, and the majority (67.8%) were caught during late ebb. The fish were primarily caught with a 184-m trammel net with an outside stretch mesh of 350 mm and an inside stretch mesh of 63.5 mm, although a few of the 1999 samples (5) and the 2000 samples (3) were obtained by using a cast net 1.8 m in diameter and equipped with 10-mm mesh. Eight fecund females were also captured by using an electroshock boat in the fresh- water and low-salinity areas of the Cooper River (one of the three rivers that make up the Charleston Harbor Estuary) in October 2000. The areas included in the present study have been sampled on a monthly basis since 1991 with trammel nets by the Inshore Fisheries Group of the South Carolina Department of Natural Resources as part of a gamefish monitoring program. During this period repro- ductively developing striped mullet of both sexes were generally observed from October through February and were presumably heading offshore to spawn (Jacot, 1920; Anderson, 1958; Arnold and Thompson, 1958). Male striped mullet that were reproductively developed were easy to discern because they were usually leaking milt and were not analyzed further. All other fish were brought back to the laboratory and eviscerated in order to determine sex and to collect ovaries for reproductive analysis. All of the samples were kept on ice and were generally examined within twenty-four hours of capture. Standard morphometric measurements included total length (TL) in mm, fork length in mm (FL), and standard length in mm (SL), total weight (TW) in grams, ovary weight (OW) in grams, sex, and maturity. Body weight (BW) was calculated as total weight minus ovary weight {BW=TW-OW). Saggital otoliths were removed for aging. A small section of the posterior end of the ovaries, at the junction of the two lobes, was also removed for histo- logical examination. The whole ovaries were fixed in 10% seawater-buffered formalin and the histological sample was fixed in 10% neutral-buffered formalin. Histological samples were processed by using standard procedures for paraffin embedding and sectioning (Humason, 1967). The sections were dried on slides and stained by using standard haematoxylin and eosin-Y staining techniques (Humason, 1967). Examination of the histological sections for maturity stage was done by using a compound microscope at lOOx magnification. Each histological section was evaluated by two separate readers to determine agreement on maturity stage. If there was a discrepancy in maturity staging for any specimen, the discrepancy was either resolved by the two readers or that specimen was not used in the analysis. There were no discrepancies between readers on any of the reproductively developing females. Maturity was as- sessed according to a modified version of the schedule used by Wenner et al. (1986) adapted to work with isochronal 824 Fishery Bulletin 101(4) 80° 30' 80° 00 80° 30' 80° 00' 79° 30' Figure 1 The South Carolina coast with the locations of estuaries and major river basins where striped mullet were collected from 1998 to 2000. spawning fish, as well as previous models of reproductive development (Stenger, 1959; Wallace and Selman, 1981) (Table 1). This evaluation method was based on identifica- tion of morphological characteristics evident in histological sections. Each specimen was evaluated by two (in some cases three) readers and discrepancies between readers were either resolved or the specimen was excluded from the analysis. A gonadosomatic index (GSI) was calculated for each specimen following the method of Render et al. (1995) where GSI was expressed as gonad weight (GW) divided by body weight (BW) such that GSI = (GWIBW) X 100. The GSI values of the fecundity specimens were compared among the three sampling years, as well as with GSI values for female striped mullet collected during the rest of the year that were mature but not reproductively active. Fecundity determinations were made from a total of 129 advanced-stage developing ovaries; 50 from 1998, 37 from 1999, and 42 from 2000. All of the ovaries were determined by histological examinations and criteria outlined in Table 1 to be actively vitellogenic, having tertiary yolk-stage oocytes. All oocytes counted for fecundity were 400 /jm or larger and in the tertiary yolk stage (Pien and Liao, 1975) (Fig. 2). This 400-/jm threshold has also been used in other studies of oocyte development in striped mullet for determining the point at which the oocytes that would be spawned during that year were identifiable (Shehadeh et al., 1973a). Unlike other species (particularly batch spawn- ers), where fecundity counts should ideally be conducted by using hydrated oocytes only, the striped mullet oocytes used for fecundity counts were still presumably several weeks from hydration and spawning. However, because mullet are synchronous spawners, it is relatively easy to distinguish the developing oocytes from the undeveloped ones because of the drastic difference in size between the two, as well as by the uniformity in size of the developing oocytes once they reach 400 pm. Fecundity was estimated by using a modified gravimet- ric method. The fixed whole ovary was patted dry and re- weighed. The ovarian lobes were divided into four discrete regions along each lobe's longitudinal axis and three sub- samples (chosen at random) were taken between the two lobes and prescn'ed in 50% isopropyl until oocyte counts could be conducted. The subsamples ranged in weight from 0.025 g to 0.033 g. The subsamples from each specimen McDonough et al.: Fecundity and spawning season of Mugil cephalus 825 Table 1 Histological criteria used to determine reproductive stage in female striped mullet {Mugil cephalus) once sexual differentiation has occurred (Wenner et al.: see footnote 2 in the general text). Reproductive stage Description 1. Immature 2. Developing 3. Running ripe 4. Atretic or Spent 5. Inactive or Resting Inactive ovary with previtellogenic oocytes and no evidence of atresia. Oocytes are < 80 ^m, lamellae still contain somatic and connective tissue bundles. Ovary wall is very thin (one or two cell layers). Developing ovary have enlarged oocytes generally greater than 120 /jm in size. Cortical alveoli are present and actual vitellogenesis occurs after oocytes reach 180 /jm in size and continue to increase in size. Abundant yolk globules with oocytes reaching a size of >600 pm. Completion of yolk coalescence and hydration in most oocytes. More than 30% of developed oocytes undergoing the atretic process. Previtellogenic oocytes only but traces of atresia possible. In comparison to immature females, most oocytes are >80 ^m, lamellae have some muscle and connective tissue bundles. Lamellae are larger, have moore oocytes, and are elongated. A thicker ovarian wall with blood vessels, muscle, and nerve tissue. were then teased apart. After separation, the oocytes were spread out on a Bogorov tray and counts of ooc5ftes, greater than 400 |jm, were made by using a dissecting microscope at 12x magnification. Each subsample was counted twice and counts were averaged. A third count was performed if the first two counts differed by more than 10%. Oocyte density was calculated by dividing the mean number of oocytes by the mean weight of all three subsamples for each specimen. The oocyte density was then used to calculate the total oocyte number for each ovary, or individual fecundity, by multiplying mean oocyte density by whole ovary weight. In order to determine mean oocyte diameter for each specimen, 20-30 oocytes were removed from each counted subsample and grouped together in a petri dish. Each oo- cyte was then measured along the longest axis by using Optimas''''^ Image Analysis software (version 6, Media Cybernetics, Bothell,WA). Mean oocyte diameter was calculated as the average of all measurements for each 826 Fishery Bulletin 101(4) ou - ^m 1998 1 1 1999 25 - ^m 2000 20 - 15 - 10 - 1 5 - 1 IJ •1 I 1 I 300-350 351-400 401-450 451-500 501-550 551-600 Size class (mm) Figure 3 Size-frequency distribution of female striped mullet used in fecundity determinations from South Carolina estuaries from 1998 to 2000. n (number offish in sample) = 50 fish for 1998, ri = 37 for 1999. and n = 42 for 2000. subsample. The overall mean oocyte diameter for each specimen resulted from the calculated average of the means of the three subsamples. Measurements were not made on fresh oocytes but shrinkage was estimated from the amount of whole ovary shrinkage because fresh ovary weight and preserved (in 10% formalin) ovary weight were known. The estimated unpreserved oocyte diameter was determined by multiplying the preserved oocyte diameter by 1 and adding the percentage of ovary shrinkage. The dif- ference in the preserved oocyte diameter and the estimated fresh oocyte diameter was then compared by using a paired t-test to determine if there was a significant difference be- tween preserved and unpreserved oocyte diameters. Age was determined by using the left sagittal otolith, which was embedded in epoxy resin. A 0.5-mm transverse section encompassing the otolith core was cut with an Isomet low-speed saw with diamond wafering blades. The thin section of the otolith was embedded in epoxy and ob- served with a dissection microscope at the magnification appropriate for the otolith's size. Age was recorded as the number of rings (annular bands) present. The otoliths were initially aged by one reader. A second reader then evalu- ated a subsample of specimens from 1998 and 2000 and all the otoliths from 1999. Ages were validated by the percent- age of agreement between the two age determinations, an analysis of variance (ANOVA) between the two groups of ages, and a paired Mest comparing the means and vari- ances of the two groups (Campana et al., 1995). Results Fecund female striped mullet (again, defined as those females with ovaries containing oocytes >4G0 ^m in the tertiary stage of vitellogenesis) were collected from late October through February; most of the specimens were caught in November and December for all three years. Size-frequency distributions did not vary over the three years of the study (Fig. 3). Fewer fish (n=37) were taken in 1999 versus 1998 (n=50) and 2000 (/i=42). The most abun- dant size class for each year was that from 401 to 450 mm. In the overall size-frequency distributions, fecund females made up greater than 44% of those fish larger than 400 mm in 1998 and comprised all of the specimens over 500 mm. In 1999 the fecund females made up 12.5"% offish in the 401-450 mm size range, 39% of the fish in the 451-500 mm size range, and 100% of the fish in the size classes over 500 mm. In 2000, fecund females made up 17.7% of the 401^50 mm size class, 35.7% in the 451-500 mm size range and, like 1998 and 1999, all of the specimens over 500 mm. The majority of females used in our fecundity study were 3 or 4 years old (Fig. 4), accounting for 80.0% of the specimens in 1998 and 73.3% in 1999. However, the age distribution in 2000 showed that the frequency of 2-, 3-, and 4-year-olds was the same and that these three ages made up 82.0% of the fecund fish sampled that year. Three-year- old fish made up the largest single group in 1998 and 1999. The age determined from the otoliths was validated as part McDonough et aL: Fecundity and spawning season of Mugil cephalus 827 25 - ■■1 1998 1 1 1999 ^H 2000 20 - 15 - ! 10 - i 5 - n r 1 0 - 1 1 1 1 1 1 .n. 1 — g — , 1 2 5 6 Age (yr) 10 Figure 4 Age-frequency distribution of fecund striped mullet in South Carolina estuaries from 1998 to 2000. n (number offish in sample) = 50 fish for 1998, n = 37 for 1999, and /; = 42 for 2000. of another study where size and age structure of striped mullet in South Carohna was examined (Wenner and Mc- Donough, unpubl. data^). A comparison of multiple read- ings of the same group of otoliths assessed aging precision. One year (1999) was chosen at random and all of the oto- liths («=1234) from that year were aged by a second reader. The ages of the two independent determinations were then compared by using a one-way ANOVA and a ^test. The variance statistic was 2.78 for the original ages and 2.81 for the second age reads, which were not significantly dif- ferent (P=0.001) and both had almost identical normalized residuals. Overall, there was an 83.4% agreement on ages between readers. The results from the ANOVA (F= 1555.0, df=10, P=0.000) and the t-tesi «=2.898, df=1233, signifi- cance |2-tailed] =0.004) both confirmed that there was no significant difference between the separate age determina- tions. Therefore, the age recorded by the first reader for all specimens was used in the analysis. The length-weight relationship for fecund female striped mullet was compared by using a linear regression of (natu- ral) log-transformed body weight against total length to see if there were any differences between years. The regression coefficients from each year were compared by Wenner, C. A., and C. J. McDonough. 2001. Cooperative research on the biology and assessment of nearshore and estua- rine fishes along the southeast coast of the U.S. Part IV: Striped mullet, Mugil cephalus. Final rep, Grant no. NA77FF0550, 82 p. Marine Resources Research Institute, South Carolina Dept. of Natural Resources, P.O. Box 12559 Charieston, S.C. 29422-2559. using a test of significance between more than two slopes (Zar, 1984). The weight measurement used was total body weight minus gonad weight (TW-OW=BW) because ovary weight had a considerable influence on total body weight in the fecund specimens (GSI values of 7.7 to 27.7). There was no significant difference in the total length to Ln body weight regressions between different years (F=9.22, P=0.001, df=129). Because there was little difference in the regressions between years and in order to increase sample size, data from all three years were combined to obtain the overall total-length to Ln-body-weight relationship of Ln BW = -11.1 + 2.92 (TL) (Fig. 5). In contrast, the length and body-weight-at-age relationships were highly variable; a wide range of sizes occurred in the 2-, 3-, and 4-year age classes (Fig. 6). The high degree of variability was also ex- acerbated by the smaller number offish age 5 or older. The gonadosomatic index (GSI) for fecund mullet ranged from 7.7 to 27.5 in 1998, 9.3 to 27.7 in 1999, and 9.5 to 26.6 in 2000. In contrast, the GSI for mature females that were not undergoing any reproductive development ranged from almost zero to 4 for all three years of the study. The rela- tionship of GSI to size (TL or BW) was not very strong in any year. However, GSI was positively correlated (P=0.01) with oocyte diameter and negatively correlated with oo- cyte density (Table 2) because of the inverse relationship of oocyte density and oocyte diameter. The correlation coef- ficient for GSI and age were very close to zero and slightly negative (Table 2). Mean GSI by month for males and females (Fig. 7) in- creased from October through April, peaking in November- 828 Fishery Bulletin 101(4) December. The duration of the reproductive season, as evi- denced by advanced reproductive condition determined by the GSI, was also confirmed from histological assessments of maturity stages for all gonads collected during this time period (not just those used for the fecundity study) which indicated that reproductively developing males were present 5 7 0 - 65 60 5.5 - 5.0 Ln BW=-]] A +29 r'=949. F= 9690.4 /I = 1 29 250 300 350 400 450 500 550 600 Total length (mm) Figure 5 Regression analysis of log-transformed (Ln) body weight on total length for fecund striped mullet in South Carolina estuaries from 1998 to 2000. n ( number of fish in sample) = 129. August through February, whereas reproductively develop- ing females were present August through April (Fig. 8). There was no significant difference in oocyte density among subsamples taken from different areas of the ovary lobe. This result was obtained by using an ANOVA of oocyte densities between the four divided areas of the ovary lobes where subsamples were taken (F=0.421, df=3). This analysis allowed us to accept the assumption that oocytes were equally distributed throughout the ovary lobes, which provided validation for the random sampling of oocytes from different areas of the lobe in order to determine individual fecundity. The regression of individual fecundity with total length (TL) was not a linear relationship, whereas the regression of fecundity on body weight (BW) was linear. Therefore, the comparisons of individual fecundity to total length (TL) and body weight (BW) were made by using both the raw data and the data with natural log transformations. The range of specimen total lengths was 291 to 600 mm in 1998, 332 to 588 mm in 1999, and 325 to 592 mm in 2000 and for body weight 242 to 2149 g in 1998, 335 to 2008 g in 1999, and 284 to 2144 g in 2000. Mean fecundity, compared between years with a two sample Mest, was significantly different between 1999 and 2000 {t=0.019, df=78, P=0.985) but was not significantly different between 1998 and 1999 (<=0.974, df=86, P=0.336) 650 Table 2 Pearson correlation coefficients, with significance values, for the morphological Carolina estuaries from 1998 to 2000. n (number offish in sample) = 129. TL diameter, ODN = oocyte density, GSI = gonadosomatic index, FEC = fecundity. variables and fecundity of striped mullet in South = total length, BW = body weight, ODM = oocyte Age TL BW ODM ODN GSI FEC Age Pearson correlation significance (2-tailed) 1.000 TL Pearson correlation significance (2-tailed) 0.117 0.186 1.000 BW Pearson correlation significance (2-tailed) 0.164 0.062 0.951** 0.000 1.000 ODM Pearson correlation significance (2-tailed) 0.004 0.964 0.101 0.254 0.029 0.741 1.000 ODN Pearson correlation significance (2-tailed) 0.575 -0.050 0.927 ^0.008 0.282 0.095 0.000 -0.628** 1.000 GSI Pearson correlation significance (2-tailed) -0.029 0.746 0.128 0.147 -0.006 0.947 0.543** 0.000 -0.645** 0.000 1.000 FEC Pearson correlation significance (2-tailed) 0.113 0.200 0.892** 0.000 0.888** 0.000 0.059 0.,'504 0.142 0.105 0.260** 1.000 0.003 " Correlation is significant at the 0.01 leve 1 (2-tailed). McDonough et al : Fecundity and spawning season of Mugil cephalus 829 and between 1998 and 2000 «=1.368, df=92, P=0.179). However, given that the mean fecun- dity was 1.18 miUion oocytes in 1998, 1.16 mil- hon oocytes in 1999, and 1.09 milhon oocytes in 2000, the difference in mean fecundity between 1999 and 2000 was probably not biologically significant. It was determined that data could be pooled across years for several reasons. The coefficients of determination for each year in- dicated that there was a similarly strong rela- tionship of fecundity to total length and body weight in all three years and the coefficients of variation for each year (0.408 for 1998, 0.594 for 1999, and 0.457 for 2000 at P=0.001) were not significantly different. By pooling the data from all three years we were able to determine two models of potential fecundity based on total length (TL) and body weight (BW) (Fig. 9): Ln Fecundity = -6.86 + 3.42(Ln Total Length) [r2=0.803,F=527.2, df=129] Ln Fecundity = 6.95 + 1.05(Ln Body Weight) [r2=0.804, F=530.6, df=129]. The r^ values for untransformed data were very close to the values obtained with trans- formed data (r2=0.795, i^=502.9 for fecundity on total length [TL] and r2=0.787, F=479.4 for fecundity on body weight [BW]). The high r^ values, as well as the high correlation coeffi- cients between fecundity and total length and body weight (Table 2) indicated that potential fecundity was size dependent. Unlike fecundity, oocyte density did not change with size (TL or BW) in 1998 and 1999 and increased with size in 2000. The increase in density in 2000 was due to a group of fish captured in freshwater in October having relatively low GSIs and high densities of oocytes that also happened to be some of the largest fish captured that year. Oocyte density was negatively correlated with GSI (Table 2), and thus indicated that increasing GSI resulted in lower oocyte densities. There was not a high degree of variability in oocyte di- ameter over the entire size range for the three years of the study. Oocyte diameter did appear to increase with age in 2000 and remained stable for 1998 and 1999. However, the increase in oocj^te diameter in 2000 was not statistically significant. In a comparison of mean oocyte diameter in each size class (total length) by month of capture, the data for all three years were pooled in order to obtain adequate repre- sentation in each month. Oocyte size ranged from 463 to 682 /jm and the mean size was 596 ^m. The largest mean oocyte diameters were found in specimens captured in Jan- uary and February. Specimens were captured during the months of November and December for all size classes and there was an increase in oocyte diameter with each pro- o 1998 700 A 1999 O 2000 o A A o 600 - k o 500 - O o i i o e O o 400 ' i o 2 # o o A 300 - o 700 A 3000 1500 - 600 o o o o B 5 6 Age (yr) Figure 6 Relationship of total length (A) and body weight (B) to age for fecund striped mullet in South Carolina estuaries from 1998 to 2000. n (number offish in sample) = 129. gressive month. In particular, females in the 400-500 mm size range (which represented the largest number of specimens) were examined and there was a progression of increasing oocyte diameter with month of capture through the reproductive season. The increase in oocyte size, as the reproductive season progressed, was more apparent when mean oocyte diam- eter by month for each year separately was examined. Specimens were collected from October through February in 1998, November through January in 1999, and October through December in 2000. Equal effort was made during all of these months of each year to capture specimens, but they were not always available for capture. There was an increase in mean oocyte size per month as the spawning season progressed in all three years (Fig. 10). Even though the largest oocyte size measured was 682 /jm, this measure- ment was that of a preserved oocyte. If we factor in a mean shrinkage of 4%, maximum oocyte size becomes 709 ^m. 830 Fishery Bulletin 101(4) The paired i-test showed no significant difference between the preserved oocytes and the predicted size of fresh oo- cytes (t=-26.2, df=128, P=0.000). Discussion Fecundity in striped mullet from South Carolina correlated highly with length and weight, but not with age. Oocyte counts ranged from 4.47 x IQS to 2.52 x 10'' in 1998 for fish ranging in total length from 331 mm to 600 mm, 2.13 x 10^ to 3.89 X 10® in 1999 for fish ranging in total length from 332 mm to 588 mm, and 3.89 x 10^ to 3.01 x 10® in 2000 for fish ranging in total length from 325 mm to 592 mm. These fecundity levels correspond with general fecundity levels (2.0 x 10'' to 14.0 X 10®) found in striped mullet in northeast Florida (Greeley et al., 1987), the Gulf Coast (Render et al., 1995) as well as studies in Europe and Asia (review by Alvarez-Lajonchere, 1982). One marked difference in fecun- dity between the present study and some in the literature was the difference in oocyte density. Render et al. (1995) found densities ranging from 798 to 2616 oocj^es/g ovary weight, whereas densities in the present study ranged from 1710 to 14,817 oocytes/g ovary weight. However, although fecundity increased with both total length and body weight in 1998 and 1999, densities did not. The lower oocyte densi- ties in the larger fish were most likely indicative of larger oocytes. This feature is common in both synchronous and asynchronous spawning fishes (Greeley et al., 1987; Render et al., 1995; Fox and Crivelli, 1998; DeMartini and Lau, 1999). Because total length and body weight were more highly correlated with increased fecundity, the larger speci- mens would have made a greater individual reproductive o O Males Females M.^Y JUN JUL AUG SEP OCT NOV Month DEC JAN FEB MAR APR Figure 7 Mean gonadosomatic index value by month for male and female striped mullet from South Carolina estuaries from 1998 to 2000. n (number of fish in sample) = 455. contribution during any given spawning season (Korhola et al., 1996; Kaunda-Arara and Ntiba, 1997; DeMartini and Lau, 1999). Making estimates of potential fecundity from age alone was difficult because of the variability in the size-at-age relationship; however, fecundity estimates from total length or body weight appeared more reliable and reflected values closer to the fecundity levels observed in the present study. Previous studies have reported that female mullet be- come reproductively mature at three years of age (Thom- son, 1951; 1963; Stenger, 1959; Chubb et al., 1981; Render et al., 1995). Greely et al. (1987) suggested that fecundity specimens collected in northeastern Florida were as young as two years at maturity. But, Greely et al. (1987) did not age the striped mullet used in their study and inferred age from a size-at-age growth schedule (Thomson, 1966). In the present study, two-year-olds made up only a small per- centage of the fecund fish in 1998 and 1999. However, the results were different for the 2000 specimens in that there was an almost equal distribution in the age frequency of two-, three-, and four-year- olds. However, three- and four- year-olds made up the greatest percentage of females with advanced ovaries. Maturing fish were those undergoing ac- tive vitellogenic development and were generally captured in and around inlets or estuaries. Our study would suggest that female striped mullet reach 50% maturity at age 2 and 100% maturity at age 3. Three- and four-year-olds made up the majority of reproductively advanced fish in all years, whereas less abundant older fish made less of a contribu- tion towards total reproductive effort. There are several possible explanations accounting for the wide age distribution in maturity stages; the most likely is that fecundity is size related despite the highly variable growth rates and the widely ranging size at age in adult striped mullet. Size at maturity has been found to range widely from 230 mm standard length (Thomson, 1963; Greeley et al., 1987; Tamaru et al, 1994) up to 410 mm standard length (Thomson, 1963; 1966; Chubb et al., 1981) for two- and three-year-old fish. The lower end of this size range agrees quite readily with the lower size range (291 mm TL=239 mm SL) found in our study. In a con- current study of maturity schedules related to size and age in South Carolina striped mullet (McDonough, unpubl. data), male striped mul- let were found to mature at two years of age and as small as 250 mm total length ( 190 mm standard length). Other species of mullet have been shown to mature over a wide range of siz- es. The Pacific mullet (Mugil so-iuy) becomes mature upon reaching approximately 430 mm total length (Okumus and Bascinar, 1997). Monthly GSI levels clearly showed that the time period of reproductive activity is from October through April. Female striped mullet in all reproductive developmental stages were observed during the course of our sampling, with the exception of stage-3 (hydrated oocytes) females and females with McDonough et al.: Fecundity and spawning season o\ Mugil cephalus 831 recently spawned ovaries (characterized by the presence of postovulatory foHicles). Atretic ovaries were observed from December through May. There were no postovu- latory follicles observed, indicating that any atretic ovaries were not from recently spawned fish. The fish with atretic ovaries were characteristically emaciated for their size (TL and BW) and were most common from January through March. The presence of females with atretic ovaries starting in December is strong evidence that spawning occurred in November, if not earlier, and females with atretic ovaries caught as late as May demonstrate that spawning may still occur as late as April. Additional evidence for the October through April spawning period has also been shown in backcalculated birth dates for juvenile striped mullet by daily growth increments (McDonough and Wenner, 2003). This evidence supports the concept of offshore spawning in striped mullet and a yet undetermined time period required for moving from the estuaries to the spawning areas and for returning again to the estuaries. Other authors have come to the same conclusion from similar evidence in estuaries throughout the southeast (Jacot, 1920; Broadhead, 1956; Anderson, 1958; Stenger, 1959; Shireman, 1975; Dindo and MacGregor, 1981; Greeley et al., 1987; Render et al., 1995; Hettler et al., 1997). All of the fecundity specimens were caught from Oc- tober through February when the mean monthly GSI was highest. Pien and Liao (1975) found that mullet oocytes reached a hydrated size of 900 to 1000 ^m. The size of oocytes used for fecundity counts in the present study ranged from 463 to 682 ^m. The maximum size of oocytes in the tertiary stage of vitellogenesis from our study was 600 ^m or greater. This result agrees with those of previous studies where the maximum size of oocytes prior to either hydration or atresia (if spawning did not occur) ranged from 600 to 700 ^m (Shehadeh et al., 1973b; Kuo et al., 1974). There was no evidence of prespawning atresia in any of the specimens used for fecundity estimates. The appropriateness of using a GSI alone to determine the level of reproductive development has been questioned, particularly for serial or asynchronous spawning fishes (De Vlaming et al., 1980; Hunter and Macewicz, 1985). Striped mullet can have a wide range of GSI values that range from practically zero to over thirty (Render et al., 1995). The GSI range for females in our study ranged from almost zero to 27.7. Because of the high variability in GSI with size, it does not appear appropriate to use GSI alone in order to assess reproductive development in striped mullet. When used in conjunction with histological analysis and mean oocyte diameter of tertiary-stage oocytes, GSI does provide excellent supporting evidence of reproductive schedules and spawning season duration. GSI is probably more ap- propriately used for isochronal spawning fishes than for serial spawning fishes because of the uniform development of oocytes in the former. However, it is still difficult to meet all the basic assumptions of the GSI index as given by De Vlaming et al. ( 1980) because of the high variability of GSI with size. Another technique that has been used in aquacul- ture situations to assess maturity and sex involves the use g. 0 o 6 c^ >^ g 200 0) 3 CT Developing males n = 340 n JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC B Developing females n = 277 I n JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Figure 8 Frequency distribution of reproductively developing male (A) and female (B) striped mullet by month in South Carolina estuaries from 1998 to 2000. of a cannula to remove oocytes from the ovaries of live fish which are then evaluated (Shehadeh, et al, 1973a; Kuo et al., 1974). This technique, although useful for determining sex and the extent or stage of reproductive development, would be inappropriate for estimating potential fecundity. Historically, reproductively developing mullet have been found in southeast United States waters from November through December (Jacot, 1920; Anderson, 1958; Stenger, 1959). During our study, reproductively developing mullet were caught in the Charleston Harbor Estuary from Oc- tober through February. Gonad development of these fish was discernible through gross morphological observation, and histological sections showed that vitellogenesis was well underway. Other studies have shown that previtel- logenic oocytes were usually less than 160 jum and that the onset of vitellogenesis began when the oocytes reached a size of 180 pm (Dindo and MacGregor, 1981; Greeley et al., 1987; Tamaru et al., 1994; Render et al., 1995). Devel- oping individuals caught during our study that were not used for fecundity counts had less-developed ovaries, GSI values less than 7, and mean oocyte diameters less than 832 Fishery Bulletin 101(4) 16 15 14 13 12 - ^ 11 1 5 O 5 16 15 14 Prediclion Inlerval (2 SD) Confidence Intenal (95%) Ln Fecundity = -6.86 + 3.42 (Ln TL) r^ = 0.W3, F = 527.2. df= 129 50 5.75 6.00 Ln Total length 6-25 6.50 13 12 Prediction Inlerval (2 SD) Confidence Interval (95%) ^ -' . ^ - ^ ,<>^-^^* - ^ • . » y^ ^^^' - • Ln Fecurulitx = 6.95 + 1.05 (Ln BW) B 1 1 r = 0.804. F = 530.6. rf/-= 129 1 1 ! 1 5.0 5.5 6.0 6.5 Ln Body weight 70 7.5 80 Figure 9 Regression of the natural log (Ln) transformation of individual fecundity on total length (A) and body weight (B) for striped mullet combined data, 1998 to 2000, from South Carolina estuaries, n (number offish in sample) = 129. 350 jjm. The vitellogenic activity in these ovaries was still in the primary and secondary stages. These specimens could not be used for fecundity counts because not all of the oocytes destined for that year's spawning batch had developed enough to be separable from the smaller oocytes that would not develop. Once the developing oocytes reached a size of 400 ^m or larger, they became more uniform in size and in appearance and it was obvi- ous which oocytes would constitute that year's spawn. At that point, fecundity could be determined more accurately because all the oocytes to be counted were significantly and equally larger. This point in particular is important in that it makes fecundity estimates from nonhydrated oo- cytes more accurate for isochronal spawning fishes, such as striped mullet. Fecundity estimates made in fishes that are batch-spawners should only be made from hydrated oocytes because of the presence of multiple developmental stages (Hunter and Macewicz, 1985). The presence of dif- ferent vitellogenic stages in the ovary of a repeat-spawn- ing fish makes it necessary to determine individual batch fecundity and spawning frequency before any estimate of annual fecundity can be made. In isochronal spawn- ing fishes, such as striped mullet, this process is made simpler by the fact that oocytes mature at a similar rate (Greeley et al., 1987). During early vitellogenesis (180 ^m to 350 fjm), there is a higher degree of variability in the rate of development and a range of developmental stages would be present from the presence of precortical alveoli through secondary and tertiary vitellogenesis (Render et al., 1995). Estimating numbers of oocytes (uniform in an individual but varying in size because of season) during this stage would naturally make oocyte density and oocyte- size relationships inconsistent. This could possibly be cor- rected by using some timing factor such as month during the spawning season. The present study did demonstrate an inverse relationship between oocyte density and oocyte diameter When month of capture was taken into consider- ation, oocyte density decreased with increasing oocyte size as the spawning season progressed. In conclusion there were several biological aspects of striped mullet reproduction demonstrated in this study. Fecundity levels in striped mullet increased with total McDonough et al,; Fecundity and spawning season oi Mugil cephalus 833 0 1998 -^ 1999 ___ 700 - O 2000 ^ 0) 1 650 - o E. 11 I <> ° S 600 - <> () o <> ■o B "" " >. o o O f ! L 500 - OCT NOV DEC JAN FEB Month Figure 10 Mean oocyte diameter per month during the spawning season for striped mullet in South Carolina estuaries 1998 to 2000. Error bars represent standard deviation. length (TL) and body weight (BW=TW-OW). Oocyte den- sity remained relatively stable with size in the fecund fish and this allowed reasonable estimates of potential fecun- dity based on total length and body weight. Age-specific fecundity was highly variable and there appeared to be no consistent relationship. The reproductive season for striped mullet in South Carolina extends from October through April as determined by mean monthly GSI levels and histological confirmation of reproductive state. Potential spawning periods would also occur within this period as evidenced by the elevated GSI levels and the presence of atretic (or possible postspawning) ovaries from December through May. The gonadosomatic index itself is more useful to evaluate reproductive potential when used in conjunc- tion with other techniques, such as histological analysis and oocyte diameter. The models of potential fecundity as they relate to size (total length and body weight) could be useful when applied to catch statistics of length and weight in populations with known size- and age- frequency distri- butions. This application would allow reasonable estimates of potential fecundity for these populations. Acknowledgments This study would not have been possible without the assis- tance of everyone in the Inshore Fisheries group at the Marine Resources Research Institute of the South Carolina Department of Natural Resources, which includes Myra Brouwer, John Archambault, Hayne Von Kolnitz, Will Hegler, Erin Levesque, Alice Palmer, Chad Johnson, Richie Evitt, Larry Goss, and Travis Waits. We especially thank Chad Altman of the South Carolina Department of Health and Environmental Control for the freshwater specimens. We also thank Myra Brouwer for assistance with map fig- ures and the anonymous referees for careful review and suggestions for this manuscript. This research was made possible by National Marine Fisheries Service MARFIN, grant no. NA77FF0550. Literature citations Anderson, W. W. 1958. Larval development, growth, and spawning of striped mullet (Mugil cephalus) along the south Atlantic coast of the United States. Fish. Bull. 58:501-519. Alvarez-LaJonchere, L. 1982. The fecundity of mullet (Pisces, Mugilidae) from Cuban waters, J. Fish Biol. 21:607-613, Arnold, E. L,, and J. R. Thompson, 1958. Offshore spawning of the striped mullet, Mugil cepha- lus, in the Gulf of Mexico. Copeia 1958: 130-132. Broadhead, G. C. 1956. Growth of the black mullet, Mugil cephalus, in west and northwest Florida. Mar. Lab. Tech, Ser 25:1-29. Campana, S. E., M. C. Annand, and J. L McMillan, 1995, Graphical and statistical methods for determining the consistency of age determinations. Trans. Am. Fish. Soc. 124:131-138. Chubb, C. F., I. C. Potter, C. J. Grant, R. C. J. Lenanton, and J. Wallace. 1981. Age, structure, growth rates, and movements of sea mul- let, Mugil cephalus L., and yellow eye mullet, Aldrichetta forsteri (Valenciennes), in the Swan-Avon river system. West- ern Austrailia. Aust. J. Mar. Freshw. Res. 32:605-628. Collins, M. R., and B. W. Stender. 1989. Larval striped mullet (Mugil cephalus) and white mullet (Mugil curema) off the southeastern United States. Bull. Mar. Sci. 45(3):580-589. 834 Fishery Bulletin 101(4) DeMartini, E. E., and B. B. Lau. 1999. Morphometric criteria for estimating sexual maturity in two snappers, Etelis carbunculus and Pristipomoides seiboldii. Fish. Bull. 97:449-458. De Vlaming, V. L., H. S. Wiley, G. Delahunty, and R. Wallace. 1980. Goldfish (Carassius auratus) vitellogenin: induction, isolation, properties and relationship to yolk proteins. Comp. Biochem. Physiol. B 67B:613-623. Dindo, J. J., and R. MacGregor 1981. Annual cycle of serum gonadal steroids and serum lipids in striped mullet. Trans. Am. Fish. Soc. 110:403- 409. Finucane, J. H., L. A. ColHns, and L. E. Barger 1978. Spawning of the striped mullet, Mugil cephalus, in the northwestern Gulf of Mexico. Northeast Gulf Sci. 2: 148-150. Fox, M. G., and A. J. Crivelli. 1998. Body size and reproductive allocation in a multiple spawning centrarchid. Can. J. Fish. Aquat. Sci. 55(3): 737-748. Greeley, M. S., D. R. Calder, and R. A. Wallace. 1987. Oocyte growth and development in the striped mullet, Mugil cephalus, during sesonal ovarian recrudescence: rela- tionship to fecundity and size at maturity. Fish. Bull. 85: 187-200. Hettler, W. F., D. S. Peters, D. R. Colby, and E. H. Laban. 1997. Daily variability in abundance of larval fishes inside Beaufort Inlet. Fish. Bull. 95:477-493. Humason, G. L. 1967. Animal tissue techniques, 426 p. W.H. Freeman and Co., San Francisco, CA. Hunter, J. R., and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax. NOAA Tech. Rep. NMFS 36:79-94. Jacot, A. P. 1920. Age, growth, and scale characters of the mullets, Mugil cephalus and Mugil curema. Trans. Am. Fish. Soc. 39(3): 199-229. Kaunda-Arara, B., and M. J. Ntiba. 1997. The reproductive biology of Lutjanus fulviflamma (Forsskal, 1775) (Pisces: Lutjanidae) in Kenyan inshore marine waters. Hydrobiologia 353(1/3):153-160. Kelly, C. D. 1990. Effects of photoperiod and temperature on ovarian maturation in the striped mullet, Mugil cephalus. Pac. Sci. 44(2):187-188 Korhola, A., P. Solemdal, P. Bratland, and M. Fonn. 1996. Variation in annual egg production in individual cap- tive Atlantic cod (Gadhus morhua). Can. J. Fish. Aquat. Sci. 53(3 ):610-620. Kuo, C. M. 1995. Manipulation of ovarian development and spawn- ing in grey mullet, Mugil cephalus L. Israel J. Aquacult. Bamidgeh 47(2):43-58. Kuo, C. M., C. E. Nash, and Z. H. Shehadeh. 1974. A procedural guide to induce spawning in grey mullet (Mugil cephalus L.). Aquaculture 3(1974): 1-14. McDonough, C. J., and C. A. Wenner 2003. Growth, recruitment, and abundance of juvenile Mugil cephalus in South Carolina estuaries. Fish. Bull. 101:343-357. Okumus, I., and N. Bascinar. 1997. Population structure, growth, and reproduction of introduced mullet, Mugil so-iuy, in the Black Sea. Fish. Res. 33:131-137. Pien, P. C, and J. C. Liao. 1975. Preliminary report of histological studies on the grey mullet gonad related to hormone treatment. Aquaculture 5:31-39. Render, J. H., B. A. Thompson, and R. L. Allen. 1995. Reproductive development of striped mullet in Loui- siana estuarine waters with notes on the applicability of reproductive assessment methods for isochronal species. Trans. Am. Fish. Soc. 124( 1 ):26-36. Rossi, A. R., M. Capula, D. Crosetti, D. E. Campton, and L. Sola. 1998. Genetic divergence and phylogenetic inferences in five species of Mugilidae (Pisces: Perciformes). Mar Biol. 131: 213-218. Shehadeh, Z. H., C. M. Kuo, and K. K. Milisen. 1973a. Validation of an in vivo method for monitoring ovar- ian development in the grey mullet (Mugil cephalus L.). J. Fish Biol. 1973(5):489-496. Shehadeh, Z. H., W. D. Madden, and T P Dohl. 1973b. The effect of exogenous hormone treatment on spermiation and vitellogenesis in the grey mullet, Mugil cephalus L. J. Fish Biol. 1973(5):479-487. Shireman, J. V. 1975. Gonadal development of striped mullet (Mugil cepha- lus) in freshwater Prog. Fish Cult. 37(4):205-208. Stenger. A. H. 1959. A study of the structure and development of certain reproductive tissues of Mugil cephalus Linnaeus. Zoolog- ica 44(2):53-70. Tamaru, C. S., C. S. Lee, C. D. Kelley, G. Miyamoto, and A. Moriwake. 1994. Oocyte gi'owth in the Striped mullet, Mugil cepha- lus L., at different salinities. J. World Aquacult. Soc. 25: 109-115. Thomson, J. M. 1951. Growth and habits of the sea mullet, Mugil dobula Gunther, in Western Australia. Aust. J. Mar Freshw. Res. 2:193-225. 1963. Mullet life history strategies. Aust. J. Sci. 25:414-416. The grey mullets. Oceanogr. Mar. Biol. Annu. Rev. 4: 301-335. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21:325-343. Wenner, C. A., B. A. Roumillat, and C. W. Waltz. 1986. Contributions to the life history of black seabass, Cen- tropristis striata , off the southeastern United States. Fish. Bull. 84:723-741. Zar, J. H. 1984. Biostatistical analysis, 2"'' ed., p. 292-305. Prentice Hall Inc., Englewood Cliffs, NJ. 835 Abstract— Fishes are widely known to aggregate around floating objects, including flotsam and fish aggregating devices (FADs). The numbers and diver- sity of juvenile fishes that associated with floating objects in the nearshore waters of the eastern tropical Pacific were recording by using FADs as an experimental tool. The effects of fish removal. FAD size, and the presence or absence of a fouling community at the FAD over a period of days, and the presence of prior recruits over a period of hours were evaluated by using a series of experiments. The removal of FAD-associated fish assemblages had a significant effect on the number of the dominant species (Abudefduf troschelii) in the following day's assemblage com- pared to FADs where the previous day's assemblage was undisturbed; there was no experimental effect on combined spe- cies totals. Fishes do, however, discrimi- nate among floating objects, forming larger, more species-rich assemblages around large FADs compared to small ones. Fishes also formed larger assemblages around FADs possessing a fouling biota versus FADs without a fouling biota, although this effect was also closely tied to temporal factors. FADs enriched with fish accumulated additional recruits more quickly than FADs that were not enriched with fish and therefore the presence of prior recruits had a strong, positive effect on subsequent recruitment. These results suggest that fish recruitment to float- ing objects is deliberate rather than haphazard or accidental and they sup- port the hypothesis that flotsam plays a role in the interrelationship between environment and some juvenile fishes. These results are relevant to the use of FADs for fisheries, but emphasize that further research is necessary for applied interests. Marine fish assemblages associated with fish aggregating devices (FADs): effects of fish removal, FAD size, fouling communities, and prior recruits Peter A. Nelson Department of Biological Sciences Northern Arizona University Flagstaff, Arizona 86011-5640 Present address: Center for Marine and Biodiversity & Conservation Scripps Institution of Oceanography University of California, San Diego La Jolla, California 92093-0202 E-mail address, pnelson@ucsd.edu Manuscript approved for publication 13 June 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:835-850 Fishes associate with floating objects in nearly all oceans of the world (Good- ing and Magnuson, 1967; Hunter and Mitchell, 1967; Klima and Wickham, 1971; Crawford and Jorgenson, 1993; Kingsford, 1993; Druce and Kingsford, 1995; Massuti et al., 1998; Hampton and Bailey, 1999; Parin and Fedoryako, 1999). Fishes also gather around fish aggregating devices (FADs), floating objects deployed to concentrate target species or bait fishes and improve the catch for artisanal, sport, or commercial fisheries. The physical attributes of a floating object, such as a FAD, may affect the ability of potential fish recruits to locate the floating object or may affect the adaptive advantages of associating with that object (or both) — a topic that has been addressed in numerous prior studies (e.g. Hunter and Mitchell, 1968; "Wickham et al., 1973; Wickham and Russell, 1974; Fedoryako, 1989; Roun- tree, 1989; Safran, 1990; Safran and Omori, 1990; Friedlander et al., 1994; Hall et al., 1999b). However, the present study is apparently the first to address empirically the effects of disturbance, fouling communities, and prior recruits by examining both the number and the diversity of fishes that aggregate around FADs. In addition, this study addresses the effect of FAD size, a factor well represented in prior studies but frequently confounded by temporal or design issues. FADs are widely used to enhance sport and commercial fisheries, but are expensive to build, deploy, and maintain; therefore better information about the effects of FAD size and foul- ing could aid design efforts. Given the bycatch associated with the FAD fishery for tuna in the eastern tropical Pacific (Hall et al., 2000), for example, we need a better understanding of how fishes use FADs in order to manage fisheries for FAD-associated species (Lennert- Cody and Hall, 2000). Finally, careful study of how differing characteristics of floating objects affect fish recruit- ment may provide important clues regarding the adaptive significance of fish associations with flotsam and drift algae — a phenomenon widely noted but poorly understood. Prior research has suggested that rates of immigration and fish removal from FADs similar to those seen in the present study were high from one day to the next (Nelson, 1999), and "Wickham and Russell (1974) reported that mid- water FADs, which were fished daily, produced a larger cumulative catch than mid-water FADs, which were un- disturbed during the same period and then fished once at the end of the con- clusion of the study. I tested the hypoth- esis that, over time, the size and diver- sity of FAD-associated fish assemblages are reduced by the repeated removal of these fishes compared with undisturbed assemblages. Effective management or use of FADs deployed for fisheries purposes and an understanding of the ecological relationship between flotsam and fishes associated with flotsam will depend in part on patterns of immigra- tion and loss to fish assemblages. 836 Fishery Bulletin 101(4) There have been numerous attempts to equate flotsam structure (size, complexity, orientation, etc.) with the number of associ- ated fishes (e.g. Hunter and Mitchell, 1968; Dooley, 1972; Wickham et al., 1973; Wickham and Russell, 1974; Rountree, 1989; Druce and Kingsford, 1995), but the results have been equivocal, except when the analysis was re- stricted to a single species (e.g. Histrio histrio, Dooley, 1972;Decapterus punctatus, Rountree, 1989). Huge aggregations have been associ- ated with very small objects — lATTC (Inter- American Tropical Tuna Commission ) records include a report of 55 metric tons of mostly yellowfin tuna {Thiinnus albacares) fished from beneath a 1-m length of floating poly- propylene rope (Hall et al., 1999b); therefore, despite the intuitive appeal, there is no clear reason to expect that size of FAD per se is an important factor in determining the size of associated assemblages. Thus, object size re- mains an unresolved problem in understand- ing flotsam-associated communities. If there are optimal FAD sizes, these may be species specific, and economical FAD design depends upon controlled experiments in the field. Fouling organisms (sessile invertebrates and algae that colonize flotsam) are believed to have a strong, positive effect on the sub- sequent recruitment and retention of fishes by commercial fishermen (Gaertmer and Medina-Gaertner, 1999; Hall et al,, 1999a; Hallier and Parajua, 1999; Suzuki, 1999). However, prior to the results presented here, there appear to have been no controlled tests of the hypothesis that the presence of fouling organisms enhances fish recruitment to a floating object. I also compared the numbers and diversity of fishes associated with FADs that are equipped with ar- tificial (lead weight) fish versus FADs without these artifi- cial fish. The latter experiment was intended to determine the importance of prior recruits to subsequent patterns of recruitment. To test a similar hypothesis over the short term (hours versus days) and using living fish instead of painted models, I also compared recruitment to FADs enriched with real fish (juvenile Abiidefduf troschelii) to unenriched FADs. I tested the hypothesis that each of these factors would affect the number of fishes associated with FADs (combined and individual species), as well as the species diversity of FAD-associated fish assemblages. Both the size of these FAD-associated fish assemblages and their species diver- sity provide insight on recruitment processes and the use of floating objects by fishes. Although the association of fishes with floating objects has been well documented, very little is known regarding the behavioral and ecological processes behind these assemblages. The results reported in the pres- ent study provide new information on the role of flotsam and FAD characteristics in determining the number and diversity of these assemblages, and some clues towards Figure 1 Location of the study site at Achotines, Panama, Central America. understanding why and how fishes aggregate beneath floating objects. Materials and methods Study site and FAD construction All research was conducted between July and October 1997 on the Pacific coast of Panama, Central America, from the Inter-American Tropical Tuna Commission laboratory at Achotines, near the tip of the Azuero Peninsula (Fig. 1). Experimental FADs were constructed of three tuna purse- seine buoys lashed together and anchored to the substrate with a 25-kg cast concrete block unless otherwise noted (Fig. 2). The length of the anchor lines allowed the FADs to rest at the surface at all tidal heights. Each buoy was roughly 25 cm in cross sectional diameter, and approxi- mately 35 cm in length. The FADs were detachable from their moorings by detaching a large (2 m diameter) loop on the anchor line that held a 2-kg line weight (Fig. 2). This design allowed me to change FADs for another treatment. The FAD arrays were deployed nearshore (within 1.5 km; Nelson: Fad characteristics and associated fish assemblages 837 Fig. 3) and in shallow water (14-25 m). The FAD treatments were not assigned randomly to FAD positions; instead, I assigned treatments uni- formly across the FAD array because the total number of FADs was relatively low (8-10) and, given the small number of experimental units, in order to reduce the possibility that the results be confounded by positional effects. The anchored FADs were spaced approximately 100 m apart. The maximum horizontal underwa- ter visibility measured was 27 m, and typically averaged much less. Assuming that vision was the principal means by which fishes located these objects, it is therefore highly unlikely that fishes treated the FAD array as a single "object" or moved from FAD to FAD within the array. I believe that it was unlikely that any fish trans- ferred from one FAD to another for the following reasons. 1) The horizontal underwater visibility was always much less than the distance between FADs. 2) Observations suggest that short-term, daytime fidelity was high; once a fish associated with a floating object, it was unlikely to leave that object during the day (Nelson, 1999). 3) Crossing an open stretch of water for another floating ob- ject (presumably within detection range) entails a potential risk for a fish. Moser et al. (1998) did note that the larger juveniles and adult fishes associated with floating Sargassum showed little apparent fidelity to this habitat and would move between their research boats, floating observation equipment, and the Sargassum habitat. However, the fishes that were observed to move between these floating objects were juvenile carangids be- tween 10 and 20 cm in length (Moser et al., 1998), whereas the fishes in the present study were gen- erally much smaller and presumably less vagile. Longshore currents ran roughly west to east through the experimental area, and rarely in the reverse (Fig. 3). I recorded an estimate of cur- rent direction using an underwater compass and the angle of the FAD anchor lines. This estimate represented the sum of the forces due to windage on the FAD buoys and currents. Censusing FADS FAD-associated fish assemblages were censused by direct visual observation by divers using mask and snorkel. Most other studies offish assemblages associated with floating material have employed nets or quantitative fishing methods for sampling purposes (e.g. Kojima, 1960; Dooley, 1972; ICings- ford, 1992, 1995), but Hunter and Mitchell (1968) compared data from net captures, automated photography and direct visual observations and found that the visual observations agreed well with the other methods, and provided behavioral information not available with the other methods. In my study, a FAD was approached by swimming Fish Aggregation Device (FAD) 2 kg line weight cast concrete with 6.25 mm polypropylene loop FAD line & anchor line 9 mm blue polypropylen to bottom & 25 kg concrete anchor Figure 2 Design of detachable fish aggregating devices (FADs) used in the present study. — '-, ^Laboratorio A^fiotines long-shore Currents North \to Los Frailles i9-12km rocky coast sand beach experimental FAD array (general location) t drifting FAD experinnents conducted in this area Figure 3 Coastline and general location of experiments and observations. The laboratory iLaboratorio Achotines) is located at approximately T'S'N lat, SOnCW long. 838 Fishery Bulletin 101(4) slowly and quietly at the surface from a distance of at least 12 m. All fishes associated (defined below) with the floating object were counted and identified; therefore the statistical unit in all of the experiments described below was a single FAD-associated assemblage of fishes at a given date and time. Horizontal underwater visibility, measured with a Secchi disk, was always sufficient to allow the identification of species and to count individual fishes from a minimum distance of 2 meters. Any fish observed within 2 m of a FAD was considered to be "FAD-associated." Very few fishes were observed outside of this range, and with rare exceptions, fishes responded to the approach of an observer first by swimming towards the observer and then by moving closer to the FAD, rather than away from it. Different species of fishes used the space around and below the FAD differently, as both Good- ing and Magnuson (1967) and Hunter and Mitchell (1967) described, but the juvenile fishes that predominated in the present study were unambiguous regarding their relation- ship to the FADs. After fishes resumed their prior positions in relation to a FAD, continued observations of these fishes revealed that there was no inclination to abandon that FAD. The appearance of potential predators invariably resulted in a tightening of the spatial distribution around the FAD. When the experiment required the capture of FAD-as- sociated fishes, I used a smaller (1.1x1.3 m) version of the diver-operated liftnet described by McCleneghan and Houk (1978). Captured fishes were preserved for further studies, held in grow-out facilities in the laboratory to verify spe- cies identification, or released 1.5 km down-current over rocky reef habitat to ensure that they had effectively been removed from the FAD array. Diversity calculations Measurements of species diversity provided a means of monitoring treatment effects on the composition of FAD- associated assemblages. I measured species diversity using species richness (S, the raw number of species observed), and the Brillouin index (HB). S is simple and widely used, but increases with sample size, and, where sample sizes are unequal, HB provides a less biased measure of diversity (Magurran, 1988). In addition, HB was chosen over one of the more commonly used information theory indices (e.g. the Shannon-Wiener index) because 1) FAD- associated assemblages are not a random sample of poten- tial recruits (different species vary in their attraction to floating objects) and 2) each of these assemblages was completely censused — not sampled (Magurran, 1988, and references therein). HB is calculated as ///? = In/V! -Z'"".^ where N = the total number of fishes of all species observed; and ;i, = the number of individuals within the ith spe- cies (Magurran, 1988). Statistical analyses 1 used a two-way repeated measures ANOVA (a=0.05) to test for treatment differences, differences among observa- tion dates and evidence of treatment-by-sample interaction for assemblage sizes (no. of fish(es)), species richness (S) and species diversity (HB). Because individual species dif- fered in their relative abundance and had different ecologi- cal requirements, there was the potential for the dominant species to bias comparisons of experimental treatments where assemblage size (a combinination of all species) was used. For all experiments except for the recruit-enriched experiment, I repeated the statistical analyses twice: once using the number of sergeant major damselfish (Abudefduf troschelii) only and again using all fishes combined but with A. troschelii removed. For the artificial fish experi- ment where rainbow runner (Elagatis bipinnulata) were particularly abundant, I ran separate analyses for A. tros- chelii alone, £. bipinnulata alone, and for all species minus the numbers of A. troschelii. When no fishes were present at a FAD, HB was undefined; the "missing" data were re- placed according to the procedures of Zar ( 1996) and the degrees of freedom were reduced accordingly. Fish-removal experiments Observations on similar FADs during a previous field season at the same location suggested that the turn- over rate of fish associated with anchored FADs is high, especially when the initial assemblage is large, but that some recognizable individuals did persist from day-to-day (Nelson, 1999). If immigration and emigration rates were as high as suspected, FADs cleared offish on a daily basis should not differ significantly from undisturbed FADs in their mean assemblage size or in the average number of species associated with these FADs. Wickham and Russell (1974) compared the catches of bait fishes associated with FADs subject to daily purse-seine sets versus those allowed to "soak" undisturbed for three days prior to a single purse-seine set and concluded that sufficient emigration and immigration occurred on a daily basis to remove any appreciable effect of daily removals. I sought to address similar questions, but by using a different system (juvenile reef fishes versus bait fishes). To test these hypotheses, 1 deployed eight identical FADs on 30 June 1997 in two lines of four FADs each, oriented roughly parallel to shore (Figs. 2 and 3). Fishes associ- ated with all eight FADs were counted and identified on a daily basis, beginning 3 July 1997. Alternate FADs were cleared of all fish, following the daily counts; the remaining FADs were left undisturbed in such a way as to distribute the treatments evenly among the FAD array. After three consecutive days of these observations (series 1), the treat- ments were reversed, and previously undisturbed FADs were cleared, and those that had been cleared regularly were left undisturbed (series 2). The treatments were re- versed in an attempt to control for possible positional effects of the FADs. However, the two series were necessarily run consecutively, not concurrently; therefore treatment posi- tion was confounded by sample date. I used a 2 (cleared vs. Nelson: Fad characteristics and associated fish assemblages 839 undisturbed) by 2 (first series vs. second series) by 6 (sample date) model and I used a repeated measures ANOVA (repeated on sample date) on the follow- ing dependent variables: assemblage size (total no. of fishes), species richness (S), and HB. I repeated analyses of assemblage-size effects looking at the number of A. troschelii only, and the total number of fishes minus the number of A. troschelii. FAD size To determine the effect of FAD size on the associ- ated assemblage size and diversity, I compared FAD-associated fish assemblages between triple- size FADs and single FADs. An existing anchored array of eight FADs (two lines parallel to the coast of four FADs each. Fig. 3) was cleared of fishes on 24 July 1997. As the fish were removed from the FADs, each FAD was replaced with a fresh (i.e. clean and unfouled) single or triple-size FAD, placed at alter- nating positions. The single FADs were constructed as described above and in Figure 2; the triple-size FADs were identical to the single FADs, except that they consisted of nine, rather than three, purse-seine buoys lashed together and had the effect of nearly tripling the wetted surface area (although inner buoys are less exposed than outer ones) and the volume of the FAD, and of increasing the maximum linear dimension of the FAD by a factor of two. Treatments were not reversed for this or subsequent experiments because sample date appeared to be the major factor determining assemblage size for any species, based on the previous experiment. Note that in each of these experiments, except for the recruit-enrichment experi- ment that used drifting FADs, treatments were assigned uniformly throughout the FAD arrays so that onshore, offshore, or longshore biases in recruitment due to oceano- graphic processes would not confound the results. Fishes at all FADs were counted and identified on three dates (26, 28, and 30 July 1997), each observation separated from the next by 48 hours. No fish were collected, with the exception of one balistid, taken from the array on 26 July because it was the first of that species to be observed associated with a FAD. Data were analyzed for experimental effects on total assemblage size, species diversity (S and HB), the number of A. troschelii, and total number of fishes minus the number of A. troschelii. Presence of absence of a fouling community To determine whether the presence of a fouling community on a floating object affected the associated fish assemblage, I compared FAD-associated assemblage sizes and species richness between fouled and unfouled (control) FADs. Con- trol FADs were scrubbed of all fouling organisms. Fouled FADs had been deployed for a minimum of 14 days (range: 14-22 days) in the study area, and had accumulated fouling that completely covered the wetted surface of the FAD with gooseneck barnacles (Lepas sp.), hydroids, and bryozoans. Grapsid crabs and polychaete worms (Amphimone vagans) Fish Aggregation Device (FAD) steel hoop to distribute model \ aggregation 1 oz lead 1 oz lead weight, weight, vertical model fish orientation (painted) to line weight and anchor Figure 4 Treatments for the artificial fish experiment involved suspending artificial fish, lead weights of an equivalent mass and volume, or nothing (control) from a steel hoop lashed beneath the FAD. were also intermittent associates of fouled FADs. Control and fouled FADs were deployed on 8 September 1997 in an alternating array of eight buoys, with four FADs per treat- ment (layout and spatial distribution of treatments follow that of the FAD size experiment). All fishes were cleared from FAD positions prior to deploying the FADs, and data collection commenced 24 hours later Data were collected on four consecutive days (9, 10, 11, and 12 September 1997) and analyzed for experimental effects on total assemblage size, species diversity (S and HB), the number of A. tros- chelii, the number of E. bipinnulata, and total number of fishes minus the number of A. troschelii. Artificial fish experiment I tested the hypothesis that potential recruits would dis- tinguish between FADs with an "assemblage" of artificial fish suspended beneath them, FADs with an "assemblage" of suspended material equal to the artificial fish in size but not resembling fish in appearance, and control FADs without anything suspended beneath them (Fig. 4). I con- structed artificial fish from 31.25-g lead fishing weights. These weights were flattened, tear-drop-shaped objects, painted a dull yellow with black bars to resemble juvenile Abudefduf troschelii and suspended, by using monofila- ment (20 lb. test) and a steel hoop, beneath the "artificial fish FADs" (Fig. 4). I suspended oblong 31.25-g lead fish- ing weights beneath "weighted FADs," and the control FADs had only a steel hoop beneath each (Fig. 4). These FADs were deployed in an anchored array, and the vari- ous treatments were distributed in an alternating pattern throughout the array. FAD positions were cleared of fishes, 840 Fishery Bulletin 101(4) Table 1 Fish species and life history stages observed at all experimental FADs (combined data) and abundance. 1 = coastal pelagic species; 2 = substrate-associated species; and 3 = po A = adult. with relative importance by frequency ssible flotsam specialists. J = juvenile; No, Species Family Stage Frequency (%) Abundance {%) 1 Abudefduf troschelii^ Pomacentridae J 34.1 814 2 Etagatis bipinnulata^ Carangidae J 16.9 8.1 3 Polydactylus approximans'^ Polynemidae J 6.4 1.8 4 Mugil sp. ■^ Mugilidae J 6.4 1.3 5 Lutjanus argentiventris^ Lutjanidae J 6.4 1.1 6 Epinephelus panamensis'^ Serranidae J 3.4 0.5 7 Hoplopagrus guntheri- Lutjanidae J 3.2 0.5 8 Canthidermis maculatus^ Balistidae J 2.8 0.4 9 Gnathanodon speciosus- Carangidae J 1.5 0.2 10 Atectis ciliaris^ Carangidae J 11 Caranx caninus^ Carangidae J 12 Caranx caballus^ Carangidae J 14.7 3.9 13 Caranx vinctus^ Carangidae J (nos. 10-14) 14 Seriola peruana^ Carangidae J 15 Tylosaurus acus pacificus^ Belonidae A 16 T. crocoditus fodiator^ Belonidae A 17 Fistularia commersonii^ Fistulariidae J 18 Syngnathus auliscus- Syngnathidae J 19 Lobotes pacificus^ Lobotidae J and A 20 Mulloidichthys dentatus^ Mullidae J 4.1 0.7 21 Sectator ocyurus^ Kyphosidae A (nos. 15-26) 22 Parapsettus panamensis^ Ephippidae J 23 Hypsoblenmus breviceps^ Blenniidae ? 24 goby (unidentified) 2 Gobiidae 7 25 Aluterus scriptus'^-^ Balistidae J and A 26 Batistes polytepis'^ Balistidae J 26 species 16 families 100 100 and treatment FADs were deployed on 24 September 1997. FADs were monitored daily as described above, from 25 September through 3 October 1997 (sampling days=9). Data were analyzed for experimental effects on total assemblage size, species diversity (S and HB), number of A. troschelii. number of £. hipinnulata, and total number of fishes minus the number of A. troschelii. Recaiit-enriched vs. nonenriched FADs I tested the hypothesis that the presence of prior recruits (juvenile sergeant major damsclhsh, Abudefduf troschelii) would have a positive effect on subsequent recruitment to a FAD. I used A. troschelii because these were the most important species associated with FADs by frequency and abundance (Table 1). It is possible that the selection of a particular species as the prior recruit might affect the sub- sequent recruitment of the same or different species (via intra- or interspecific competition for example), but I had no basis for predicting the direction of such effects. Given the strong day-to-day changes in assemblage sizes, this test required frequent, short-interval observations of the experimental FADs. I used drifting, rather than an- chored, FADs to provide a more realistic (and conservative) test of the effect. (Drifting objects should result in fewer chance encounters by potential fish recruits carried by cur- rents through a fixed FAD array, but anchored FADs are much easier to track for longer experiments.) I deployed four drifting FADs (constructed from 3 buoys — the "single" size) in the stippled area indicated in Figure 3. Two of these FADs were enriched with nine A. troschelii per FAD, previously collected from anchored FADs and released in close proximity to drifting FADs immediately after deploy- ment. The two control FADs received no sergeant majors to start. Both groups were checked immediately following deployment to verify that the fish had associated with the experimental FADs and to check against quick recruitment to the control FADs. To minimize the potential transfer of fish with the boat, I accelerated sharply when leaving a FAD enriched with sergeant majors and when checking Nelson: Fad characteristics and associated fish assemblages 841 the FADs, entered the water from the boat a minimum of 10 m from each FAD. The FADs were deployed from an inflatable boat at 50-m intervals in a roughly linear array, and checked at hourly intervals. The FADs did not maintain their initial spatial arrangement, but I did not move any FAD once the drift began unless FAD-to-FAD distance had been reduced to less than 10 m. In this instance, I moved one or more FADs to a minimum FAD-to-FAD distance of 50 m after checking for any FAD-associated fishes. In none of these instances were any FAD-associated fishes observed. I monitored the drift for four hours; deteriorating weather and fading light, however, did not permit additional observations. I used linear regression to test the hypothesis that the number of FAD-associated fishes changed over time for the enriched FADs and for the nonenriched FADs. I used a ^-test to compare the slopes of the two regression models and to test the hypothesis that the treatments accumulated fish at different rates. Results Twenty-six species of fishes from 16 families were recorded, including species associated with reef, soft bottom, and coastal pelagic habitats as adults (Table 1). Only juvenile specimens were observed clearly associated with FADs, with the exception of Ahiterus scriptus and Lobotes paci- ficus, of which both juvenile and adult forms were observed in close, continuous proximity to the FADs. Two needlefish species (Tylosaurus acus pacificus and T. crocodilus fodia- tor) appeared occasionally in close proximity to the FADs, but they were not clearly associated with the FADs. An adult Lobotes pacificus (tripletail) was observed once and a single adult Al uterus scriptus (scrawled filefish) were observed on three separate instances. Horizontal under- water visibility averaged 13.4 m (±1.7 SE) for all sampling days combined. Juvenile sergeant major damselfish (Abudefduf trosche- lii) were the dominant species by frequency of occurrence and numerical abundance (Table 1) for all experiments. The damselfish was followed in rank overall by juvenile rainbow runner (Elagatis bipinnulata), although this spe- cies was observed with the FADs only during the fouling and model fish experiments. Juvenile threadfin {Polydac- tylus approximans), mullet (Mugil sp.), and yellow snapper (Lutjanus argentiuentris) were equally frequent but dif- fered slightly in abundance (P approximans>Mugil sp.>L. argentiventns; Table 1). The latter pattern was consistent across all experiments. Specimens from a suite of juvenile carangids (excluding E. bipinnulata) were also observed frequently. Fish-removal experiments Sample date, series, and treatment combined to have a significant effect on A. troschelii abundance (three way interaction, P=0.03), but there was no clear pattern; the remaining species (combined species less numbers of A. troschelii) were influenced by sample date (date by series 30 25 20 IS 10 LU U) A 0 A Abudefduf troschelii only □ observed □ removed -j- 1 ^ 8-1 0) 1 r 3-Jul 4-Jiil 5-Jul 6-Jul 7-Jul 8-]ul 9-Jul 10-|ul B combined spp. - A. trosclielii T MA i I 3-Jul 4-Jiil 5-Jul 6-Jul 7-Jiil 8-JliI 9-Jul 10-]ul Figure 5 Repeated fish-removal effects (disturbed vs. undisturbed [observed only]) on aggregation size (mean no. of fishes) for Abudefduf troschelii alone (A) and for all species combined less A. troschelii (B). See Tables 3-5 for sample sizes and ANOVA results for assemblage size and diversity measures. interaction, P<0.01) but not by treatment (P=0.73, Table 2, Fig. 5). Measures of diversity varied between series (series: S, P<0.01; HB, P=0. 01) but were unaffected by treatment. Thus, fish removal or fish disturbance may contribute to assemblage sizes for individual species (e.g. A. troschelii), but, in the present study, the total number of combined species was unaffected. FAD size Abudefduf troschelii was strongly affected by a combina- tion of treatment and sample date (date by treatment interaction, P=0.03, Table 3, Fig. 6). Results from the remaining species combined were comparable with larger total numbers at the larger FADs, although not statisti- cally significant (treatment, P=0.07). Although both mea- sures of diversity (S and HB) suggested that the treatment may have had a positive effect on diversity (S, treatment, P=0.02), species richness was positively correlated with sample size. HB, a diversity measure comparatively unaf- 842 Fishery Bulletin 101(4) Table 2 Repeated measures ANOVA results (cleared FADs vs. undisturbed FADs, n. =^2=12) in fish-removal experiment. Dependent variable Factor(s) F df P 1-/J Number of fishes treatment 0.12 1, 12 0.73 0.06 (all species combined) series 35.3 1,12 <0.01 >0.99 treatment x series 0.64 1, 12 0.44 0.11 date 5.53 2,24 0.01 0.81 date X series 48.5 2,24 <0.01 >0.99 date X treatment 3.22 2,24 0.06 0.55 3-way interaction 0.06 2,24 0.94 0.06 Number offish treatment 16.7 1,12 <0.01 0.97 (A. troschelii only) series 5.92 1,12 0.03 0.61 treatment x series 0.27 1, 12 0.61 0.08 date 3.10 2,24 0.06 0.53 date X series 7.71 2,24 <0.01 0.93 date X treatment 2.95 2,24 0.07 0.51 3-way interaction 4.24 2,24 0.03 0.69 Number of fishes treatment 0.05 1, 12 0.82 0.06 (all spp. minus A. troschelii) series 13.3 1, 12 <0.01 0.93 treatment x series 0.01 1,12 0.94 0.05 date 2.93 2,24 0.07 0.51 date x series 4.06 2,24 0.03 0.66 date X treatment 0.07 2,24 0.93 0.06 3-way interaction 0.31 2,24 0.74 0.09 Species richness (S) treatment 0.63 1,12 0.44 0.11 series 11.8 1,12 <0.01 0,90 treatment x series 0.63 1,12 0.44 0.11 date 0.43 2,24 0.66 0.11 date X series 0.63 2,24 0.54 0.14 date X treatment 0.69 2,24 0.51 0.15 3-way interaction 2.08 2,24 0.37 0.37 Species diversity (UB)' treatment 1.05 1, 10 0.33 0.15 series 8.79 1, 10 0.01 0.79 treatment x series 0.54 1,10 0.48 0.10 date 1.43 2,22 0.26 0.27 date X series 2.80 2,22 0.08 0.49 date X treatment 0.12 2,22 0.12 0.07 3-way interaction 0.61 2,22 0.55 0.14 ' Missing data were replaced accordi to interactions between treatment ng to the directions in Zar (1996), and the degrees of freedom series, and dates. were reduced accordingly. "3-way interaction" refers fected by sample size (Magurran, 1988), was marginally nonsignificant (HB, treatment, P=0.07, Table 3). cies diversity (HB), though not richness (S), was signifi- cantly affected by sample date (P=0.02). Presence or absence of a fouling community Treatment and sample date combined to have a significant efTect on the number of A. troschelii (date by treatment interaction, P<0.01) — an effect contributing to the similar significant interaction effect for all species combined (Fig. 7, Table 4). Although the mean numbers of fish(es) were consistently higher at fouled FADs i'or E. bipinnulata alone and for all species minus A. troschelii, the only significant main efTects were due to sample date (Table 4, Fig. 7). Spe- Artificial fish experiment Experimental treatments (FADs with model fish, with lead weights or with nothing, Fig. 4) had no effect on any mea- sured parameter — combined species, A. troschelii alone, E. bipinnulata alone, combined species less A. troschelii, spe- cies richness and diversity (Fig. 8, Table 5). All measures were significantly affected (P<0.01) by sample date except for£. bipinnulata alone (date, P=0. 48). Although individual FADs varied in the number of associated E. bipinnulata. Nelson: Fad characteristics and associated fish assemblages 843 Table 3 FAD-size effects and repeated measures ANOVA results (single FADs vs triple-size FADs fll=n2=12). Dependent variable Factor(s) F df P 1-/3 Number of of fishes treatment 24.1 1,6 <0.01 0.99 (all species combined) date 10.7 2,12 <0.01 0.97 date X treatment 2.56 2,12 0.12 0.41 Number offish treatment 10.9 1,6 0.02 0.79 (A. troschelii only) date 18.2 2,12 <0.01 >0.99 date X treatment 4.55 2,12 0.03 0.66 Number of fishes treatment 4.84 1,6 0.07 0.45 (all spp-A. troschelii) date 0.42 2,12 0.67 0.10 date X treatment 0.29 2, 12 0.75 0.09 Species richness (S) treatment 11.3 1,6 0.02 0.81 date 0.38 2,12 0.69 0.10 date X treatment 0.38 2,12 0.69 0.10 Species diversity (HB) treatment 5.00 1,6 0.07 0.46 date 3.39 2,12 0.07 0.52 date X treatment 0.40 2,12 0.68 0.10 these numbers were strikingly constant across sample date and, to a lesser extent, across treatments (Fig. 8). Recruit-enriched vs. nonenriched FADs Enriched FADs showed significantly higher rates of recruitment than nonenriched FADs: the regression line for the enriched FADs had a significant slope (f , j g|=20.76, P<0.01), but the regression line for the nonenriched FADs did not (F,i g)=2.29, P=0.17; Fig. 6). All additional fish were juvenile sergeant major damselfish, Abudefduf troschelii . These slopes are significantly different (f=3.05, 2 tailed test, v=6, P=0.02; Fig. 9); enriched FADs accumulated fish at a significantly higher rate (2.5 fish per hour) than did nonenriched FADs that accumulated fish at a rate of 0.1 fish per hour Horizontal underwater visibility was 15 m at the beginning of the experiment. Discussion FAD size, the presence of a fouling community, and the presence of prior recruits all had positive effects on the size of FAD-associated assemblages, although the latter factor was assessed over a period of hours, whereas the former were assessed over days. The repeated removal of an exist- ing assemblage also had significant effects due at least partially to treatment, but in all of these analyses sample date appeared to play the largest role in determining the numbers of fish(es) at these FADs. The presence of artifi- cial fish or comparable-size weights did not significantly affect assemblage sizes. There was little support for the hypothesis that any of these factors might affect the spe- cies diversity of these assemblages; only species richness was significantly increased along with an increase in FAD size and this result may be an effect of assemblage size rather than object characteristics. Where treatment effects did significantly affect the numbers of fishes, their effects 50- 45- 40- 35- 30- 25 20 15- 10- 5 0 10- 9- 8- 7 6 5- 4- 3- 2- 1- 0-t A Abudefduf troschelii only D single D triple 26-Jul 28-Jul 30-Jul B comb ned spp. - A. troschelii m 26-Jul 28-Jul 30-Jul Figure 6 FAD-size effects (single FADs vs. triple-size FADs) on aggregation size (mean no. of fishes) for Abudefduf trosche- lii alone (A) and for all species combined less A. troschelii (B). See Tables 3-5 for sample sizes and ANOVA results for assemblage size and diversity measures. 844 Fishery Bulletin 101(4) A Abudi'fduf irosdiclii only .w- D Clean T 2S- n Fouled 20- X. Ky?. T m. ]b- % ID- ■/>. S- T '■* T T 0- 1 1 ' ^ 6 5 4- 3- 9-Sep 10-Sep 11 -Sep 12-Sep B combined spp. - A. troschclii p.,, 0- m W 6- 5- 4- 3- 2 - 1- 9-Sep 10-Sep 11 -Sep C Lla%at'\s bipi)nuilnta only 12-Sep mi 9-Sep 10-Scp 11 -Sop 12-Sep Figure 7 Effects of fouling community (fouled vs. unfouled [control]) on aggregation size (mean no. of fishes) ior Ahudefduf tros- chelii alone (A), for all species combined less A. troschelii (B), and for Elagatis bipinnulata alone (C). See Tables 3-5 for sample sizes and ANOVA results for assemblage size and diversity measures. on Abudefduf troschelii were generally the strongest. It is not clear whether this is a species-specific effect or if these results are due to the fact that A. troschelii was the most numerically important species. The absence of a significant treatment main effect in the fish-removal experiments suggests that recruitment and loss from these anchored FADs is sufficiently rapid so that I20-| A Abudefduf troschelii only 1011- n Control ■- so- □ VVi'ighls J 60- ■ Models r 40- _ -p ■r T 20- 0 J fi.Jf1 \\ \i\ 1 C^D-D-Q-Ci-C-t; r, T, ^ ^ ^ rj Z} zj f^ r~^^ r^ 0.99 (A. troschelii only) sate 19.7 3,18 <0.01 >0.99 date X treatment 25.2 3,18 <0.01 >0.99 Number offish treatment 1.06 1,6 0.34 0.14 (£, bipinnulata only) date 4.89 3,18 0.01 0.84 date X treatment 2.15 3,18 0.13 0.45 Number of fishes treatment 3.01 1,6 0.13 0.30 (all spp. -minus A. troschelii) date 3.88 3,18 0.03 0.73 date X treatment 2.27 3,18 0.12 0.47 Species richness (S) treatment 2.49 1,6 0.17 0.26 date 3.00 3,18 0.06 0.60 date X treatment 0.60 3,18 0.62 0.15 Species diversity (HB) treatment 1.74 1,6 0.23 0.19 date 4.35 3,18 0.02 0.79 date X treatment 1.91 3,18 0.16 0.40 Table 5 Repeated measures ANOVA results (control [no weights and no artificial fish] vs. weights and vs. artificial fish, n =^2=13=27) in the artificial fish experiment. Dependent variable Factor(s) F df P 1-/3 Number of fishes treatment 0.13 2,6 0.88 0.06 (all species combined) date 3.52 8,48 <0.01 0.97 date X treatment 0.37 8,48 0.98 0.20 Number offish treatment 0.12 2,6 0.89 0.06 (A. troschelii only) date 3.34 8,48 <0.01 0.95 date X treatment 0.41 8,48 0.97 0.22 Number of fishes treatment 0.29 2,6 0.76 0.08 (all spp. minus A. troschelii) date 4.82 8,48 <0.01 >0.99 date X treatment 0.73 8,48 0.75 0.41 Number offish treatment 0.05 2,6 0.95 0.06 (E. bipinnulata only) date 0.95 8,48 0.48 0.38 date X treatment 0.36 8,48 0.94 0.20 Species richness (S) treatment 0.41 2,6 0.68 0.10 date 9.87 8,48 <0.01 >0.99 date X treatment 0.90 8,48 0.58 0.50 Species diversity (HB)' treatment 0.13 2,5 0.88 0.06 date 4.44 8,47 <0.01 0.99 date X treatment 1.00 8,47 0.45 0.56 ' Missing data were replaced according to the directions in Zar (1996), and the degrees of freedom were reduced accordingly. results are consistent with those obtained by Wickham and Russell (1974). A similar result would occur if these FADs had a predictable carrying capacity and recruitment was suf- ficiently rapid that removal of the assemblage was followed by its replacement before the next observation. However, assemblage sizes within treatments varied widely from one day's observations to the next; therefore recruitment, rather than carrying capacity, seems to determine assemblage size. 846 Fishery Bulletin 101(4) f 1?- E 10- - Abudcfduf twfchelii enriched i-ADs Y =- 2.2X + s.y non-enriched FADs Y = 0.1X-0,1 0-$ 00:00 -*- -^ ■V 01:00 02:00 03:00 04:00 Elapsed time (tih.mm) Figure 9 Changes in the number of fish associated with enriched (solid circles) and nonen- riched (open circles) drifting FADs over time. Enriched FADs showed significantly higher rates of recruitment than did nonenriched FADs; the slopes of the regression lines are significantly different ((=3.05, 2-tailed test, v=6,P=0.02). There appears to be insufficient time or stability for such factors as competition or predation to influence the size or diversity of these FAD-associated fish assemblages. De- spite these results, individual fish do remain with a specific FAD for days: On at least five separate occasions associated with some of the other experiments described in this paper, individuals recognizable by scars and bite marks were sighted repeatedly as many as six days after the initial observation (Nelson, unpubl. data). Although the FADs and the associated fishes described in the present study are not directly comparable to FADs and fishes targeted in fisheries- scale operations, these experiments are among the first con- trolled efforts at understanding the effects of disturbance or fishing for FAD-associated assemblages. The average assemblage size for all experiments and treatments varied considerably, often significantly, over time (Tables 2-5). Significant interaction effects between sample date and FAD treatments may be indicative of day- to-day recruitment fluctuations, dependent upon recruit- ment variability. A significant interaction may result when these effects are large and are in evidence regardless of the experimental treatment (i.e. occur in concert across treat- ments). Significant sample date effects (and series effects in the fish-removal experiment, Table 2) are likely a result of temporal fluctuations in the numbers of fishes available to recruit to the FADs. Note that the two series in the fish-removal experiment differed not only in which FADs were given a particular treatment (positional effects), but also in time — the two scries were necessarily run consecutively, not concurrently. I believe, however, it to be unlikely that positional effects influenced any of the results reported in the present study: treatments were assigned to FADs within the arrays in such a way as to ensure that inshore, offshore, or longshore positions were equally weighted among treatments. Signifi- cant series main effects, independent of additional factors, were found only for species richness and diversity (HB) — a result I attribute to changes in the availability of potential recruit species. Temporal patterns of juvenile reef fish re- cruitment are often variable and may be affected by such factors as spawning periodicity (Love et al., 1990), variable predation (Nelson, 2001), or changing physical oceano- graphic processes (Doherty, 1991; Levin, 1994; Kingsford and Finn, 1997). Rountree ( 1989), also, found that the mean numbers of the most abundant species observed around a FAD array off South Carolina varied widely during FAD deployment, albeit over a much longer time period (nearly 200 days). Thus, differences in assemblage size and diver- sity over time are not unexpected. FAD size had significant, positive effects on assemblage size and species richness. Although tripling the FAD size resulted in a nearly threefold increase in the number of as- sociated fishes (combined species), the response may not be linear. (Note, however, that Rountree (1989) demonstrated that the number of Decapterus punctatus associated with midwater FADs exhibited a significant, positive linear re- sponse to FAD size.) Further research will be necessary to resolve the effect of FAD size on numbers of aggregating fishes. Also of interest is the significant increase in species richness attributable to increased FAD size. Bortone et al. (1977) suggested that species diversity may be a function of "clump size" for Sargassum-associated fish assemblag- es, and Moser et al. (1998) found greater numbers offish species under large ( 10-20 m diameter) mats of floating Sargassum than they did under smaller clumps (<1 m diameter) or in open water. However, the changes in spe- cies richness from this experiment could well be an effect of assemblage size; treatment effects on species diversity measured using the Brillouin index (HB) were marginally nonsignificant (Table 3, P=0.07). Significant sample date differences in treatment and evenness are due to large fluctuations in the abundance of the dominant species, Abudefduf troschelii, ranging at the triple-size FADs from 1 to 55 individuals over the course of 11 days. Fishes were five times more numerous on average at fouled FADs than they were at comparable FADs lacking fouling organisms, but measures of diversity showed no significant treatment effect (Table 4). There was a signifi- cant interaction between treatment and sample date for the present experiment (Table 4) that may have been due to fluctuations in assemblage sizes among sample dates across both FAD treatments. The species composition of these assemblages was similar to that of other experi- ments, except that Elagatis bipinnulata were regularly observed: Abudefduf troschelii were the dominant species by abundance, followed by E. bipinnulata, and Mugil sp. All were small, young-of-the-year fishes (the largest E. bi- pinnulata individuals reached approximately 80 mm SL) and seemed not to be feeding on the larger invertebrates forming much of the colonizing community. During casual observations of FAD-associated fishes, I observed fish feed- ing on plankton carried past the FADs, but no physical contact with the FAD or fouling organisms. Ibrahim et al. Nelson: Fad characteristics and associated fish assemblages 847 (1996) reported that none of the gut contents from FAD- associated fishes included sessile organisms found on their FADs (fish size ranges included specimens 8-14, 15-99, and >100 mm SL — the first two size categories are comparable to the fishes in the present study). Larger, piscivorous fishes do feed at least occassionally on smaller fishes associated with floating objects (Gooding and Magnuson, 1967), but published gut content studies are conflicting. Some sug- gest that piscivorous species that associate with flotsam rely on other sources of food (e.g. Gooding and Magnuson, 1967; Hunter and Mitchell, 1967; Brock, 1985), while others suggest that flotsam- (or algae-) associated fishes form an important food resource for these larger piscivorous fishes (Dooley, 1972; Manooch et al., 1984; Coston-Clements et al, 1991). Morgan et al. (1985) noted the occurrence of at least two members of the Sargosswrn -associated invertebrate fauna among the stomach contents of several species of pelagic fishes. From the perspective of flotsam- or FAD-as- sociated fishes, opportunistic predation by piscivores that do not associate with FADs may be more important than predation by other members of the assemblage. Additional gut content data from juvenile and nonpiscivorous fishes are sorely lacking. I address possible explanations for the results of the present study below. I recorded no significant treatment effect attributable to differences between FADs with artificial fish, FADs de- ployed with artificial-fish-sized weights, or control FADs. I attribute significant sample date effects to day-to-day changes in constituent individuals and the fluctuating availability of potential recruits. Numbers of £. bipin- nulata were strikingly constant across treatments and sample dates in this experiment (Table 5, Fig. 8) and in the fouling experiment (Fig. 7) and seemed to indicate an apparently unusual characteristic of this species — individ- uals remaining associated with a given FAD for multiple days. Although the experiment was intended to distinguish between FADs with prior recruits versus FADs without prior recruits, the lack of a significant treatment effect does not negate the possibility that potential recruits would distinguish between occupied and unoccupied FADs. The painted artificial fish and lead weights clearly lacked many attributes of living fish. However, comparable numbers of recruits found at all treatments suggest that a change in the structural complexity of the FADs did not affect assem- blage size or diversity. Although the addition of four small lead weights (artificial fish were painted and oriented differently but were still lead weights) did not appear to increase appreciably the visible surface area of those FADs, the subsequent experiment with live fishes instead of ar- tificial fish had a dramatic effect on recruitment; therefore sizeable changes in the physical size of a FAD may be nec- essary to yield a response in fish recruitment. The potential roles of structural complexity and orientation of FADs will be informative areas for future research. Past investiga- tions in these areas (e.g. Hunter and Mitchell, 1968; Klima and Wickham, 1971; Wickham et al., 1973) have provided a useful beginning, but more work is needed. Although sample sizes were small, the presence of prior Abudefduf troschelii "recruits" (enriched FADs) had a sig- nificant effect on patterns of subsequent recruitment; this effect contrasted sharply with FADs lacking fish at the start of this experiment (nonenriched FADs). For this spe- cies, these results point to a social aspect to these aggrega- tions, and sociality may also be involved in the recruitment of other species, particularly the schooling fishes Caranx spp., Polydactylus approximans, and Mugil spp., as sug- gested for some scombrids (e.g. Dagorn and Freon, 1999). The addition of fishes below a FAD may increase recruit- ment rates by rendering the object more visible, although the artificial fish experiment indicated that simply adding fish-size objects beneath a FAD does not affect recruitment. Comparisons between these two experiments are tenuous, however, because the artificial fish experiment employed anchored FADs observed over a period of days, whereas the enriched FAD experiment used drifting FADs observed over a course of hours. Why do FAD size, the presence of a fouling community, and the actual presence of prior recruits at a FAD each have the effect of increasing the size and, possibly, the di- versity of FAD-associated assemblages of juvenile fishes? The simplest explanation is that these factors contribute to the target strength of the object, increasing the visual, olfactory, or auditory stimulus (or some combination) of the floating object. Larger objects should be easier to find, especially if potential recruits rely on vision to explore their environment. Kellison and Sedberry (1998) found that the fishes associated with mid-water floating struc- tures that were tethered to an artificial reef decreased in abundance over time (193 days), and suggested that the loss in buoyancy associated with the development of a fouling community may have reduced the effective size of these floating objects, accounting for fewer associated fishes (see also Hunter and Mitchell, 1968; Rountree, 1989). To account for the positive effects of a fouling community observed in the present study, it seems reasonable to sup- pose that fouling organisms may be detected by olfactory means; Sweatman ( 1988) has shown that some larval fishes use olfactory cues for settlement on reefs. Further experi- ments, for example experiments controlling for FAD size, odor cues, and visibility of the FADs, are needed to deter- mine why some of these factors exhibit these effects. Future research on the role of flotsam as shelter from predators and as a conveyance to suitable habitat could yield evolutionary explanations for the attraction to float- ing objects. For these small fishes, such objects likely repre- sent a shelter from predators (Mitchell and Hunter, 1970). Some species do respond to the approach of an observer by positioning themselves so that the FAD is between them and the observer. Particularly during daylight and crepus- cular hours when visually-oriented predators are most active, flotsam may offer refuge in a habitat where there is little alternative refuge. During the day, when onshore winds drive drifting objects towards shallow water, flotsam and drift algae, unlike anchored FADs, may also offer a comparatively safe conveyance to more suitable habitat. Thus, there may be adaptive advantages for juvenile reef fishes in associating with floating objects. Although the juvenile fishes associated with the FADs used in the present study are not of interest to any fishery, the patterns observed from them may be relevant to FADs 848 Fishery Bulletin 101(4) deployed commercially to aggregate fish species at various life history stages. FAD size is clearly relevant to those in- terested in studying potential improvements to FAD design. Carefully controlled studies on the importance of surface area versus volume and the orientation of FAD structures are needed. The role of a fouling community, too, deserves further investigations. Although a fouling community may weigh down streamers (trailing pieces of buoyant material intended to increase the subsurface of area of a FAD), such a community may also improve recruitment and possibly retention of recruits around a FAD. Finally, the importance of the initial recruits to a floating object should be studied further. Enriching a FAD may increase the speed at which additional fishes are recruited. Improved artificial fish may prove more effective than the items used in the present re- search. FADs are an important tool in a number of artisanal (small-scale fishery based on traditional methods), sport, and commercial fisheries, especially in tropical waters where FAD fisheries particularly target tunas (Scomb- ridae), jacks (Carangidae), and Coryphaena spp. (Galea, 1961; Klima and Wickham, 1971; Beets, 1989; Hilborn and Medley 1989; Friedlander et al., 1994; Higashi, 1994; Hall et al., 1999b). Due largely to the potential for fisheries enhancement, considerable research has been focused on the importance of floating-object characteristics and the numbers of fishes attracted to such objects; however, the results have been difficult to interpret and are often con- flicting (Rountree, 1989; Kingsford, 1993; Druce and Kings- ford, 1995). Because log sets in tuna purse-seine fisheries (where fishermen target fish associated with drifting logs or FADs) are associated with high levels of bycatch (Hall, 1998; Lennert-Cody and Hall, 2000), the behavior and ecology of flotsam-associated species is in urgent need of study so that a means of reducing bycatch may be devised. This study made use of FADs floating at the surface; studies by other researchers have employed similar tools or they have used FADs tethered in mid-water. No one has examined the effects of FAD position in relation to the sur- face, and the implicit assumption appears to be that there is no biologically significant difference. This assumption has not been tested, although comparisons between data from floating structures, whether at the surface, mid-water, or tethered close to the bottom, are common in the literature. I have made comparisons between my data from surface FADs and results from mid-water FADs (e.g. Wickham and Russell, 1974; Rountree, 1990); such comparisons may be misleading and should be interpreted with caution. The results from the present study indicate that turn- over rates at nearshore anchored FADs are high and that undisturbed FAD assemblages may show little difference in these rates from disturbed FADs. Fishes recruiting to these FADs discriminate among potential floating objects, forming larger, more species-rich assemblages around tri- ple-size FADs than around single FADs. FADs possessing a fouling biota also attract larger (though no more diverse) assemblages than do clean FADs. The latter effect was complicated by temporal fluctuations that overlay these treatment effects, resulting in day-to-day changes in the total numbers of fishes in both treatments (Table 4, Fig. 7). Further, the presence of prior recruits in the enrichment experiment had a strong effect on subsequent recruitment. Thus, the association of juvenile fishes with floating objects is not a haphazard process, and floating-object character- istics play potentially important roles in fish recruitment to these objects. These results suggest that associating with flotsam may be adaptive, rather than an accidental behavior and support Kingsford's hypothesis (Kingsford, 1993 ) that floating material is an important environmental component in the relationship between environment and some juvenile fishes. Acknowledgments For help in the field, I am grateful to I. Nelson and D. Mansue. W. L. Montgomery, S. Shuster, and to two anony- mous reviewers who provided helpful criticism. Funding was provided by the American Museum of Natural His- tory (Lerner-Gray Fund), American Society of Ichthyolo- gists and Herpetologists (Raney Award), Animal Behavior Society, International Women's Fishing Association (Max Coan Memorial Scholarship), Seaspace/Houston Underwa- ter Club, Smithsonian Tropical Research Institute (STRI), and Sigma Xi. D. R. Robertson of STRI and D. Margulise, R. Olson, and V. Scholey of the Inter-American Tropical Tuna Commission provided invaluable advice and logisti- cal support. Literature cited Beets, J. 1989. Experimental evaluation of fish recruitment to com- binations offish aggregating devices and benthic artificial reefs. Bull. Mar. Sci. 44:973-983. Bortone, S., P. A. Hastings, and S. B. Collard. 1977. The pelagic-Sargassum ichthyofauna of the eastern GulfoflMexico. Northeast Gulf Sci. 1:60-67. Brock, R. E. 1985. Preliminary study of the feeding habits of pelagic fish around Hawaiian fish aggregation devices or can fish aggregation devices enhance local fisheries productivity? Bull. Mar Sci. 37:40-49. Coston-Clements, L., L. R. Settle, D. E. Hoss, and F. A. Cross. 1991. Utilization of the Sar^assum habitat by marine inver- tebrates and vertebrates — a review. NOAA Tech. Memo. NMFS-SEFSC-296:32. Crawford, R. E., and J. K. Jorgenson. 1993. Schooling behaviour of arctic cod, Boreogadiis saida, in relation to drifting pack ice. Environ. Biol. Fish. 36: 345-357. Dagorn, L., and P. Freon. 1999. Tropical tuna associated with floating objects: a simu- lation study of the meeting point hypothesis. Can. J. Fish. Aquat. Sci. 56:984-993. Doherty, P. J. 1991. Spatial and temporal patterns in recruitment. InThe ecology of fishes on coral reefs (P. F. Sale, ed.), p. 261-294. Academic Press, San Diego, CA. Dooley, J. 1972. Fishes associated with the pelagic Sargassum com- plex, with a discussion of the Sargassum community. Con- Irib Mar Sci. 16:1-32. Nelson: Fad characteristics and associated fish assemblages 849 Druce, B. E., and M. J. Kingsford. 1995. An experimental investigation on the fishes associ- ated with drifting objects in coastal waters of temperate Australia. Bull. Mar. Sci. 57;378-392. Fedoryako, B. I. 1989. A comparative characteristic of oceanic fish assem- blages associated with floating debris. J. Ichthyol. 29: 128-137. Friedlander, A., J. Beets, and W. Tobias. 1994. Effects offish aggregating device design and location on the fishing success in the U.S. Virgin Islands. Bull. Mar. Sci. 55:592-601. Gaertmer, D., and M. Medina-Gaertner. 1999. An overview of the relationship between tunas and floating objects in the south of Caribbean Sea. lATTC (Inter-American Tropical Tuna Commission), Spec. Rep. 11:66-86. Galea, J. A. 1961. The "Kannizzati" fishery. Gen. Fish. Counc. Mediterr. Sess. Rep. 6:85-91. Gooding, R. M., and J. J. Magnuson. 1967. Ecological significance of a drifting object to pelagic fishes. Pac. Sci. 21:486-497. Hall, M. A. 1998. An ecological view of the tuna-dolphin problem: impacts and trade-offs. Rev. Fish Biol. Fish. 8:1-34. Hall, M. A., D. L. Alverson, and K. I. Metuzals. 2000. By-catch: problems and solutions. Mar. Pollut. Bull. 41:204-219. Hall, M. A., M. Garcia, C. Lennert-Cody, R Arenas, and F. Miller. 1999a. The association of tunas with floating objects and dol- phins in the eastern Pacific Ocean: a review of the current purse-seine fishery. lATTC, Spec. Rep. 11:87-194. Hall, M. A., C. Lennert-Cody, M. Garcia, and P. Arenas. 1999b. Characteristics of floating objects and their attrac- tiveness for tunas. lATTC, Spec. Rep. 11:396-446. Hallier, J. P., and J. I. Parajua. 1999. Review of tuna fisheries on floating objects in the Indian Ocean. lATTC, Spec. Rep. 11:195-221. Hampton, J., and K. Bailey. 1999. Fishing for tunas associated with floating objects: review of the western Pacific fishery. lATTC, Spec. Rep. 11:222-284. Higashi, G. R. 1994. Ten years of fish aggregating device (FAD) design development in Hawaii. Bull. Mar Sci. 55:651-666. Hilborn, R., and P. Medley. 1989. Tuna purse-seine fishing with fish-aggregating devices (FAD): models of tuna FAD interactions. Can. J. Fish. Aquat. Sci. 46:28-32. Hunter, J. R., and C. T Mitchell. 1967. Association of fishes with flotsam in the offshore waters of Central America. Fish. Bull. 66:13-29. 1968. Field experiments on the attraction of pelagic fish to floating objects. J. Cons. Int. Explor Mer 31:427-434. Ibrahim, S., M. A. Ambak, L. Shamsudin, and M. Z. Samsudin. 1996. Importance offish aggregating devices (FADs) as sub- strates for food organisms offish. Fish. Res. 27:265-273. Kellison, G. T, and G. R. Sedberry 1998. The effects of artificial reef vertical profile and hole diameter on fishes off South Carolina. Bull. Mar. Sci. 62: 763-780. Kingsford, M. J. 1992. Drift algae and small fish in coastal waters of north- eastern New Zealand. Mar. Ecol. Progr. Ser. 80:41-55. 1993. Biotic and abiotic structure in the pelagic environ- ment: importance to small fishes. Bull. Mar. Sci. 53:393- 415. 1995. Drift algae: a contribution to near-shore habitat com- plexity in the pelagic environment and an attractant for fish. Mar Ecol. Progr Ser. 116:297-301. Kingsford, M. J., and M. Finn. 1997. The influence of phase of the moon and physical pro- cesses on the input of presettlement fishes to coral reefs. J. Fish Biol. 51:176-205. Klima, E. F, and D. A. Wickham. 1971. Attraction of coastal pelagic fishes with artificial structures. Trans. Am. Fish. Soc. 1:86-99. Kojima, S. 1960. Fishing for dolphins in the western part of the Japan Sea. V. Species of fishes attracted to bamboo rafts. Bull. Jpn. Soc. Sci. Fish. 26:379-382. Lennert-Cody, C. E., and M. A. Hall. 2000. The development of the purse seine fishery on drift- ing fish aggregating devices in the eastern Pacific Ocean: 1992-1998. In Peche thoniere et dispositifs de concentra- tion de poissons: colloque DCP, Martinique, Octobre 1999, p. 78-107. Institut Frangais de Recherche pour I'Exploita- tion de la Mer, Issy-les-Moulineaux, France. Levin, P S. 1994. Fine-scale temporal variation in recruitment of a tem- perate demersal fish: the importance of settlement versus post-settlement loss. Oecologia 97:124-133. Love, M. S., M. H. Carr, and L. J. Haldorson. 1990. The ecology of substrate-associated juveniles of the genus Sebastes. Environ. Biol. Fish. 30:225-243. Magurran, A. E. 1988. Ecological diversity and its measurement, 179 p. Princeton Univ. Press, Princeton, NJ. Manooch, C. S., Ill, D. L. Mason, and R. S. Nelson. 1984. Food and gastrointestinal parasites of dolphin Cory- phaena hippurus collected along the southeastern and Gulf coasts of the United States. Bull. Jpn. Soc. Sci. Fish. 50: 1511-1525. Massuti, E., S. Deudero, P. Sanchez, and B. Morales-Nin. 1998. Diet and feeding of dolphin (Coryphaena hippurus) in western Mediterranean waters. Bull. Mar. Sci. 63: 329-341. McCleneghan, K., and J. L. Houk. 1978. A diver-operated net for catching large numbers of juvenile marine fishes. Calif. Fish Game 64:305-307. Mitchell, C. T, and J. R. Hunter. 1970. Fishes associated with drifting kelp, Macrocystis pyr- ifera, off the coast of southern California and northern Baja California. Calif Fish Game 56:288-297. Morgan, S. G., C. S. Manooch III, D. L. Mason, and J. W. Goy 1985. Pelagic fish predation on Cerataspis, a rare larval genus of oceanic penaeoids. Bull. Mar. Sci. 36:249-259. Moser, M. L., P. J. Auster, and J. B. Bichy. 1998. Effects of mat morphology on large Sargassum-assoc'i- ated fishes: observations from a remotely operated vehicle (ROV) and free-floating video camcorders. Environ. Biol. Fish. 51:391-398. Nelson, P. A. 1999. The ecology and behavior of flotsam-associated ma- rine fish aggregations. Ph.D. diss., 137 p. Northern Ari- zona Univ., Flagstaff, AZ. 2001. Behavioral ecology of young-of-the-year kelp rock- fish, Sebastes atrovirens Jordan and Gilbert (Pisces: Scor- paenidae). J. Exp. Mar. Biol. Ecol. 256:33-50. 850 Fishery Bulletin 101(4) Parin, N. V., and B. I. Fedoryako. 1999. Pelagic fish communities around floating objects in the open ocean. lATTC, Spec. Rep. 11:447-458. Rountree, R. A. 1989. Association of fishes with fish aggregation devices: effects of structure size on fish abundance. Bull. Mar. Sci. 44:960-972. 1990. Community structure of fishes attracted to shallow water fish aggregation devices off South Carolina, U.S.A. Environ. Biol. Fish. 29:241-262. Safran, P. 1990. Drifting seaweed and associated ichthyofauna: fioat- ing nursery in Tohoku waters. Mer 28:225-239. Safran, P., and M. Omori. 1990. Some ecological observations on fishes associated with drifting seaweed off Tohoku coast, Japan. Mar. Biol. 105:395-402. Suzuki, Z. 1999. Distribution of floating logs in the Pacific and purse seine sets on tunas associated with logs by Japanese boats in the tropical western and central Pacific. lATTC, Spec. Rep. 11:459-479. Sweatman, H. P. A. 1988. Field evidence that settling coral reef fish larvae detect resident fishes using disolved chemical cues. J. Exp. Mar Biol. Ecol. 124:163-174. Wickham, D. A., and G. M. Russell. 1974. An evaluation of mid-water artificial structures for attracting coastal pelagic fishes. Fish. Bull. 72:181-191. Wickham, D. A., J. W. Watson Jr., and L. H. Ogren. 1973. The efficacy of midwater artificial structures for attract- ing pelagic sport fish. Trans. Am. Fish. Soc. 3:563-572. Zar, J. H. 1996. Biostatistical analysis, 3rd ed., 662 p. Prentice Hall, Upper Saddle River, NJ. 851 Abstract— The bastard grunt 'Pomada- sys incisus) is one of the most abundant coastal demersal fishes inhabiting the Canary Islands. Age and growth were studied from samples collected between October 2000 and September 2001. Growth analysis revealed that this species is a fast growing and moderately short-lived species (ages up to seven years recorded). Length- at-age was described by the von Ber- talanffy growth model (L_=309.58 mm; A=0.220/year; ((,=-1.865 year), the Schnute growth model (>'j= 126.66 mm; .V2=293.50 mm; a=-0.426; 6= 5.963), and the seasonalized von Bertalanffy growth model (L„=309.93 mm; k=0:218/ year; <„= -1.896 year; C=0.555; (^=0.652). Individuals grow quickly in their first year, attaining approximately 60% of their maximum length; after the first year, their growth rate drops rapidly as energy is probably diverted to repro- duction. The parameters of the von Bertalanffy weight growth curve were W, =788.22 mm; *=0.1567/year; <„= -1.984 year Fish total length and otolith radius were closely correlated, r'^=Q.912. A power relationship was estimated between the total length and the oto- lith radms (a=49.93; v'=0.851). A year's growth was represented by an opaque and hyaline (translucent) zone — an annulus. Backcalculated lengths were similar to those predicted by the growth models. Growth parameters estimated from the backcalculated sizes at age were L_=315.23 mm; ft=0.217/year; and (q= -1.73 year Age and growth of the bastard grunt (Pomadasys incisus: Haemulidae) inhabiting the Canarian archipelago. Northwest Africa Jose G. Pajuelo Jose M. Lorenzo Departament of Biology University of Las Palmas de Gran Canaria Campus Universitano de Tafira 35017 Las Palmas de Gran Canaria, Spam E-mail address (for J G Pajuelo) |paguelo@dbio.ulpgc.es Muriel Gregoire Faculty of Biology University of Liege Liege, Belgium Present address: Departament of Biology University of Las Palmas de Gran Canaria Campus Universitano de Tafira 35017 Las Palmas de Gran Canaria, Spain Manuscript approved for publication 19 June 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:851-859 (2003). The family Haemulidae consists of 16 genera (126 species), including the genus Pomadasys, This genus is represented by 37 species distributed around the world. Members of this family are commonly referred to as grunt (Bauchot and Huraeu, 1990). Only the bastard grunt (Pomadasys incisus (Bowdich, 1825)) is found off the Canary Islands. The bastard grunt is a coastal demersal fish species inhabiting marine and brackish waters along the eastern central Atlantic coasts from the Strait of Gibraltar to Angola, and also in the Canaries, Azores, and Cape Verde Islands (Bauchot and Hureau, 1990). In the Canary Islands, where the bastard grunt is one of the most abundant spe- cies, it has been observed in high densi- ties in schools along coastal waters. Information on the biology of the bastard grunt is not available any- where in the world. Despite its wide- spread occurrence, the bastard grunt has no commercial value for its low quality meat and it is discarded in the Canarian archipelago. The need for a biologically based discard management strategy and the paucity of data avail- able on the biology of this species have prompted an investigation into aspects of its life history. We report aspects of age and growth, which are important parameters in models for managing the population of the bastard grunt off the Canary Islands. Materials and methods A total of 878 individuals of P. incisus were collected at weekly intervals from discarded commercial catches taken between October 2000 and September 2001 off Gran Canaria (Canary Islands, central-east Atlantic, 27°57'24"N- 15°35'23"W). Each fish was measured to the nearest mm for total length (L,) and weighed to the near 0.1 g for total body weight (Wp. Sex was assessed visually and sagittal otoliths were removed, cleaned, and stored dry for later age determination. Age estimation was made by identify- ing and counting annuli following Wil- liams and Bedford (1974). An annulus was defined as a hyaline zone formed annually in the winter season when there is low growth and an opaque zone formed annually in the summer season when there is increased growth. The whole otoliths were placed in a black- ened bottom watch glass containing water and examined under a compound microscope (lOx) with reflected light. Counts of the growth bands were made 852 Fishery Bulletin 101(4) Figure 1 Saggital otoliths of P. incisus collected off the Canary Islands: (A) from a two-year-old (182 mm); (B) from a three-year-old (202 mm); and (C) from a five-year-old (265 mm) fish. Ri is the radius of the i"' band (distance from the centre of the otolith to the outer margin of the annulus), Re is the radius of the otolith at capture, and MI is the marginal increment measured. by two readers without knowledge of the size and sex of the specimens, or previous counts of the other reader. Counts were made for otoliths of each individual on two separate occasions, and only coincident readings were accepted. The same approach was used to determine the final number of bands in each specimen, and a consensus was reached between readers on the final counts. Reproducibility of the resultant age estimates was evaluated with the coefficient of variation (CV) (Chang, 1982) and the index of average percent error (lAPE) (Beamish and Fournier, 1981). To validate that rings were formed annually in the bastard grunt, we analyzed the monthly mean marginal increment (Hyndes and Potter, 1997). Marginal increment, estimated as the distance between the outer edge of the outermost annual ring and the periphery of each otolith, was measured (to the nearest 0.01 mm) with an ocular micrometer (Fig. 1). Measurements were always made along the longest axis of the otolith. The pattern expected in the marginal increment would be a minimal value at the start of the growth period, increasing with time until the measurement fell to a minimum again at the formation of the next period of growth (Pilling et al., 2000). The size of the marginal increment varies both with the age of the fish and the time of sampling during the year. Because older fish grow slower than younger fish, a smaller marginal increment is expected. For this reason, to assess the pos- sibility of false annulus formation among either younger or older bastard grunt, quantitative marginal increment analyses should be standardized for age. Therefore, we used age class to standardize our analyses. Owing to the wide range of ages encountered, however, there were insuf- ficient samples to fully accomplish this standardization. It was necessary to combine the ages in two or more age groups representing fast, moderate, and slow-growing in- dividuals (Pilling et al., 2000). Mean marginal increments were plotted against month of capture, and the minimum was used to indicate the month of annulus formation. Once the periodicity and timing of ring formation were verified, the age of each fish was determined from the num- ber of annuli, the assumed birthdate, and the sampling date. It was assumed that annulus formation began 1 January, corresponding to the peak of spawning in the species (Gregoire'). The difference between the date of capture and the birthdate was used to estimate a fractional age (Gordoa and Moll, 1997). This fraction was added to the number of annuli read in the otoliths to avoid any potential bias in growth estimates due to differences in sampling date (Gordoa and Moli, 1997). Length-at-age was described by the three-parameter specialized von Bertalanffy growth model (Ricker, 1973): L, =LJl-e -kH-l„) ), the four-parameter Schnute growth model (Schnute, 1981): Gregoire, M. 2001. Edad y crecimiento del roncador Pomadasys incisus (Bowdich, 1825) de Gran Canaria (Islas Canarias). Unpubl.data. ULPGC Socrates/Erasmus Research Report, 34 p. Departament of Biology, University of Las Palmas de Gran Canaria. 35017 Las Palmas de Gran Canaria. Spain. Pajuelo et al.: Age and growth of Pomadasys incisus 853 Z., = .y +(.v;' and the seasonalized von Bertalanffy growth model (Pitcher and Macdonald, 1973) L, = L \-e -Al/-/„ )-{ — sin(2;r(/-/, )) where Lc = the length-at-capture; and V = a constant derived from the relationship of total length to otolith radius (Francis, 1990). The von Bertalanffy growth curve was fitted to the backcal- culated length-at-age by means of Marquardt's algorithm for nonlinear least squares parameter estimation (Gay- anilo et al., 1996). where A, = the smallest age in the sample; A., = the largest age in the sample; Tg = the age at zero length; L, = the length-at-age; L„ = the predicted asymptotic length; yj = the estimated mean length of Aj-year-old fish; ^2 = the estimated mean length of A2-year-old fish; C = the amplitude of the fluctuation in seasonal growth; t^ = the point of the minumum growth {t^=WP+ 0.5); and k = the Brody growth constant (Schnute, 1981; Sparre and Venema, 1995). The models were fitted to data with the Marquardt's algo- rithm for nonlinear least squares parameter estimation (Gayanilo et al., 1996). A nonparametric, one-sample test was applied to test for residual randomness and a Bartlett's test was used to test for their homoscedasticity. Von Bertalanffy growth model parameters were also esti- mated for observed W, as a function of age by substituting weights in place of lengths in the growth equation and incorporating b derived by the weight-length regression (Sparre and Venema, 1995): W, = WJ]-e -l(i-r„l,* where W, = the weight at age; and W^= the predicted asymptotic weight. Backcalculated size of each fish at the time of forma- tion of each annulus was determined by a backcalculation formula consistent with the body proportional hypothesis (Campana, 1990; Francis, 1990). The measurements were the following: the radius of the ('*' band (Ri, 0.01 mm, dis- tance from the center of the otolith to the outer margin of the annulus) and the radius of the otolith at capture (Re, 0.01 mm, distance from the center of the otolith to the pe- riphery). These measurements were always made along the longest axis of the otolith ( Fig. 1 ). The relationship between the radius of the otolith at capture iRc) and the total length was estimated as a power function (nonlinear relationship). It is estimated by fitting the data by a regression of log(L,) on log(i?c) consistent with the body proportional hypothesis (BPH). The length of an individual when the i"' band was laid down {Li, mm) was calculated as Li = (Ri/Rc)'Lc, Results Of the 878 fish examined, 377 (42.9%) were males and 412 (46.9%) females. The remaining 89 (10.1%) individuals were immature and could not be identified macroscopically. Fish varied in size from 103 to 304 mm L,, and weighed between 8.7 and 137.1 g W,. Males varied from 143 to 298 mm L,, and total mass was from 9.3 to 114.7 g W^. Female total length varied between 134 and 304 mm and total mass was from 13.2 to 137.1 g. Immature fish varied from 103 to 186 mm L, and from 8.7 to 29.6 g W,. No significant differences were found between males and females in mean size (Student's i-test, <=1.03<-f'o 05 2 1560^' therefore, the specialized von Bertalanffy growth model 1 - 0.8" -- 0.6" n r~ 0.4" R -p 1 rn T I S 0.2' i r 1 p-| -L ± ? A- ^ E 0 E 08" c £ 0.6- o — o 1 0,4- s n:' -- -L -r- -[- -r- s I s^ 0.8- 0.6- 0.4 T -r 15 T -r- 0.2- ^r*- 1 1 1 1 1 1 1 1 10 II 12 1 2 3 4 5 6 7 8 9 I 2000 I 2001 I Month Figure 2 Mean monthly marginal increment from otoliths with one and two, three, and more than three annuli, representing fast, mod- erate, and slow-growing individuals of P. incisus off the Canary Islands. Standard errors are identified by the bars. was chosen because it has fewer parameters making it sta- tistically more robust, its parameters are commonly used in mortality estimates and per recruit modelling, and because it allows for comparison between growth studies conducted on other species (Booth, 1997). No significant differences were found between mean lengths-at-age of males and fe- males with a Student's t-test {t=0.52 ;gQggjQ = 1.65). Backcalculatcd size at time of annulus formation was used to provide length-at-age data unbiased by differences in sampling date and to estimate the von Bertalanffy equation (Table 3). Backcalculated lengths were similar to those pre- dicted by the growth models. Growth parameters estimated from the backcalculated sizes at age were L„=315.23 mm; A=0.217/year; and tg=-1.733 year. The data was pooled as a single growth model because no significant differences in the growth parameters were found between males and females (HoteUing's T'-'-test, 7^=48.3>ro2Q.o5,3,784=7-89). Discussion Age estimation in fishes is complicated by the phe- nomenon of "stacking" of growth zones towards the otolith margin, particularly in older fish (Buxton and Clarke, 1991). In many cases age determination is difficult because whole otoliths are so thick that light does not pass through (Buxton and Clarke 1991); how- ever, in the bastard grunt off the Canary Islands the translucency of the otoliths allows aging with relative ease. The values of the lAPE and the CV suggested that the precision levels obtained are according to the reference point values indicated by Campana (2001). The oldest age estimate obtained in the present study was seven years and the phenomenon of stacking was not evident. The otoliths of the bastard grunt have a ring pat- tern common to teleost fishes. Marginal increment analysis demonstrated that one annulus, consisting of one opaque zone and one hyaline zone, is formed an- nually. These rings are believed to be deposited during periods of fast and slow growth, respectively (Williams and Bedford, 1974). Seasonal growth cycles might be related to physiological changes produced by the influ- ence of temperature, feeding regime, and reproductive cycle (Morales-Nin and Ralston, 1990). The seasonal- ized von Bertalanffy growth model reveals the reduc- Pajuelo et al.: Age and growth of Pomodasys incisus 855 Table 1 Sample size per age group off the Canary Islands, n intervals. (age-length key) and percentage (in parentheses) within each age group for all fish of P s the number fish by age class and SD is the standar deviation. Total length classes are ncisus collected given in 10-mm Size (mm) Age groups (year) 0 I II III IV V VI VII 100 2(40.0) 3(60.0) 110 1(8.1) 10(90.9) 120 1(5.6) 17(94.4) 130 19(100.0) 140 15(71.4) 6(28.6) 150 4(17.4) 19(82.6) 160 33(100.0) 170 47(95.9) 2(4.1) 180 69(70.4) 29(29.6) 190 52 (46.6) 61(53.4) 200 25 (20.5) 96(78.6) 1 (0.9) 210 14(12.8) 89(81.6) 6(5.5) 220 5 (5.9) 57(67.0) 23(27.1) 230 14(27.0) 38(73.0) 240 2(10.0) 17(85.0) 1(5) 250 1 (6.6) 9(60.0) 4 (26.8) 1(6.6) 260 2(20.0) 6(60.0) 2(20.0) 270 1(25.0) 1(25.0) 2 (50.0) 280 1(33.3) 1(33.3) 1(33.3) 290 1 (100) 300 1(100.0) Mean 107 125 176 212 236 261 275 295 n 4 68 270 351 97 13 7 2 SD 5.77 12.64 17.28 18.06 12.36 11.43 11.40 14.14 Table 2 Parameters estimates, standard errors, and 95% cofidence intervals for the specialized von Bertalanffy, Sehnute, and seasonalized von Bertalanffy growth models for all P. incisus collected off the Canary Islands. All models were pooled without the age-0 class. Parameter 95% confidence intervals Estimate Standard error Lower Upper Specialized von Bertalanffy growth model (r^=0.9l) L_ (mm) 309.58 8.06 294.07 325.08 k (/year) 0.220 0.031 0.157 0.283 ;,i(year) -1.865 0.055 -2.007 -1.723 Sehnute growth model (r^=0.87) V, (mm) 126.66 1.51 124.54 128.78 v., (mm) 293.50 9.50 286.68 300.32 A -0.426 0.077 -0.578 -0.274 B 5.963 0.664 4.659 7.267 Seasonalized von Bertalanffy growth model (r'-=0.91) L_ (mm) 309.93 7.68 295.21 324.65 if (/year) 0.218 0.032 0.153 0.282 ;„(year) -1.896 0.049 -2.067 -1.725 C 0.555 0.212 0.138 0.971 /, 0.652 0.061 0.532 0.773 856 Fishery Bulletin 101(4) 350 - 300 ^ jm^^^^^jm • • ^-^^^ ^J©"*"^ * 251) - ? ^ 200 c OJ — 150 TO ...s^^"^' |5 ^Btt*"* • ^'^^ Bertalanffy curve ion ^ ZjSf H Seasonalized von Bertalanffy curve -•-Schnule curve -♦Observed data 50 - 0 . : 1 , , T 0 12 3 4 5 6 7 8 Age (year) Figure 3 Von Bertalanffy specialized and seasonalized growth curves and the Schnute growth curve for all individuals of P incisus collected off the Canary Islands. 4(H) 350 - .. 300 ,• vr\ ' 3 250 g 2110 m ,o *- 150 l(K) - * *** JjaUKBlwir*^ " H>788(l^-"«"*""')='" V^MfWrnnZ^^^^*^ 50 - • • mW)^^^^ •' ''="*» 1 m-'^^iVt ' * ^"SUS^ " - 1 1 1 I 0 2 4 6 8 Age (year) Figure 4 Von Bertalanffy growth curve derived from observed weight at age for all P. incisiis collected off the Canary Islands. tion of somatic growth and the formation of the hyahne zone during the winter months. The high correlation found between L, and otolith radius indicates that otoliths are a useful structure for estimating the age and for indicating the past growth history of bastard grunt. The coefficients of determination for each fitted curve show that the three models explain more than 88% of the growth pattern. The use of the von Bertalanffy model to describe growth has been criticized for several reasons (Booth, 1997). These include the use of parameters that Age and growth of Pomadasys incisus 857 320 280 - « ^'^ • • ^^X* 240 •liilK^*^^** mm) iliililW^'^ £ .,.^^^'' al leng O 1 "*() F J^'V* * * L,=49.93 R"*^' **tl* • n=812 80 - • • r=0.912 40 - 0 1 1 1 1 1 ' 12 3 4 5 6 7 8 Otolith radius (mm) Figure 5 The relation of otolith length to total length for P. incisus collected off the Canary Islands. Table 3 Backcalculated 1 ength-at-age for all P. incisus collected off the Canary Islands. Annu us number Age Number (year) offish I 11 Ill rv V VI VII 1 68 119 2 270 123 179 3 351 127 172 219 4 97 132 168 212 235 5 13 129 178 205 245 26 6 7 122 170 221 233 257 275 7 2 128 174 214 239 260 279 293 Mean 126 173 214 238 259 277 293 Number of backcalculated lengths at age 808 740 470 119 22 9 2 Annual growth ncrement (mm) 126 47 41 24 21 18 16 Annual growth ncrement (%) 43.0 16.1 13.9 8.2 7.2 6.1 5.4 have little biological meaning (Schnute, 1981) and the absence of parameters that take into account seasonal changes in growth rate (Pauly, 1980; Moreau, 1987). Nev- ertheless, the von Bertalanffy growth model has been used extensively to describe the growth of grunts. The growth model provides a simple description of growth which can be compared between species and species groups (Booth, 1997). The special and the seasonal forms of the von Ber- talanffy growth model were chosen for the present study because they contain fewer parameters than the Schnute growth model. Backcalculated lengths-at-age are in close agreement with the length estimated from otoliths readings. The results obtained with the backcalculation method are very satisfactory because they show the consistency in the in- terpretation of the sequence of growth increments of the bastard grunt off the Canary Islands and reduce the effect of size-selective sampling bias on the length estimates for youngest fish in the sample (Campana, 2001). The growth parameters obtained are reasonable because the theoretical maximal length value is higher than the size of the largest fish sampled and the growth coefficient 858 Fishery Bulletin 101(4) 1 - 0.8 - / ©^^ \ „~ P an>\reus~\^ sfficlent (year o a- 1 v.y \^*«\ 1 0.4^ o 0.2 - P sirtalus P \/-\ \ P opfnitluris ' ~---___^ jubenile ^— N. ^\ \ incistis ^^^/^^ \s, y^ ^ P argmleus ^--.^^^^^ /\# \ P fiaculalus p-SigTOmAieoo/mii 0 - \. 1 1 1 ' ■ ^^ 0 200 400 600 800 1000 1200 Asymptotic length (mm) Figure 6 The relation of asymptotic length to the growth coefficient for different species of the Pomadasys genus. value indicates a relatively fast attainment of maximum size, characteristic of the moderately short life cycle for this species. However, the estimations of ^g tend to be negative and different from zero for values affected by the small sample size of smaller fish. These estimations suggest that the von Bertalanffy growth model does not accurately de- scribe growth in the early stages. Pomadasys incisus grows quickly in its first year, attaining approximately 60% of its maximum length. After the first year, the annual growth rate drops rapidly. This change in growth rate is attribut- able to the utilization of available energy for reproduction instead of somatic growth; in the study area the maturation process begins in the second year of life (GregoireM. Two different patterns of growth rate in relation to as- ymptotic length are observed for Pomadasys species (Fig. 6). The pattern off! incisus is similar to that observed for P. striatus, P. jubeline, P. kaakan, P. maculates, and P. commersonnii — species characterized by a high or moder- ate asymptotic length and low or moderate growth coef- ficient (Latif and Shenouda, 1972; Wallace and Schleyer, 1979; Edwards at al., 1985; Iqbal, 1989; Al-Husaini et al., 2001; Pauly''^). However, it differs substantially from that observed for P. argyreus, a species with a very low asymp- totic length (<151 mm) and a very high growth coefficient (0.62-0.83/year), and for P. opercularis and P. argenteus, species characterized by a high asymptotic length and a high growth coefficient (550-741 mm; 0.28-0.52/years) (Deshmukh, 1973; Nzioka, 1982; Brothers and Mathews, 1987; Majid and Imad, 1991; Ingles and Pauly^). Results of models used in fisheries management, e.g. analytical yield-per-recruit models (Beverton and Holt, 1957), are sensitive to uncertainty in the estimates of input parameters such as the von Bertalanffy growth parameters. Several estimations of growth in Pomadasys species have been derived through length-based methods, which for slow growing species are uncertain. The growth parameters from this study are the first otoliths-based es- timates of growth for P. incisus. Similar estimates obtained from different growth models and methods suggest that the current estimation could be considered a good estimation of the growth pattern for the species and adequate for use as an input parameter in models for the management of the species. Acknowledgments The authors are grateful to the three anonymous review- ers for their constructive, critical, and useful comments on the manuscript. ^ Pauly, D. 1978 A preliminary compilation of fish lengtli growth parameters. Ber Inst. Meereskunden, Universitat an der Kiel 55, 200 p. Institut fur Meereskunde, Dusternbrooker Weg 20, 24105 Kiel, Germany. '' Ingles, J., and D. Pauly. 1984. An atlas of the growth, mortal- ity and recruitment of Philippine fishes, ICLARM Technical Repport 13, 127 p. World Fish Center (ICLARM) Jalan Batu Maung, Batu Maung, 1 1960 Bayan Lepas, Penang, Malaysia. Paiuelo et al,: Age and growth of Pomadasys Incisus 859 Literature cited Al-Husaini, M., S. Al-Ayoub, and J. Dashti. 200 1 . Age validation of nagroor, Pomadasys kaakan ( Cuvier, 1830) (family: Haemulidae) in Kuwaiti waters. Fish. Res. 53:71-81. Bauchot, M. L., and J. C. Hureau. 1990. Haemulidae. //; Check-list of the fishes of the eastern tropical Atlantic (J. C. Quero, J. C. Hureau, C. Karrer, A. Post, and L. Saldanha, eds. ), p. 786-787. UNESCO, Paris, France. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish Aquat. Sci. 38:982-983. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish- eries Investigations Series II, XIX, 533 p. Her Majesty's Stationery Office, London, UK. Booth, A. J. 1997. On the life history of the lesser gurnard (Scorpaeni- formes: Triglidae) inhabiting the Agulhas Bank, South Africa. J. Fish. Biol. 51:1155-1173. Brothers, E. B., and C. P. Mathews. 1987. Application of otolith microstructural studies to age determinations of some commercially valuable fish of the Arabia Gulf Kuwait Bull. Mar Sci. 9:127-157. Buxton, C. D., and J. R. Clarke. 1991. The biology of the white musselcracker Sparadon durbanensis (Pisces: Sparidae) on the Eastern Cape Coast, South Africa. S. Afr J. Mar Sci. 10:285-296. Campana, S. E. 1990. How reliable are growth back-calculations based on otoliths? Can. J. Fish. Aquat. Sci. 47:2219-2227. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Chang, W. B. 1982. A statistical method for evaluating the reproduc- ibility of age determinations. Can. J. Fish. Aquat. Sci. 39: 1208-1210. Deshmukh,V. M. 1973. Fishery and biology of Pomadasys hasta (Bloch). Indian J. Fish. 20:497-522. Edwards, R. R. C, A. Bakhader, and S. Shaher 1985. Growth, mortality, age composition and fishery yields offish from the Gulf of Aden. J. Fish. Biol. 27:13-21. Francis, R. I. C. C. 1990. Back-calculation of fish length: a critical review. J. Fish Biol. 36:883-902. Gayanilo, F. C, R Sparre, and D. Pauly. 1996. FAO-ICLARM stock assessment tools (FiSAT) user's guide, 126 p. FAO (Food Agric. Organ. U. N.) Comp. Inf Ser (Fish.)8. Gordoa, A., and B. Moll. 1997. Age and growth of the sparids Diplodus vulgaris, D. sargus and D. annularis in adult populations and the dif- ferences in their juvenile growth patterns in the north-west Mediterranean Sea. Fish. Res. 33:123-129. Hyndes, G. A, and I. C. Potter 1997. Age, growth and reproduction ofSillago schomhurgkii in south-western Australin, nearshore waters and compari- sions of life history styles of a suite ofSillago species. En- viron. Biol. Fish. 49 435-447. Iqbal, M. 1989. A note on the population dynamics of Pomadasys kaakan (Haemulidae) from Pakistan. Fishbyte 7:4-5. Latif, A., and S. Shenouda. 1972. Biological studies on Rhonciscus striatus (family Pomadasidae) from the Gulf of Suez. Bull. Inst. Ocean. Fish. 2:103-134. Majid, A., and A. Imad 1991. Growth of Pomadasys kaakan (Haemulidae) off the coast of Pakistan. Fishbyte 9:30-33. Morales-Nin, B., and S. Ralston. 1990. Age and growth of Lutjanus kasmira (Forskal) [sic] in Hawaiian waters. J. Fish Biol. 36:191-203. Moreau, J. 1987. Mathematical and biological expression of growth in fishes: recent trends and further developments. In Age and growth offish (R. C. Summerfelt and G. E. Hall, eds.), p. 81-113. Iowa State Univ. Press, Ames, lA. Nzioka, R. M. 1982. Biology of the small spotted grunt Pomadasys opercu- laris (Playfair 1866) (Pisces: Pomadasyidae) around Malindi in Kenya. Kenya J. Sci. Tech. 3:69-81. Pilling, G. M., R. S. Millner, M. W. Easey, C. C. Mees, S. Rathachasen, and R. Azemia. 2000. Validation of annual growth increments in the otoliths of the lethnnid Lethrinus mahsena and the lutjanid Apnon virescens from sites in the tropical Indian Ocean, with notes on the nature of growth increments in Pristipomoides filamentosus. Fish. Bull. 98:600-611. Pauly, D. 1980. On the interrelationships between natural mortality, growth parameters, and mean enviromental temperature in 175 fish stocks. J. Cons. Int. Explor Mer 39:175-195. Pitcher, T J., and R D. M. Macdonald. 1973. Two models for seasonal growth in fishes. J. Appl. Ecol. 10:597-606. Ricker, W. E. 1973. Linear regressions in fishery research. J. Fish. Res. Board Can. 30:409-434. Schnute, J. 1981. A versalite growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. Sparre, P., and S. C. Venema. 1995. Introduccion a la evaluacion de recursos pesqueros tropicales, parte 1, manual, 420 p. FAO (Food Agric. Organ. U. N.) Doc. Tec. Pesca 306(1). Wallace, J. H., and M. H. Schleyer 1979. Age determination in two important species of South Africa angling fishes, the knob {Argyrosomus hololepidotus Lacep) and the spotted grunter (Pomadasys commersonnii Lacep). Trans. R. Soc. S. Afr. 44:15-26. Williams, T., and B. C. Bedford. 1974. The use of otoliths for age determination. In The ageing of fish (T. B. Bagenal, ed.), p. 114-123. Unwin Brothers, Old Woking, Surrey, UK. 860 Abstract— Management of West Coast groundfish resources by the Pacific Fishery Management Council involves Federal government and academic scientists conducting stock assess- ments, generally using the stock syn- thesis framework, applying the 40-10 rule to determine harvest guidelines for resources that are not overfished and conducting rebuilding analyses to determine harvest guidelines for resources that have been designated as overfished. However, this manage- ment system has not been evaluated in terms of its ability to satisfy the National Standard 1 goals of the Sus- tainable Fisheries Act. A Monte Carlo simulation framework is therefore outlined that can be used to make such evaluations. Based on simulations tai- lored to a situation similar to that of managing the widow rockfish {Sebastes entomelas) resource, it is shown that catches during recovery and thereafter are likely to be highly variable (up to ±30'?f from one year to the next). Such variability is far greater than has been presented to the decision makers to date. Reductions in interannual vari- ability in catches through additional data collection are, however, unlikely. Rather, improved performance will probably arise from better methods for predicting future recruitment. Rebuild- ing analyses include quantities such as the year to which the desired probabil- ity of recovery applies. The estimates of such quantities are, however, very poorly determined. Evaluating the efficacy of managing West Coast groundfish resources through simulations Andre E. Punt School of Aquatic and Fishery Sciences University of Washington 1122 NE Boat Street Seattle, Washington 98195-5020 E-mail address aepunt@u. Washington edu Manuscript approved for publication 24 April 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:860-873 (2003). National Standard 1 of the Sustainable Fisheries Act (SFA) of 1996 states that "Conservation and management mea- sures shall prevent overfishing while achieving, on a continuing basis, the optimum yield from each fishery for the United States industry." The need to satisfy this National Standard has led inter alia to the requirement for the eight Regional Fishery Management Councils to develop control rules that are used to assess whether overfishing is occurring^ or a stock is in an over- fished state (e.g. Restrepo and Powers, 1999). In addition, the SFA specifies that a rebuilding plan has to be devel- oped for any fish stocks that are des- ignated as overfished. This plan needs to include the time period by which the stock will be rebuilt to fi^gY (the aver- age biomass associated with maximum sustainable yield, MSY), and the strat- egy by which the stock is to be rebuilt. The Pacific Fishery Management Council (PFMC) has adopted the "40- 10" rule to manage groundfish stocks that are not designated as being over- fished. This rule determines the harvest guideline for each groundfish stock by computing the catch corresponding to an i^MSY proxy (F^q,,/ for flatfish, FgQ.^ for rockfish in the Sebastes complex, and Fjjr; for other species ) and reducing it if the spawning output is estimated to be less than 40% of the estimated Bq. This reduction in catch is linear with spawn- ing output, being 0 at 0.4Sq and lOO'J at O.IBq. For stocks that are designated as being in an overfished state (defined for West Coast groundfish as being that the spawning output is less than 0.25Bq) a rebuilding plan is developed. ■* The main features of the technical as- pects of a rebuilding plan (referred to as a rebuilding analysis) identified by the Scientific and Statistical Committee of the PFMC are outlined in Appendix 1. In brief, the rebuilding analysis used by the PFMC involves projecting the best estimates of the current age-structure of the overfished population forward under a range of alternative fishing mortality rates and selecting the fish- ing mortality rate that has a Council- selected probability that the population recovers to the proxy for Bygv of 0.4Sg within a time frame consistent with the specifications of the SFA. Detailed stock assessments are avail- able for only a small subset of the 81 species included in the PFMC Ground- fish Management Plan. Of these species, nine (bocaccio [Sebastes paucispinis], canary rockfish [Sebastes pinniger[, cowcod [Sebastes levis], darkblotched rockfish [Sebastes crameri], lingcod [Ophiodon elongates], Pacific ocean perch [Sebastes alutus]. Pacific whiting [Merluccius productus], widow rockfish [Sebastes entomelas], and yelloweye rockfish [Sebastes ruberhmiis]) have been designated overfished and rebuild- ing plans have been or are being devel- oped for them. The direct consequences ' In the present study, and consistent with usage by the Pacific Fishery Management Council, "overfishing" means that the level of fishing mortality exceeds that associ- ated with MSY and "being in an overfished state" means that the current spawning output is less than 25'7c of the pre-exploi- tation equilibrium spawning output. Sq (spawning output is the product of egg production-at-age and numbers-at-age). 2 F^,^ is the fishing mortality rate at which the spawning output-per-recruit is reduced to x% of its unfished level. 3 One implication of this is that the 40-10 rule is not actually used if the stock is assessed to be below 0.25B„. Punt: Managing West Coast groundfish resources through simulations 861 for industry of the implementation of a rebuilding plan can be substantial (e.g. a reduction in the catch of canary rockfish from 883 metric tons (t) in 1999 to only 90 t in 2001), although there are also indirect consequences in the form of reductions in the harvest of nonoverfished species to prevent overharvesting of overfished species through technical interactions. The performance of the method commonly used for as- sessments of West Coast species has been evaluated to some extent (e.g. Sampson and Yin, 1998; lanelli, 2002). However, the performance of this assessment method in combination with the rules used to determine harvest guidelines has not been evaluated. Management procedures'* are combinations of stock as- sessment methods and catch control laws that have been evaluated by means of Monte Carlo simulation to assess the extent to which they are able to satisfy the manage- ment objectives for a fishery. Evaluation of management procedures by means of Monte Carlo simulation has been argued to be essential because "if a management procedure is unable to perform adequately in the ideal world repre- sented on a computer, what basis is there to assume that it will perform adequately in the real world?" (Sainsbury^). One caveat to this argument is that it is only possible to evaluate a management procedure if it is fully specified and if it will be followed for several years in reality. Management procedures have been adopted by the In- ternational Whaling Commission for managing commer- cial and aboriginal whaling (e.g. IWC, 1992, 2001) and by southern African nations for managing a variety of pelagic and demersal resources (Butterworth and Bergh, 1993; Co- chrane et al., 1998; Geromont et al., 1999). Management procedures are under consideration in Australia (Punt et al., 2001) and New Zealand (Starr et al., 1997). If it can be assumed that the same rules will be applied to modify rebuilding plans each time new information on abundance and year-class strength becomes available, it is possible to consider the combination of the assessment method, the default 40-10 rule, and rebuilding plans as a "manage- ment procedure" and evaluate it by means of Monte Carlo simulation. This study therefore involves determining from past practice the "management procedure" being applied by the PFMC. However, this "management procedure" has not been formally adopted in any way and the approach to managing West Coast groundfish could change in time. This paper first outlines a simulation framework (a management procedure evaluation, MPE, framework) within which the expected performance of the approach used by the PFMC to determine harvest guidelines can be evaluated. It then evaluates variants of this approach for scenarios similar to that of managing the fishery for widow rockfish. '' Also referred to as "harvest strategies" ( Punt et al ., 200 1 ), "man- agement decision rules" (Starr et al., 1997 ), "fisheries control sys- tems" ( Hilborn, 1979 ), and "operational management procedures" (Barnes, 1999). ^ Sainsbury, K. G. 2001. Personal commun. CSIRO Marine Research, Castray Esplanade, Hobart, TAS 7000, Australia. Materials and methods The steps in evaluating management procedures are as follows: 1 Identification of the management objectives and rep- resentation of these by using a set of quantitative performance statistics. 2 Identification of a range of alternative management procedures. 3 Development and parameterization of a set of alterna- tive structural models (called operating models) of the system. 4 Simulation of the future use of each management procedure to manage the system (as represented by each operating model). For each year of the projection period, the simulations involve the following steps; a Generation of the data available for assessment purposes. b Application of a method of stock assessment to the generated data to determine key assessment- related quantities (e.g. current age-structure, spawning output in relation to target and limit levels, historical trends in recruitment) and any inputs to the catch control law. c Application of the catch control law element of the management procedure to determine a harvest guideline. d Determination of the biological implications of this harvest guideline by setting the catch for the "true" population represented in the operat- ing model based on it. The step can potentially include "implementation uncertainty" (Rosenberg and Brault, 1993). The harvest guideline is not updated every year in the simulations described in this article, but rather every third year (co-incident with the results from each new survey) and thus reflects the intended frequency with which assessments for West Coast groundfish species are conducted. Each simulation trial (i.e. each combination of an operating model variant and candidate management procedure) involves 100 simulations of an 80-year manage- ment period. The four steps listed above are discussed in detail below. Note that for the application considered in this paper then, there are three "models": 1) the operating model that represents "reality" for the simulations, 2) an assessment model (a stock synthesis-like approach), and 3) a model to calculate the harvest guidelines. The data available to the last two models are generated from the first model. The operating model The operating model has been taken to be virtually iden- tical to that on which the population assessments and rebuilding analysis calculations are based (Appendix 1), with two exceptions: 1) the approach used to generate recruitment and 2) the allowance for variability over time in commercial selectivity. Commercial selectivity is given 862 Fishery Bulletin 101(4) o _ 8 O -1 B CM \ Selectivity 3 •" CO > 3 O O o ■ in \ AA elative 1.0 ^ \ /\^ o ^/-'"^^ '^ S. s J J 0 o ' C ) 5 10 15 20 0 10 20 30 40 Age (yr) Year Figure 1 Biological parameters (A) and catch history (B) for widow rockfish iSebastes entomelas). by the following double-logistic equation: 5,, „ = S[ ^, / max 5' ,,. I 1 (1) 5,' =- \ + e ■6f(u-iiy^+y, I l + e -Slti^-a) where S^^ = the selectivity on fish of age a during yeary; a^Q, oIq, 5j, ^2 = the parameters of the double-logistic equation; 7 = the deviation from the average selectiv- ity pattern in yeary: Yy = Ps rv-l + ^v el ~ N(0;a;), pg = the interannual correlation in the de- viation from average selectivity; and Gg = a measure of the standard deviation of the interannual deviations from aver- age selectivity. Recruitment is assumed to be governed by a Beverton-Holt stock-recruitment relationship: ' 4/i + (5/i-l)(B, /B„-l) E^~N{0-ol). (2) where Rq = the "virgin recruitment" (the number of zero- year-olds at the pre-exploitation equilibrium level); By - the spawning output at the start of year y; h = the"steepness"of the stock-recruitment rela- tionship (the fraction of virgin recruitment expected at 0.2B„); and a^ - the standard deviation of the logarithms of the random fluctuations in recruitment about its expected value. The biological parameters of the operating model are set to those for widow rockfish (Fig. lA), and the catches for Table 1 The baseline parameters of the operating values used in the tests of sensitivity. N/A model and the = not available. Parameter Baseline value Sensitivity values Ps 0.707 N/A ^s 0.4 N/A h 0.4 0.25; 0.7 cffl 0.6 0.4; 1 M Spawning output in year 41 0.15/yr 0.2S„ N/A 0.1Bo;0.4Bo the 40 years prior to the year in which the management procedure is first applied (referred to as "projection year 1") are set to the actual catches for widow rockfish (Fig. IB). The baseline values for the parameters /;, (T^, pg, and Og (Table 1) are educated guesses. The baseline choice for steepness, /;, is lower than the posterior mean for this quantity (0.65) obtained by Dorn (2002) because, increas- ingly. West Coast rockfish are being found to be less pro- ductive than initially anticipated (e.g. lanelli, 2002). The value assumed for the extent of variation in recruitment, a^, although based on the collection of estimates of this parameter by Beddington and Cooke (1983), is neverthe- less also largely an educated guess. Sensitivity to the values for both h and a^j is explored. The biomass at the start of year 1 is assumed equal to Bq, which is defined as the mean of the distribution for the unfished biomass which would arise given variability in recruitment about its expected value. However, this specification has little impact on the results. For example, the alternative that is defined to be the median of the distribution for the unfished biomass would only change B„ by about 5%. The value for Bg for each simulation is selected so that the spawning output at start of year 41 (projection year 1) equals a prespecified fraction of Bq (baseline fraction Punt: Managing West Coast groundfish resources through simulations 863 Table 2 The parameters on which the generation of future data is based n'' is the sample size for the multinom al distribution. Data source First year collected Frequency Precision Catch rates 14 Every year a'= 0.4 Fishery age-composition 21 Every year ^"=200 Survey indices 13 Every third year CT''=0.5 Survey age-composition 13 Every third year n'=200 0.2 — i.e. just below the level that defines an overfished stock). Sensitivity to alternative values for the ratio of the spawning output at the start of year 41 to Bq is explored (Table 1). Generating future data The data available for assessment purposes are survey indices of relative abundance, age-composition data from surveys, catch-rate-based indices of relative abundance, and age-composition data from the commercial catches. Table 2 lists the baseline specifications regarding the fre- quency at which the various data sources are collected and the parameters that determine the sampling variability associated with each data source. The survey and catch-rate indices are generated by using the equations B':">" = B'/'-"'''''\ /= By' -'"■'■ '\ ef ~/v(0;(CT'r); (3a) f;:~/v(0;(cr')-); (3b) where £*■"''''' = the survey index for year j; B^ = the survey selected-biomass during yeary: fi: = £H-„5;,/V„,„.-''""; (4a) w^ = the mass of an animal of age a; Sj, = the selectivity of the survey gear on animals of age a (assumed to be governed by a logistic function and to be independent of time); ^\a ~ ^^^ number of animals of age a at the start of year .v; Z,,^ = the total mortality on animals of age a during year y\ cTj = the standard deviation of the random fluctuations in survey catchability; "max - ^^^^ oldest age considered in the operating model; /^, = the catch-rate index for year y; B^ = is the exploitable biomass during yeary; fi:=X»'.%^^..,(" -<-"'■■'); (4b) CT'' = the standard deviation of the random fluctuations in fishery catchability. Note that Equations 3a and 3b assume that the survey and fishery catchability coefficients are unity. This assumption can be made without loss of generality because the stock assessment method is not provided with this information and instead estimates these catchability coefficients. Note also that the key difference between the survey index and the catch-rate index is that selectivity for the latter changes over time (see Eq. 1), whereas selectivity for the former is time-invariant. The age-composition data are generated by selecting a sample multinomially from the age-composition of the survey catch and of the fishery catch (see Eqs. 5a and 5b for the relative survey and fishery catches-at-age): Sln^. •7^«v.a(l-^ )• (5a) (5b) F^ = the fully selected fishing mortality during year y; and The PFMC management procedure The "PFMC management procedure" (see Fig. 2 for an overview) involves first conducting a stock assessment by fitting an age-structured population dynamics model to the available data by maximizing a likelihood func- tion. This approach mimics the common use of the stock synthesis framework (Methot, 2000) when conducting assessments of West Coast groundfish resources. The likelihood function is determined by assuming that the age-composition data are multinomially distributed (in the simulations with effective sample sizes given by the actual effective sample sizes) and by assuming that the survey and catch-rate series are log-normally distributed about the appropriate model quantities. The estimable parameters of the model are the annual recruitments, the annual fishing mortalities, the catchability coef- ficients, and the parameters that determine selectivity (the survey and fishery selectivity are [correctly] assumed to be governed by logistic and double-logistic equations). The values for the remaining parameters (weight-at-age, fecundity-at-age, and natural mortality) are assumed to be known without error. The key outputs from the assess- ment are time-series of recruitments and spawning out- 864 Fishery Bulletin 101(4) i Conduct a Stock Assessment i Select an approach for generating future recruitment (recruits or recruits / spawner) No Currently under a Yes Rebuilding Plan'' Is stock assessed to be currently overfished'' Apply the 40-10 rule Yes ♦ Declare stock overfished Has stock recovered to above the t.arget level'' Apply the rebuilding software (Do not update the target rebuild year unless the stock has just been declared overfished) ^ Apply the constraints on ^ i the extent of inter-annual variation in harvest guidelines Constrain the harvest guideline to fall within the maximum and minimum limits Figure 2 Flowchart of the Pacific Fishery Management Council management procedure. puts, and the age structure at the start of the last year of the assessment. An estimate of the pre-exploitation equilibrium spawn- ing output (i.e. Bq) is obtained by multiplying the average recruitment for the first ten years of the assessment period by the spa-wning output-per-recruit in the absence of fish- ing. This approach to estimating Sg has been used for sever- al rebuilding analyses for West Coast groundfish species. If the estimate of the current spawning output exceeds 0.4Sg or if it exceeds 0.25 fig and the resource is not currently under rebuilding (i.e. has not yet been declared to be in an overfished state), a raw harvest guideline is computed us- ing the 40-10 rule. On the other hand, if the estimate of the current spawning output is less than 0.25 fig or the stock is currently under a rebuilding plan and the spawning output has not yet recovered to 0.4 fig, the raw harvest guideline is based on the application of the rebuilding analysis (see Appendix 1 for further details). It is necessary to know the maximum possible rebuilding period, T^^^, when using a rebuilding analysis to calculate a harvest guideline. If the stock is declared overfished in the present year, T^^^ is computed as described in Appen- dix 1. On the other hand, if the stock is currently under a rebuilding plan, T^^^ is taken to be the value computed when the stock was first declared overfished. Therefore, the implementations of the rebuilding plans considered in this paper are based on the assumption that the 7",,,,^,^ and the probability of recovery by T^^g^ are set when the first rebuilding analysis is conducted and not changed thereafter. The probability of recovery by T„^^^ is taken to be 0.6 in this paper because this is the probability on which management of widow rockfish is currently based. This probability ranges between 0.55 and 0.92 among the seven overfished groundfish resources for which it has been selected by the PFMC. Calculation of a harvest guideline using the 40-10 rule and application of the rebuilding analysis requires the ability to generate future recruitment. For the purposes of the present study (and consistent with current practice), future recruitment is either generated from the estimates of recruitment from the assessment or by multiplying the spawning output by a generated value for the recruits-per- spawning output ratio. The pool of recruitment to recruits- per-spawning output is taken to be those for the last 23 years of the assessment period less those for the last three years. The last three years are excluded because of their known poor precision. The approach used to generate re- cruitment therefore leads to the set of recruitments used to conduct projections changing with time. Allowing the set of recruitments to change with time is needed to avoid an inconsistency between the recruitments used for projections and the recruitments on which the estimate of Bg is based. Allowance is made for the raw harvest guideline to be constrained so as not to change by more than a prespecified percentage from that for the previous year and not to fall outside of specified limits, although this option is not part of the baseline simulations. One aspect of the actual management process that is ignored in the simulation of the PFMC management pro- cedure is the time-lag between the collection of data and their use in assessments (for example, catch-at-age infor- mation from surveys conducted in one year would usually not be available for use in the assessments conducted in the following year) and that between assessments Punt: Managing West Coast groundfish resources through simulations 865 Table 3 The performance statistics used in the present study. For consistency with the definition of recovery used by the Pacific Fishery Management Council, "recovery by year x" is defined as the spawning output being larger than OABq at or before year x. Some of the statistics are based on the "actual" (i.e. operating model) spawning output and others are based on the "assessed" (i.e. assess- ment model) spawning output. Abbreviation rec °decl 5%D/50%D AAV Description The fraction of the simulations in which the stock is assessed to be overfished at the start of the first projection year that actually recover by the maximum possible recovery year determined from the rebuilding analysis conducted in projection year 1. The median year in which the actual spawning output first reaches QABg. The proportion of simulations in which the spawning output is assessed to be below 0.25Bq (i.e. overfished) at the start of projection year 1. The lower 5th and median of the distribution of the actual spawning output in projection years 20 and 60 expressed in relation to the actual pre-exploitation spawning output, Bq. Average annual absolute change in catch evaluated after 20 and 60 years, i.e. where C^ is the catch during yeary. Average annual catch over projection years 1-20 and 1-60. The fraction of simulations in which actual spawning output reached 0.4Bo sometime between projections years 1 and 20 and between projection years 1 and 60 (but may have dropped below 0.4Sq again). being conducted and their being used for management purposes. The performance statistics A variety of performance statistics are considered (Table 3). These consider both the performance of the management procedure in terms of the behavior of the rules used for management (statistics F^^^, Y^^^, and P^g^^) and of satis- fying the goals established by the SFA in relation to the status of the population and the fishery (statistics 5%D, 50%D, C, AAV, and P^J. The choice of years 20 and 60 in the definitions of the latter five statistics is meant to capture "short"-term and medium-term considerations. For instance, recovery should have occurred by year 60 in most cases and the population should be well above 0.25Bq after 20 years. The catch and catch variability statistics for the first 20 years provide an indication of the likely impacts of recovery on the industry. The need to examine aspects of the behavior of the man- agement rules can be understood from Figure 3, which shows results for four simulations for the combination of a PFMC management procedure and an operating model variant. The solid lines are the "true" time-trajectories of spawning output (expressed in relation to the pre-exploi- tation level) and the dotted lines reflect the estimates of this ratio each time an assessment is conducted (every third year for the analyses shown in Fig. 3). The up ar- rows indicate when the assessment first indicates that the population is overfished (based on the model estimates of spawning output) — note that a population may be identi- fied to be overfished more than once during a given simula- tion. The down arrows indicate the years in which recovery is predicted by the rebuilding analysis software (with the estimates from the assessment) to occur with 60% probabil- ity. The solid bar parallel to the j:-axis indicates the years in which management is based on the rebuilding plan (rather than the 40-10 rule). The bar will stretch from the up ar- row to the down arrow unless the population is assessed to have recovered to 0.4Bq (when management reverts to being based on the 40-10 rule). There are several possible impacts of the difference between the perceived and true state of the system. For example, the population can erroneously be assessed not to be overfished in the first projection year (e.g. simulation 1 in Fig. 3). The statistic Pj^^, is designed to capture the frequency of this possibility. Even if the population is as- sessed to be overfished, there is no guarantee that it will recover with the expected probability and in the "correct" year. For example, for simulation 1, the stock assessment indicates that recovery occurs in year 71 (the solid bar con- sequently stops in year 71) even though the true population size is less than 30% of fig at that time. The statistic F^^^ attempts to capture whether the rebuilding analysis per- forms as expected given that the population is assessed to be overfished at the start of the first projection year. There are other aspects to evaluating the behavior of the management rules in relation to the perceived and true state of the system (e.g. the difference between the true and estimated biomasses and recruitments). Although it is straightforward to evaluate these aspects (e.g. Patter- son and Kirkwood, 1995; Punt et al., 2002), they are not considered in detail in this paper to reduce the volume of results presented. 866 Fishery Bulletin 101(4) Simulation 1 60 80 100 Simulation 3 Simulation 2 60 80 100 120 Simulation 4 Year Figure 3 Time-trajectories of the "true" and the assessment model-estimated ratio of the spawning output to Sq (depletion) for four simulations. The up arrows indicate the years in which the stock was declared to be in need of rebuilding and the down arrows show the values of T^^^. The horizontal bars indicate the years during which the stock is under a rebuilding plan. Year 41 is the first "projection year," i.e. the first year in which the management procedure is used to determine the catches (the catches for the years prior to year 41 are set equal to the historical catches — see Fig. lA) Results and discussion Detailed results for a single operating model variant and management procedure Figures 4 and 5 and Table 4 summarize aspects of a simulation trial in which the operating model has its baseline parameterization (Tables 1 and 2) and in which the management procedure used to set harvest guidelines is the PFMC management procedure with no constraints on interannual variation in harvest guidelines other than an upper limit of 10,000 t. The lack of any constraints on changes in harvest guidelines has been imposed because the PFMC has not adopted any such constraints. The har- vest guideline is updated every third year. Figure 4 shows the time-trajectories of catch, spawning output in relation to the pre-exploitation equilibrium level ("true" and estimated), and the perceived fishing mortality on which the harvest guideline is based for three of the 100 simulations that constitute a simulation trial. The horizon- tal bars on the x-axis again reflect the year during which the stock is managed by using the results from the rebuild- ing analysis rather than the 40-10 rule. The most notable feature of Figure 4 is the high variability in annual catches. This variability arises for several reasons: 1) the additional information on population biomass obtained each time a survey occurs changes the perceived status of the resource and hence how far the spawning output is from the target level of OABq-, 2) an extension of the assessment period changes the set of recruitments on which generation of future recruitment is based; and 3) a change from being under a rebuilding plan to being managed by means of the 40-10 rule can lead to marked changes in catch. The lat- ter is evident by the change in fishing mortality and catch when the spawning output is estimated to reach 0.4Sg (i.e. the end of the horizontal bar). A marked impact due to the addition of data for a single 3-year period may appear surprising. However, effects of this nature have already been observed for West Coast species (see, for example, the 2002 update to the sablefish [Anoplopoma fimbria] stock assessment [Schirripa and Methot^l). '^ Schirnpa. M. J., and R. Methot. 2002. Status of the sablefish resource off the continental U.S. Pacific Coast in 2001. In Stock assessment and fishery evaluation: appendix to the status of the Pacific Coast groundfish fishery through 2001 and acceptable bio- logical catches for 2002, x + 122 p. Pacific Fishery Management Council, 7700 NE Ambassador Place, Portland, OR 97220. Punt; Managing West Coast groundfish resources through simulations 867 r o E Simulation 1 Simulation 2 Simulation 3 Year Year Figure 4 Time-trajectories of catch (upper panels), spawning output in relation to the pre-exploitation level (solid line is "true"; dotted line is estimated) (center panels), and perceived fishing mortality (used to set the harvest guideline [solid line]; dotted line=F^,gY proxy) (lower panels) for three individual simulations. The results in this figure pertain to the baseline operating model and baseline Pacific Fishery Management Council manage- ment procedure. o o o 1 ID ^ . *" Depletion 0.4 08 Catch (t) 15000 1h \ .^,<'rrrrrT7mTrrr: § - o 6 y |^^^^^_^-,.wv„-— -^ 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Year Year Figure 5 Piecewise medians (solid lines) and 90% intervals (dotted lines) for spawning output in relation to the pre-exploi- tation equilibrium level (left panel) and catch (right panel). The results in this figure pertain to the baseline operating model and baseline Pacific Fishery Management Council management procedure. The extent of variability in catch in Figure 4 differs markedly from the way advice on expected catches during the rebuilding period is presented to the decision makers (e.g. Fig. 6). One way to improve the presentation of in- formation on expected catches would be to include some individual catch trajectories from those on which the rebuilding analysis is based. However, even these would severely underestimate the actual extent of uncertainty 868 Fishery Bulletin 101(4) Table 4 Performance statistics (see Table 3 for definitions) for six alternative management procedure variants. All of the calculations table relate to the baseline operating model. PFMC = Pacific Fishery Management Council. N/A = not applicable. m this Management procedure F^^^. Yre. ^decl Results after 20 years Results after 60 years 5%D 509-fD AAV C ^rec 5%D 50%D AAV C P rec Baseline 0.22 72 0.82 0.22 0.33 0.33 1759 0.32 0.23 0.36 0.25 2847 0.80 With constraints 0.27 61 0.82 0.24 0.40 0.38 591 0.54 0.24 0.41 0.17 2440 0.89 No 10 years and estimated f ^jsy 0.42 68 0.82 0.24 0.34 0.30 1652 0.27 0.25 0.41 0.24 2649 0.84 Preferred 0.59 62 0.82 0.25 0.39 0.31 950 0.49 0,28 0.54 0.21 1961 0.96 PFMC (baseline) N/A 95 N/A 0.19 0.29 0.23 2273 0.07 0.24 0,33 0.20 2851 0.55 PFMC (preferred) N/A 64 N/A 0.23 0.36 0.30 1239 0.45 0.30 0.48 0.20 2265 0.93 O ° 1 Catch (t) 1000 _^— :^^^ o 2000 2005 2010 2015 2020 2025 2030 Year Figure 6 Time-trajectories of catch (median and 95% intervals) for the annual catch for widow rockfish based on a rebuilding analysis conducted in 2002. because they are conditioned on knowing the age-structure of the population at the start of the projection period and are based on fixed levels of fishing mortality during the rebuilding period. The impact of estimation uncertainty is also evident in Figure 4. The following are three examples of this: 1) management based on the rebuilding plan only starts in year 53 in simulation 1 because, prior to this year, the stock assessment indicates (erroneously) that the stock is above rather than below 0.25Bq; 2) the resource is predicted to have recovered to 0.4Bg in year 71 in simulation 1 (and hence management is based on the 40-10 rule thereaf- ter)— however, the spawning output is really only slightly larger than O.Sfiy at this time; and 3) in simulation 3 the assessment model indicates that the spawning output has recovered to above 0.4Bg in year 65 when, in fact, it recov- ered to 0.4Bq three years earlier. The results of all 100 simulations are summarized by the time-trajectories in Figure 5. The trajectories of catch in Figure 5 are notably less variable that the individual trajectories in Figure 4 because, for instance, the 5"^, median, and 95'^ intervals for the catch in year 80 are obtained by sorting all 100 year-80 catches and taking the appropriate percentiles. Unlike the individual trajectories, the median trajectories of catch and spawning output show quite smooth changes over time. This result highlights the importance of the AAV statistic that captures interannual variation in catches within individual simulations. Overall, there is a high probability (0.82) that the as- sessment model identifies that the spawning output is less than 0.25Bq at the start of the projection period (Table 4). However, the probability that recovery occurs at or before the Tjjjax y6ar predicted from the rebuilding analysis con- ducted in projection year 1 is rather low (0.22) and 50% of simulations exceed 0.4Sq only in year 72 (i.e. after 30 years). The probability of being below the overfished level of 0.25Bq still exceeds 5% after 60 years of management with this management procedure although there is an 80% probability that the spawning output recovers to 0.4B„ sometime during the first 60 years of management with the management procedure. It should be noted that the impact of recruitment vari- ability and assessment errors following recovery to OABq can be consequential. For example, the probability of hav- ing reached 0.45q after 60 years of management by using the management procedure exceeds 0.8 but the median value of the ratio of the spawning output in year 60 to Bq is nevertheless still less than 0.4 (Table 4, Fig. 5). One rea- son for the spawning output not stabilizing at 0.4 Bg is a dis- crepancy between the fishing mortality rate that stabilizes the population at Bq (deterministically) and Fgg,. For the baseline steepness of 0.4, the fishing mortality required to stabilize the spawning output at 0.4 Bq actually corresponds to a lower fishing mortality than ^50% (closer to J^63%^- Sensitivity to alternative management procedures Table 4 includes results for a range of variants of the baseline management procedure designed to improve its performance. The following are areas where improved performance is desirable: 1) the extent of interannual variability in catches; 2) the similarity between the year Punt: Managing West Coast groundfish resources through simulations 869 in which the rebuilding analysis indicates recovery will occur and the year at or before which it actually occurs; and 3) the probability of being below the overfished level after 20 and 60 years. The first variant of the baseline management procedure ("with constraints" in Table 4) involves imposing maximum and minimum catch limits of 30 and 8000 t and constrain- ing changes in harvest guideline not to exceed 25% from one year to the next, except in the first year when reduc- tions of up to 99% are allowed. This variant leads to much lower interannual variation in catches when a 60-year pe- riod is considered ( 17% compared with 25% ) but the AAV is actually higher for the first 20 years. This variant also leads to higher probabilities of recovery. However, there is still a large discrepancy between the actual year of recov- ery to 0.4Sg and the year that underlies the management procedure (the value of F^^^ in Table 4 is only 0.27 for the "with constraints" variant). The second variant considered ("no 10 year and esti- mated ^msy" see Table 4) drops the requirement that T^^^ be defined as 10 years if the resource can be recovered in 10 years and instead always sets T^^^ to T^^^^ plus one mean generation. It also allows the Fj^gy proxy used when apply- ing the 40-10 rule to differ from the default value of Fgg^^ by setting it to F (Jakobsen, 1993) ii^ F is lower than ^MSY- Estimating (rather than fixing) F^^gy is consistent with the recommendation of Brodziak (2002). The major performance difference between this variant and the base- line management procedure is the increased value of F^.^^. The "preferred" variant in Table 4 combines the features of the "with constraints" and "no 10 years and estimated ^msy" variants. Compared with the baseline management procedure, it leads to a markedly increased value for F^^^ (remarkably close, in fact, to the desired value of 0.6), slightly lower catch variability, a less than 5% chance of being overfished after 20 years, and higher probabilities of being recovered to 0.4Bg after 20 and 60 years of manage- ment. The major disadvantage of this variant is the lower catches and that it leaves the spawning output well above 40% of Bg after 60 years (see row "preferred" in Table 4). Prior to the adoption of Amendment 11 of its Groundfish Management Plan, the PFMC set harvest guidelines using only the 40-10 rule.^ Table 4 therefore also lists results for management procedures based on the 40-10 rule. When the 40-10 rule is applied without any constraints ("PFMC (baseline)" in Table 4), the probability of recovery and the values for the "50%'D" statistic are lower (particularly the former) than for the "preferred" variant. In contrast, application of the 40-10 rule with constraints ("PFMC (preferred)" in Table 4) leads, arguably, to no more than a slight difference in cateh (the 40-10 rule achieves higher catches) and probability of recovery (the "preferred" vari- ant achieves a higher probability of recovery ). The remain- ing analyses of this paper focus on the "preferred" variant. Future consideration of management procedures for West Coast groundfish resources should consider a management procedure that is based simply on the 40-10 rule and has no associated rebuilding analysis component, at least for ' Albeit with different target fishing mortality levels. comparative purposes. At present, however, such a man- agement procedure would be inconsistent with the SFA because it would not specify the time to recover to the proxy for 5^,gY 'even if the results of this paper suggest that there is considerable uncertainty associated with the estimation of this particular quantity). Sensitivity to alternative operating model specifications The values assumed for h and a^ in the baseline operating model are somewhat arbitrary. Table 5 therefore examines the sensitivity of the results for the "preferred" manage- ment procedure to varying the values assumed for these parameters, as well as that of the size of spawning output at the start of the first projection year to fig. The results are, as expected, sensitive to all three of the factors considered. Increasing a^ from 0.4 through 0.6 to 1 leads to lower and more variable catches, a slightly higher probability of recovery in the first 20 years and a markedly higher value of 50%D after 60 years (0.74 for <7;j=l com- pared to 0.46 for cr^=0.4). The ability to detect an overfished stock declines slightly as the extent of variation in recruit- ment increases. The management procedure behaves as expected as steepness is increased from 0.25 through 0.4 to 0.7; the probability of recovery is markedly higher for high values of steepness even though the management proce- dure does identify cases with low steepness, and accordingly sets very low harvest guidelines in such cases. However, it is perhaps noteworthy that the probability of correctly identifying that the resource is overfished is lowest for the least productive scenario. The catches for the scenario in which the spawning output is 10% of Bg at the start of the first projection year are much lower than for the baseline scenario, particularly over the first 20 years. However, these lower catches are necessary to achieve recovery (the median value of the statistic 50%I> after 60 years is 0.52 and there is a 0.93 probability of the spawning output having recov- ered to 0.4Bg after 60 years for this scenario). The behavior of the management procedure can be evalu- ated in terms of whether it eventually allows the stock to recover to 0.4Bg and whether it keeps the stock away from the overfished level of 0.25Bq. The "preferred" management procedure can be argued to satisfy this criterion, except possibly for the scenario with the lowest steepness but, even in this case, the probability of recovery is 0.6 after 60 years. The value for the F^^^ statistic varies markedly depend- ing on steepness and the ratio of the spawning output at the start of the first projection year to Bg. Although the "preferred" management procedure performs well for the baseline scenario in terms of recovering the resource by the predicted value for T^^^, this good performance is clearly a fortunate anomaly. However, it does help to highlight that predictions of the year-to-recovery from rebuilding analy- ses should be interpreted with considerable caution. Sensitivity to data quality The data-related specifications for the baseline trial (Table 2) could be considered to be data-rich. It is therefore 870 Fishery Bulletin 101(4) Table 5 Performance statistics (see Table 3 for definitions) for 10 variants of the base) ine operating model. All of the calculations in this | table relate to the preferred management proce dure. N/A = not applicable. Operating model scenario f'.ec ^ec ■fdecl Results after 2C years Results after 6C years 5%D 50%D AAV C ^rec 5<7rD 50%D AAV C P,e, Baseline 0.59 62 0.82 0.25 0.39 0.31 950 0.49 0.28 0.54 0.21 1961 0.96 Structural changes a^ = 0.4 0.59 63 0.86 0.24 0.38 0.26 1242 0.44 0.25 0.46 0.18 2379 0.87 a^=l 0.59 61 0.72 0.23 0.41 0.43 417 0.54 0.32 0.74 0.32 592 0.96 /i=0.25 0.15 94 0.76 0.20 0.28 0.76 86 0.02 0.23 0.38 0.50 126 0.60 /!=0.7 0.84 53 0.87 0.31 0.46 0.16 3427 0.93 0.40 0.61 0.14 3951 1.00 Initial spawning out = 0.1 B„ 0.42 72 1.00 0.19 0.29 0.43 417 0.05 0.27 0.52 0.23 1375 0.93 Initial spawning out = 0.4 Bg N/A N/A N/A 0.31 0.50 0.21 2881 0.92 0.30 0.66 0.19 2849 0.97 Data-related changes Deterministic data 0.68 61 0.84 0.29 0.38 0.30 957 0.51 0.31 0.55 0.20 2050 0.98 n'=5Q 0.68 60 0.82 0.26 0.39 0.32 785 0.56 0.29 0.55 0.22 1938 0.97 a<-=l 0.56 62 0.79 0.20 0.39 0.31 987 0.48 0.31 0.57 0.22 1962 0.97 5-yr update frequency 0.55 62 0.80 0.21 0.38 0.27 1160 0.49 0.27 0.53 0.19 1980 0.95 important to assess the sensitivity of the results to the quahty of the data. The row "deterministic data" in Table 5 provides results for a trial in which the survey biomass index, the catch-rate index, and the age-composition data are known without error. The results from this trial provide an upper bound on the impact of improved data quality on the assessment results.^ Somewhat surprisingly, the results for this trial are not notably better than for the baseline trial — the most notable difference between the baseline trial and the "deterministic data" trial being the higher values for the "5%D" statistics for the latter trial. The lack of major improvement in performance arises because, even with perfect information on spawning output and recruitment, it is still not possible to estimate Bq exactly by multiplying average recruitment for the first 10 years of the assessment period by spawning output-per- recruit in the absence of fishing (hence the value of 0.84 for Pjed'- Furthermore, the rebuilding analyses are still based on generating future recruitment by using spawn- ing output and recruitment data for only 20 years, which is clearly a major source of variability in the predictions from the rebuilding analysis. Decreasing the catch-at-age sample size from 200 to 50 has relatively little impact on the values for the perfor- mance statistics (the AAV statistic is marginally higher and the average catch, particularly for the 20-year pro- jection horizon, is lower). Decreasing the precision of the catch-rate data has a rather larger impact. This is most evident in the value for the "5%D" statistic which is 0.2 rather than 0.25, as is the case for the baseline trial. The " The assessment still ignores interannual changes in selectivity; therefore the assessment results will not be exactly the same as the true values. "5-yr update frequency" scenario in Table 5 examines the implications conducting assessments every fifth rather than every third year. The results are not markedly sensi- tive to the interassessment period although the lower val- ues for the "5%D" statistics are perhaps noteworthy. General remarks The framework developed in this paper provides an objec- tive basis for contrasting different management procedures and evaluating their sensitivity to uncertainty. Given such a framework, it becomes possible to compare variants of one class of management procedure (e.g. Table 4) and to compare variants among different classes of management procedure. The management procedure options presented in this paper are but a small subset of those possible. In particular, it should be possible to improve performance by modifying the approach used to generate future recruitment when conducting rebuilding analyses to make use of some form of stock-recruitment relationship. One reason for expected improved performance is that it may then be feasible to estimate the fishing mortality rate corresponding to 0.4Bg rather than having to set it to the default value of F^j,^, or basing it on F . Other possible management procedure options include 1) not increasing the rebuilding fishing mortality rate selected when the rebuilding analysis was first conducted if a stock is recovering faster than initially anticipated; 2) not decreasing the rebuilding fishing mor- tality rate as long of the probability of recovery by T^^^^ is at least 0.5; and 3) smoothing the discontinuity that arises when a stock changes status from being under a rebuild- ing plan to being managed with the 40-10 rule when the Punt: Managing West Coast groundfish resources through simulations 871 stock has recovered to 0.4fig. In terms of the last option, one of the issues considered an early rebuilding analysis for widow rockfish involved fishing mortality increasing to its target level as the stock approaches 0.4BQ(MacCall^). The values for the F^^^ statistic highlight that the predic- tions of the time to recovery (even in a probabilistic sense) from rebuilding analyses are highly uncertain. The uncer- tainty of this estimate of the time to recovery is due to the uncertainty about current stock size and that associated with making long-term predictions based on a short time- series of spawning output and recruitment data. Although the performance of the management proce- dures is less than ideal, the results are almost certainly optimistic because the operating model is extremely simple and considers no major structural uncertainties (except for variability in selectivity over time). In contrast, Punt et al. (2002) found that including spatial structure in an oper- ating model and assessing the stock by using a spatially aggregated assessment approach led to assessments that were markedly in error. However, the simulations con- ducted by Punt et al. (2002) were developed for a far more data-poor situation than that for West Coast groundfish, although there is also clearly spatial structure in the West Coast groundfish fishery. Another source of uncertainty not considered in this paper but that may be of critical impor- tance to the management of West Coast groundfish species is the impact of environmental regime shifts, which have been argued to impact long-term trends in recruitment (e.g. Francis et al., 1998). An important aspect of this study is the ability to focus on the relationship between the overall performance of a man- agement procedure and the performance of its constituent parts. For example, the results for the "deterministic data" scenario in Table 4 show that given the approach used to conduct the future projections, even perfect information from surveys and very large age-composition samples are unlikely to lead to marked improvements over the current situation if that situation is adequately modeled by the baseline operating model. Identification of the key sources of uncertainty could be used to focus future management- related research activities. The computational requirements of the calculations out- lined above are substantial. In particular, the need to apply a fairly complicated method of stock assessment once every three years means that rapid evaluation of management procedures is (currently) computationally not feasible. It is possible, in principle, to simplify the management procedure considerably by assuming that the results from a stock assessment can be mimicked by generating a bio- mass estimate based on the "true" biomass but with some random error (e.g. Hilborn et al., 2002). However, although such an approach may be satisfactory for some manage- ment procedures (e.g. those that set the harvest guideline equal to some fraction of the current biomass), this is not the case for PFMC-type management procedures that de- pend on the (assessed) age-structure of the population. 9 MacCall, A. D. 2002. Personal commun. NMFS Santa Cruz Laboratory, 110 Shaffer Rd, Santa Cruz, CA 95060. It needs to be recognized that any simulation study is by design case-specific. However, the conclusions of this study may be relevant to a fairly broad set of West Coast rock- fish species owing to their similar biology and exploitation history — the two factors most likely to impact the relative performance of different management procedures. Acknowledgments Discussions with Alec MacCall, John DeVore, and Richard Methot are gratefully acknowledged as are the comments on an earlier version of this paper by Pamela Mace and two anonymous reviewers. This work was funded through NMFS grant NA07FE0473. Literature cited Barnes, W. R. 1999. Viewpoint: an industry view of the application of operational management procedures to setting total allow- able catches for the South African pelagic fishery. ICES. J. Mar. Sci. 56:1067-1069. Beddington, J. R., and J. G. Cooke. 1983. The potential yield of fish stocks. FAO Fish. Tech. Pap. 242:1-47. Brodiak, J. 2002. In search of optimal harvest rates of west coast groundfish. N. Am. Fish. Manage. 22:258-271 Butterworth, D. S., and M. O. Bergh. 1993. The development of a management procedure for the South African anchovy resource. In Risk evaluation and biological reference points for fisheries management (S. J. Smith, J. J. Hunt, and D. Rivard, eds. ), p. 83-89. Can. Spec. Publ. J. Fish. Aquat. Sci. 120. Cochrane, K. L., D. S. Butterworth, J. A. A. De Oliveira, and B. A. Roel. 1998. Management procedures in a fishery based on highly variable stocks and with confiicting objectives: experiences in the South African pelagic fishery. Rev Fish. Biol. Fish. 8:177-214. Dorn, M. W. 2002. Advice on west coast rockfish harvest rates from Bayesian meta-analysis of stock-recruit relationships. N. Am. Fish. Manage. 22:280-300. Francis, R. I. C. C. 1992. Use of risk analysis to assess fishery management strategies: a case study using orange roughy (Hoplostethus atlanticus) on the Chatham Rise, New Zealand. Can. J. Fish. Aquat. Sci. 49:922-930. Francis, R. S., A. Hare, A. Hollowed, and W. Wooster. 1998. Effects of interdecadal variability on the oceanic eco- systems of the NE Pacific. Fish. Oceanogr. 7:1-21. Geromont, H. F, J. A. A. De Oliveira, S. J. Johnston, and C. L. Cunningham. 1999. Development and application of management proce- dures for fisheries in southern Africa. ICES J. Mar. Sci. 56:953-966. Hilborn, R. 1979. Comparison of fisheries control systems that uti- lize catch and effort data. J. Fish. Res. Board Can. 36: 1477-1489. 872 Fishery Bulletin 101(4) Hilborn, R.. A. Parma, and M. Maunder. 2002. Exploitation rate reference points for west coast rock- fish: are they robust and are there better alternatives? N. Am. Fish. Manage. 22:365-375. lanelli.J. N. 2002. Simulation analyses testing the robustness of pro- ductivity determinations from West-Coast Pacific ocean perch stock assessment data. N. Am. Fish. Manage. 22: 301-310. IWC (International Whaling Commission). 1992. Report of the Fourth Comprehensive Assessment Workshop on Management Procedures. Rep. Int. Whal. Comm. 42:305-321. 2001. Report of the standing working group on the develop- ment of an Aboriginal subsistence whaling management procedure. J. Cetacean Res. Manag. 4(suppl.): 148-177. Jakobsen, T. 1993. The behaviour of F^^^, F^^^ and F(„g|, in response to variation in parameters used for their estimation. /;! Risk evaluation and biological reference points for fisheries man- agement (S. J. Smith, J. J. Hunt, and D. Rivard, eds.), p. 1 19-125. Can. Spec. Publ. J. Fish. Aquat. Sci. 120. Methot, R. D. 2000. Technical description of the stock synthesis assess- ment program, 56 p. NOAA Tech. Memo., NMFS-NWFSC- 43. Patterson, K. R., and G. P. Kirkwood. 1995. Comparative performance of ADAPT and Laurec- Shepherd methods for estimating fish population param- eters and in stock management. ICES J. Mar. Sci. 52; 183-196. Punt, A. E., A. D. M. Smith, and G. Cui. 2001. Review of progress in the introduction of management strategy evaluation (MSE) approaches in Australia's South East Fishery Mar Freshw. Res. 52:719-726. 2002. Evaluation of management tools for Australia's South East Fishery. 2. How well do commonly-used stock assess- ment methods perform? Mar. Freshw. Res. 53:631-644. Restrepo, V. R., and J. E. Powers. 1999. Precautionary control rules in US fisheries manage- ment: specification and performance. ICES. J. Mar. Sci. 56:846-852. Rosenberg, A. A., and S. Brault. 1993. Choosing a management strategy for stock rebuilding when control is uncertain. In Risk evaluation and biologi- cal reference points for fisheries management (S. J. Smith, J. J. Hunt, and D. Rivard, eds.), 243-249. Can. Spec. Publ. J. Fish. Aquat. Sci. 120. Sampson, D. B., and Y. Yin. 1998. A Monte Carlo evaluation of the stock synthesis assessment program. In Fishery stock assessment models (R Funk, T J. Quinn II, J. Heifetz, J. N. lanelli, J. E. Powers, J. F. Schweigert, P. J. Sullivan, and C. I. Zhang, eds.), p. 315-338. Alaska Sea Grant, Fairbanks, AK. Starr, P. J., P. A. Breen, R. H. Hilborn, and T H. Kendrick. 1997. Evaluation of a management decision rule for a New Zealand rock lobster substoek. Mar Freshw. Res. 48: 1093-1101. Appendix 1 : An overview of the technical aspects of the PFMC's rebuilding analysis The key steps of the PFMC's rebuilding analysis are 1) to select the maximum allowable rebuilding time iT^^^), R, if a = a„,„ W,,_ ,„-,<-""*'"-" if a„,„ PFMC (Pacific Fishery Management Council). 2001. SSC terms of reference for groundfish rebuilding analysis, 9 p. Pacific Fishery Management Council, 7700 NE Ambassador Place, Portland, OR 97220. Punt: Managing West Coast groundfish resources through simulations 873 It should also be noted that no account is taken of uncer- tainty regarding the current age structure, natural mortal- ity, selectivity, etc., although the projections do account for uncertainty about future recruitment year in which the spawning output first reaches 0.48^. T^^^ is the median of the distribution for this year constructed by conducting projections for many different (random) realizations of future recruitment. Selecting the maximum allowable rebuilding period The maximum allowable rebuilding time, T^^^, is defined as the maximum of 10 years and the sum of the mean generation time and the minimum possible rebuilding time. This specification implements the requirement of the SFA to "take into account the status and biology of any overfished stocks offish, [and] the needs of fishing commu- nities." The minimum possible rebuilding period for a given future projection is computed by projecting the population forward with zero fishing mortality and by identifying the Calculating the target fishing mortality rate The target fishing mortality rate and hence the harvest guideline are determined by projecting the population forwards many times ( 100 times for the purposes of this paper), each time with a different sequence of future recruitment and for a variety of alternative Fs and then identifying the level of F that corresponds to the spawning output having reached 0.4B(, by T^^^ with the prespecified probability p. 874 Abstract— Between March 2000 and April 2001 two commercial fishing ves- sels fished for toothfish (Dissostichus eleginoides) off South Georgia using pots. A significant number of lithodid crabs (three species of Paralomis spp.) were caught as bycatch. Paralo- mis spinosissima occurred in shallow water, generally shallower than 700 m. Paralomis anamerae, not previously reported from this area and therefore representing a considerable southerly extension in the reported geographic range of this species, had an interme- diate depth distribution from 400 to 800 m. Paralomis formosa was present in shallow waters but reached much higher catch levels (and, presumably, densities) between 800 and 1400 m. Differences were also noted in depth distribution of the sexes and size of crabs. Depth, soak time, and area were found to significantly influence crab catch rates. Few crabs (3% of P. spi- nosissima and T7c of P. formosa) were males above the legal size limit and could therefore be retained. All other crabs were discarded. Most crabs ( >99% of P. formosa, >97% off! spinosissima, and >90% of P. anamerae) were lively on arrival on deck and at subsequent discard. Mortality rates estimated from re-immersion experiments indi- cated that on the vessel where pots were emptied directly onto the factory conveyor belt 78-89% of crabs would survive discarding, whereas on the vessel where crabs were emptied down a vertical chute prior to being sorted, survivorship was 38-58%. Of the three, P. anamerae was the most vul- nerable to handling onboard and sub- sequent discarding. Paralomis spino- sissima seemed more vulnerable than P. formosa. Distribution, demography, and discard mortality of crabs caught as bycatch in an experimental pot fishery for toothfish {Dissostichus eleginoides) in the South Atlantic Martin G. Purves Marine and Coastal Management P.O. Box X2, Roggebaai 8012 Cape Town, South Africa Present address: Resources Assessment Group 47 Pnnce's Gate London SW7 7QA, United Kingdom E-mail address; m purvesigimpenal acuk David J. Agnew Renewable Resources Assessment Group Imperial College Royal School of Mines Prince Consort Road London, SW7 2BP, United Kingdom Guillermo Moreno Tim Daw MRAG Ltd 47 Pnnces Gate London, SW7 2QA, United Kingdom Cynthia Yau Zoology Department University of Aberdeen Aberdeen, AB24 2TZ, United Kingdom Graham Pilling Centre for Environment, Fisheries and Aquaculture Science (CEFAS) Pakefleld Road Lowestoft, NR33 OHT, United Kingdom Manuscript approved for publication 17 April 2003 by Scientific Editor Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish Bull. 101:874-888 (2003). The Commission for the Conservation of Ajitarctic Marine Living Resources (CCAMLR) and its Scientific Committee were pioneers in the development of the "ecosystem approach" for the manage- ment of fisheries. Using this approach the Commission is bound to consider the impact of any fishery on both the target, dependent, and related species. Currently, the most important fishery in CCAMLR waters is the longline fishery for Patagonian toothfish (Dissostichus eleginoides) and fishing grounds near South Georgia Island and Shag Rocks in CCAMLR subarea 48.3 (South Atlantic sector) are among the most important. Mitigation measures, including require- ments for setting at night, in the winter, and with specialized gear, have been introduced to reduce incidental mortal- ity of birds being hooked by longlines. However, these measures impose severe Purves et al.: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 875 operational restrictions on the fishery, and low levels of bird mortality still occur (CCAMLRM. Pot fishing for toothfish has recently been tried around South Georgia (Agnew et al., 2001), and although pots do not catch birds, they do take lithodid crabs as bycatch. Crabs are largely a "nuisance" catch when fishing for toothfish, but this bycatch is clearly of concern in rela- tion to crab populations, and must be considered within the CCAMLR's ecosystem approach. A small amount of exploratory crab fishing has already taken place around South Georgia Island and Shag Rocks. Only 798 metric tons (t) of crabs have been taken in directed crab fisheries since 1992; by the FV Pro Surveyor (July-August 1992; 299 t; CCAMLR2), the FV American Champion (September 1995-January 1996; 497 t; CCAMLRM), and the FW Argos Helena (August 1999; 2 t) (CCAMLR''). A pot fishery for toothfish is likely to take place in deep water where cur- rent longline fishing concentrates (around the 1000 m con- tour; Agnew et al., 1999) rather than in shallower waters (<400 m) where crab fishing has taken place (Otto and Macintosh, 1996; Watters, 1997 ). Toothfish pot fishing may therefore impact different crab population components from those impacted by the crab fishery. We investigated the likely effects of toothfish pot fishing on crabs caught as bycatch on board two commercial fish- ing vessels. These vessels conducted three separate trials fishing for toothfish around South Georgia between March 2000 and April 2001. This paper reports on the species of crab taken during these trials, as well as the distribution of crabs and their biological characteristics. In common with many other crab fisheries (Hoggarth, 1991; Schmidt and Pengilly, 1993) retention size limits are set for the fishery around South Georgia. Only males greater than 102 mm carapace width for Paralomis spinosissima and 90 mm for P. formosa may be retained. All undersize and female crabs from both the toothfish pot fishery and the crab fishery must be discarded. We also report results of experiments on survival of such discards. Methods During the first cruise (March to May 2000), two observ- ers were deployed on board the FV Argos Georgia (cruise Gl). Detailed information was collected on the number of toothfish and the numbers and species composition of ' CCAMLR (Commission for the Conservation of Antarctic Marine Living Resources). 1999. Report of the Working Group for Fish Stock Assessment, 110 p. Annex 5 in the report of the eighteenth meeting of the Scientific Committee. CCAMLR, RO. Box 213, North Hobart, Tasmania 7002, Australia. - CCAMLR. 1992. Report of the working group for fish stock assessment, 164 p. Annex 5 in the Report of the eleventh meet- ing of the Scientific Committee. CCAMLR, RO. Box 213, North Hobart, Tasmania 7002, Australia. ^ CCAMLR. 1997. Report of the working group for fish stock assessment, 169 p. Annex 5 in the Report of the sixteenth meet- ing of the Scientific Committee. CCAMLR, P.O. Box 213, North Hobart, Tasmania 7002, Australia. ■• CCAMLR. 2000. CCAMLR statistical bulletin, 153 p. CCAMLR, RO. Box 2 13, North Hobart, Tasmania 7002, Australia. I ■ P. spinosissima n P formosa aP anamerae Depth range (m) Figure 1 Catch in numbers per pot of the three dominant crab species by depth range as found during fishing opera- tions of the Argos Georgia from May to April 2000. Pots were set between 211m and 1651 m depth. crabs caught in the pots. Information was also collected on fishing results, such as catch rates, fish bycatch, and the commercial viability of this fishing method (Agnew et al., 2001), and the diet of toothfish (Pilling et al., 2001). In the second and third toothfish pot cruises single observers were deployed on the Argos Georgia (cruise G2) and another vessel, the FV Argos Helena (cruise H), which fished simul- taneously from January to April 2001 . Fishing gear and the configuration of gear was similar for all three cruises. The semiconical pots of approximately 80 cm height were con- structed of steel frames and covered with 80-mm polysteel (Movline) mesh. A collapsible funnel entrance was situated on the side of the pot, orientated horizontally, and tapering to the pot's interior. A drawstring held the bottom mesh together in the middle. This configuration allowed the pots to be emptied easily when hauled aboard and to be stacked on top of each other when not in use. A panel was sewn into the pots by using biodegradable sisal twine to ensure that crabs could eventually escape from lost pots and to prevent "ghost fishing." However, catch handling methods were different on the two vessels; the pots were emptied directly onto the factory conveyor belt on the Argos Georgia and emptied down a chute to the factory level on the Argos Helena. During the first cruise (March to May 2000) depth of fish- ing, determined as water depth by onboard echo sounders, was related to the species of crab caught. Paralomis spino- sissima were generally caught in relatively shallow waters, whereas P. formosa tended to be caught in much greater numbers in deeper waters (Fig. 1). Paralomis anamerae 876 Fishery Bulletin 101(4) Figure 2 Map of areas used in the study. SR=Shag Rocks. NG=North South Georgia, EG=East South Georgia, SG=South South Georgia, and the 200-m, 1000-m and 3000-m bathymetric contours are shown. The total number of hauls in each of these areas were as follow: SR=145, NG=50, EG=30, and SG=15. Inset shows the position of the study area in relation to the South Atlantic. (Macpherson, 1988) were caught at intermediate depths. Other influences on crab catch rates were investigated by using two cruises in 2001, which covered the study area more evenly (most hauls in cruise Gl were concentrated in a small area around southeast Shag Rocks; see Fig. 2). For each species two generalized linear models were con- structed with Splus statistical software (version 2000, Mathsoft Engineering & Education, Inc., Cambridge, UK): a binomial link model for the probability of encountering a crab of that species (pe) and a Gaussian link model for the natural logarithm of CPUE (catch in numbers per pot) for all nonzero catches. Both models were of the form A=pjX depth+p.2xarea-(-p3Xvessel+p4Xsoak time, where for the binomial models, A was set to 1 if crabs had been caught and 0 if they had not, and for the Gaussian models, A was set to ln(numbers per pot) for all sets catching crabs. Area and vessel parameters were factors. Depth and soak time were modeled as linear continuous variables, except in the case of the Gaussian model for P. forniosa, where a third- order polynomial best described the relationship between CPUE and depth. Predictions from the two models were combined to predict crab catch rates per pot, (} = ;7(c)xexp[H±1.96x5£], where/'/ = the predicted In(CPUE) from the Gaussian model; and SE = the standard error from the Gaussian model. Biological data were collected from all crabs in randomly selected pots. Carapace widths, carapace lengths, chela height, and chela length were measured to the nearest millimeter below by using calipers. Weights were measured to the nearest 5 g below with spring-balanced scales. Sex, maturity stage, condition of the carapace, and an index of vitality (Table 1) were recorded for a subsample of crabs, selected as a random portion across species from the contents of selected pots. Identification of the uncommon species, Paralomis anamerae, Neolithodes diomedeae, and Lithodes murrayi, was confirmed by using dried specimens in London (at Imperial College and the British Museum of Natural History). Male size at maturity was determined from the allometric relationship between carapace length (L) and right (domi- nant) chela size (height, CH, or length, CD. The slope of the L-CH or L-CL relationship is assumed to change when crabs reach maturity. Around South Georgia, Otto and Macintosh (1996) used L and CH to determine the size at maturity of male P. spinosissima, and around the Falkland Islands Hoggarth (1991) used L and CL. For both P. spino- sissima and P. fonnosa we found that the intersection of the two lines corresponding to the onset of maturity was not easy to identify from the relationship between CL and L. We therefore used CH in our relationships. Following the methods ofSomerton and Otto (1986), two linear regression lines were fitted to natural logarithms of L and CH or CL. These lines represent juvenile and adult phases in the L-CH or CL relationship and intersected at a point taken as the L at which males became mature. The regression lines of best fit were determined by minimizing the combined residual sums of squares, and standard errors were estimated by using 500 bootstraps (sampling with replacement). Females were classified into two categories, "eggs absent" and "eggs present." The vast majority of "eggs absent" fe- males were small immature animals, but some large ani- mals were also encountered in this category. Consequently, for estimating size at maturity, females in the "eggs pres- ent" category were defined as "mature" and those in the "eggs absent" category were classified as "immature" up to the size at which the proportion of females with eggs (i.e. mature) reached 90%, after which they were classified as "mature without eggs." Female size at maturity was deter- Purves et al.: Distribution, demography, and discard mortality of crabs caugtit as bycatch in the South Atlantic 877 Relative Table 1 index used for assessing vitality in Paraloniis spp. Vitality index Description Characteristics 1 2 3 4 Lively Lively but limp Dead Dead and eaten Limbs supported and held out. Limbs resist manipulation, objects. Can hang on smooth end of forceps by 1 claw (weakest Legs hang when picked up. Claws weak and can be opened — not own weight on forceps. Mouthparts move, indicating life seawater. No signs of life. Mouthparts do not move when submerged. Only shell or carapace remaining. Crabs actively pinch on large crabs). capable of supporting when submerged in mined by plotting the proportion of mature females against size (carapace width) and determining the point of 50% maturity (Sragg). Logistic curves of the form {proportion mature = 1/(1 + expi-ricarapace width-Sm^Q))}] were used to estimate 50% maturity (Sm^g) and its standard error by using the nonhnear fitting function in Splus. Three different experiments were conducted between March 2000 and April 2001 to assess crab discard survival rates. During cruise Gl a number of alive and active crabs from one haul were tagged through one of the lateral plates of the abdomen with Hallprint plastic T-bar tags and main- tained in running seawater before they were placed in pots prior to the next setting. Once these pots were rehauled, the vitality of the tagged crabs was assessed by using the four point relative scale shown in Table 1. A control group of crabs were similarly tagged and kept onboard in running seawater to monitor any effect that tagging might have had on sur- vival. Survival experiments conducted on cruises G2 and H differed in that crabs were selected at random and included individuals that were already "limp" prior to re-immersion. Crabs were tagged with thin strips of masking tape around their legs prior to re-immersion. To ensure that the same crabs were assessed for vitality after rehauling, pots were marked and sealed off to prevent any new captures. Estimates of the total survival that can be expected after discarding were made in the following manner. By observing crabs on arrival on deck we determined the pro- portion of animals that arrived on deck as lively {lively^), or limp (limp J, or dead. Using the data from the survival experiments we set p(liuely,lively) as the probability that a lively animal that is discarded will recover to a lively condition (this was estimated by calculating the proportion of experimentally re-immersed lively animals that were recovered as lively). We defined p(limp,liuely) similarly. The proportion of discarded animals that were lively and would continue to be lively is lively ^xpdiuely, lively) and the proportion of discarded animals that were limp but would recover to a lively condition is limp ^xpdimpjively). The overall survivorship, S, is then S = lively ^xpilivelyMvely) + limp^xpilimp, lively). In our experiments, some of the damage may have occurred on rehauling the pots after re-immersion, a situation that would not normally occur once crabs have been discarded. p(lively,lively) can be corrected for this by adding to it the proportion of animals that were not lively when first hauled (i.e. 1-livelyJ, termed the rehaul correc- tion. For instance, suppose that 1% of the crabs were not lively on the first hauling, and in the experiment 4% of the re-immersed lively crabs were not lively on recovery. The rehaul correction would indicate that 1% of the re-im- mersed crabs would have been damaged anyway simply by the hauling process, and that therefore the correct damage rate attributed simply to the initial capture and discarding would be 3%. Results Crab catch The majority of the bycatch comprised two species of litho- did crabs (Anomura: Lithodidae), Paralomis spinosissima and P. formosa. Both species have been previously reported in catches around Shag Rocks and South Georgia (Otto and Macintosh, 1996) and have formed a large proportion of the total catch of crab (Table 2). Crab species formed 69.5% of the total weight of all species caught, including toothfish, and 98.2% of the total numbers of individuals caught. Three other species of crab were also identified during the pot trials. The most abundant of these was Paralo- mis anamerae (Macpherson, 1988), which like the other Paralomis species, was subject to detailed biological sam- pHng. 12,370 individuals of this species (1 721 kg) were caught. All individuals were discarded because they were smaller than the minimum size limit for the smaller of the two regulated species, P. formosa. Two other species, Neolithodes diomedeae and Lithodes murrayi, were also caught in small numbers. Distribution Crab distribution was investigated for the areas as defined in Figure 2. There were too few data from South of South Georgia; therefore the analysis was restricted to data from Shag Rocks, North South Georgia, and East South Georgia. A few pot strings had been left for several days because 878 Fishery Bulletin 101(4) Table 2 Species proportion (no. of crabs and percentage) and discard rates for crabs caught during the pot trials around South Georgia | during the period March 2000 to April 2001. Number Number % total crab catch % total crab catch Discard kept discarded by number by weight rate Paralomis formosa 22,803 300,660 70.1% 63.1% 93% Paralomis spinosissima 3576 121,580 27.1% 35.4% 97% Paralomis anamerae 0 12,370 2.7% 1.4% 100% Neolithodes diomedeae 0 <0.1% <0.1% Lithodes murrayi 0 <0.1% <0.1% Table 3 The results of fitting generalized linear models on the probability of encountering crabs, and the catch rate of crabs (numbers per pot) for nonzero catches. ANOVA results; the significance of adding each parameter to the general linear model is presented. For the binomial model of the probability of encounter, a chi-squared significance test was used; for the Gaussian model of ln(CPUEl an F test was used. Coefficients p,..£)j are given with standard error in parentheses. For those models including "area" the standard case was for East South Georgia (EG), and the parameters for North South Georgia and Shag Rocks (NG, SR) are given. Analyses were performed with Splus statistical software. Final models were constructed by using only the significant parameters (italicized in this table), n = number of crabs in sample. Poly (1), Poly (2), and Poly (3) are the coefficients of each of the three orders in a 3rd order polynomial. Probability of encountering Parameter P. formosa ANOVA results n 101 Depth (m) <0.001 Area >0.05 Vessel >0.05 Soak time (h) >0,05 Coefficients P[ p^ n 101 Intercept -2.2358(1.4373) Depth (m) 0.0098(0.0035) Area Soak time (h) Ln(CPUE) for nonzero catches of P. formosa Probability of encountering P. spinosissima Ln(CPUE) for nonzero catches of P. spinosissima 91 <0.001 >0.05 >0.05 <0.05 91 1.7229(0.4301) Poly(l) Poly(2): Poly(3): 4.6387 (5.3605) -8.4647 (6.2257) -1.3614 (4.4808) 101 <0.001 >0.05 >0.05 >0.05 101 8.5065(1.5822) 0.0106 (0.0022) 0.0447 (0.0221) 82 <0.001 <0.01 >0.05 <0.01 82 3.217(0.4877) -0.0055 (0.0008) NG: 1.1304(0.3241) SR: 1.2771(0.2875) 0.0638(0.0191) of bad weather, and to eliminate these the maximum soak time was Hmited to 39 hours. The mid-depth of the set (average of the depth at the start and end of the set) was used to indicate setting depth, and the analysis was restricted to sets whose depth range was less than 200 m. The data from 45 sets (19% of total) were omitted from analyses because they did not meet these criteria for area, depth, or soak time. For the binomial encounter models, the only significant factor was depth (Table 3). F'or the Gaussian catch per pot model, depth, and soak time were significant for both spe- cies and area was significant for P. spinosissima. However, the depth effects were opposite for the two species, so that P. spinosissima decreased in abundance with depth and P. formosa increased in abundance with depth, at least up to about 1000 m depth (Fig. 3). There were insufficient data from the cruises in 2001 to establish the effect of depths shallower than 300 m and deeper than 1100 m, although the limited sampling at depths greater than 1200 m in year 2000 (Fig. 1) suggests that catch rates of P formosa would continue to decline at these depths, as suggested by the generalized linear models in Figure 3. The sex ratio of P. formosa was skewed towards males in shallow water (<800 m) and females in deep water (Fig. 4). The mean size of P. formosa of both sexes also decreased significantly with an increase in depth (Fig. 5). Although catch rates in numbers were usually much higher in deeper water, smaller crabs of no commercial Purves et al.: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 879 value often dominated catches. Only about 38% of P. spinosissima sampled between 200 and 400 m were males, whereas this proportion increased to about 76% in depths of 600 to 800 m (Fig. 4). The mean carapace width of males remained relatively constant between 200 to 800 m, but female P. spi- nosissima decreased in size with an increase in depth (Fig. 5). Size frequencies Males achieved larger sizes than females in all three species (Fig. 6). Only 5.7% of the sampled P. spinosissima individuals were of legal size (carapace width greater than 102 mm), of which only 6% were females. A difference was also noted in the percentage of legal-size P. spinosissima for the different areas: 10% at South Georgia and only 3.8% at Shag Rocks. For P. formosa only 11.6% (/! = 1012) were larger than the minimum legal size of 90 mm. Of these legal crabs only 6%- (n=63) were females, indicating that few females would be processed if carapace width was the only criterion used to select crabs that could be taken legally. No obvious difference was noted in the percentage of legal-size crabs caught in the two main fishing areas. Although a legal-size limit is not specified for P. anamerae (sizes ranged from 39 to 96 mm), only two crabs (0.2% of sample) were larger than the legal limit for P. formosa (90 mm). A peak in the length distribution of this species occurred be- tween 55 and 57 mm, and few crabs were larger than 77 mm. Differences in the size-frequency distribution of the sexes was more pronounced for P. formosa than for the other two species; females peaked at 65-72 mm and males peaked at between 85 and 90 mm carapace width (Fig. 6). For P. spinosissima female size distribution peaked between 78 and 82 mm and males peaked at 87 to 92 mm carapace width (Fig. 6). Males and females of P. anamerae had a relatively even size distribution up to cara- pace widths of 65 mm. Most of the larger crabs were males; only 6.8% of individuals larger than 65 mm were females. Maximum widths recorded were 121 mm (P. spinosissima), 120 mm {P. for- mosa), and 91 mm (P. anamerae). Size at maturity There was no significant difference between female size at maturity at Shag Rocks and South Georgia for either species (^-tests on Sm^g esti- mated by fitting logistic models to the proportion mature at size, P>0.05). The close allometric relationship found between carapace width and length (Table 4) made it possible to alter between these two types of measurements (Table 5). Combining the two areas, female size at maturity was 55.1 mm carapace length 200 300 400 500 600 700 800 Depth (m) — P formosa - -*- - P spinosissima. EG 900 1000 1100 1200 .fi. ■ . P spinosissima. SR Figure 3 General-linear-model-predicted CPUE (numbers per pot) for the two species (Paralomis formosa and P. anamerae) at different depths standardized for a soak time of 18 h (the average used in the study). For P. spinosissima predictions at East South Georgia and Shag Rocks are presented separately (EG, SR, respectively). n = 1988 80 70 ■ 60 - 50 40 30 20 10 % IVIales of P formosa % IVIales of P spinosissima 200-400 400-600 600-800 800- 1,000 1 ,000- 1,200 1,200- 1,400 Depth range (m) Figure 4 The percentage of males of Paralomis formosa and P. spinosissima found at different depth ranges during random sampling of the crab bycatch; all data from cruises Gl (Argos Georgia cruise 1), G2 (Argos Georgia cruise 2), and H (Argos Helena cruise) were used. (57.1 mm carapace width) for P. formosa and 61.2 mm carapace length (67.7 mm carapace width) for P. spinosis- sima (Fig. 7). These sizes are very similar to the 61.7 mm 880 Fishery Bulletin 101(4) 90 1 ~ 80 70 65 60 P formosa M PT?^~-^-.=^_^-- n=1987 P formosa F ^^^~-^^....^^^ P spinosissima M ,\ N V P spinosissima F n= 1504 n= 1749 200-400 400-600 600-800 800-1.000 Depth range (m) 1,000-1.200 1.200-1,400 Figure 5 Mean carapace width of the males (M) and females (F) of P formosa and P. spinosissima found at different depth distributions. Table 4 Parameters of the regression carapace width - -- carapace length x slope -1- intercept (all ages and sexes combined), n = number of crabs in sample. Species Intercept Slope Correlation coefficient n P. spinosissima 6.457 0.994 0.977 367 P. formosa 1.976 1.032 0.969 351 P. anamerae 5.700 0.917 0.979 28 The size at maturity (Sm^^) for ma are shown in parentheses. Table 5 es and females of both Paralomis species. The standard errors (SE) for male carapace lengths P. spinosissima P. formosa Carapace length (mm) Carapace width (mm) Carapace length (mm) Carapace width (mm) Males 67.3(0.10) Females 61.2 73.4 64.0(0.16) 68.0 67.7 55.1 57.1 carapace length reported by Otto and Macintosh (1996) for P. spinosissima from Shag Rocks and South Georgia. Unfortunately, owing to limited sampling of male crabs around South Georgia, estimates of male sexual maturity were only available for Shag Rocks. Too few samples off! anamerae were available for determination of the onset of female or male maturity in either area. Male size at maturity (Sm^g) was determined at a carapace length of 67.3 mm (SD=2.3 mm derived from bootstrap resampling with replacement) for P. spinosissima and at 64.0 mm (SD= 3.6 mm) for P. formosa (Fig. 8). Crab survival rate Approximately 5000 crabs were examined for carapace damage on the cruises in 2001. The results (Table 6) indi- cated that the level of visible damage to these crabs prior to discarding was very low (2% of crabs of all species). The vitality of these crabs was also assessed (according to the index in Table 1). Most crabs were lively on arrival on deck and prior to discard (Table 6). Differences were, however, noted in both carapace condition and the vitality of crabs between the two fishing vessels. Pairwise x^ comparisons Purves et a\: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 881 P spinosissima P. anamerae 50 60 70 80 90 Carapace width (mm) Figure 6 Carapace-width frequency distribution for the three Paralomis species (mm). (A) P. spinosissima, (B) P. formosa, (C) P. anamerae. The minimum legal size (102 mm for P. spinosissima and 90 mm for P. formosa) is indicated with a vertical line and arrow. between the vitality indices of P. formosa and P. spinosis- sima clearly showed that both species displayed signifi- cantly lower vitality on the Argos Helena than on the Argos Georgia (Table 7). The processing environment of the two vessels may ex- plain these differences. On the Argos Helena crabs were likely to sustain more damage as pots were emptied down a vertical chute before entering the processing area below deck. On the Argos Georgia pots were emptied on a hori- zontal conveyer belt leading to the factory. Interestingly, there was no significant difference between the vitality displayed by P. anamerae between the two vessels, al- though this may be a result of the smaller sample size for this species. For the Argos Georgia, the vitality of P. ana- merae was significantly lower than for either P. formosa or P. spinosissima. The results of the three survival experiments are shown in Table 8. Experiment 1 re-immersed only lively crabs and included a control set of animals retained on deck in a large tank for the same length of time as the re-im- 882 Fishery Bulletin 101(4) 1 0.9 0.8 0.7 0 6 0.5 0.4 - 0.3 0.2 0.1 0 1 0.9 0.8 0.7 ■ 0.6 0.5 0.4 0.3 0.2 0.1 0 10 20 30 40 50 60 70 80 90 100 110 120 B p. formosa 0 10 20 30 40 50 60 70 80 90 100 110 Carapace width (mm) Figure 7 Size at female maturity for (A) P. spinosissima and (B) P. formosa. The size at maturity U^q) carapace width was 67.7 mm and 57.1 mm for the two species, respectively. Table 6 The number and percentages of crabs of the different species assessed for carapace condition and vitality prior to being discarded during 2001 iG2=Argos Georgia cruise no 2, H=A rgos Helena ). Species and cruise Carapace condition Vitality ind ex Damaged Undamaged Total Lively Limp Dead Total No. % No. % No. % No. % No. % Paralomis formosa G2 48 1.9 2431 98.1 2479 2294 98.9 26 1.1 0 0.0 2320 H 23 2.7 814 97.3 837 745 97.4 19 2.5 1 0.1 765 Total 71 2% 3245 98% 3316 3039 98% 45 1% 1 0% 3085 Paralomis spinosissima G2 17 1.5 1143 98.5 1160 1122 98.4 17 1.5 1 0.1 1140 H 18 2.0 874 98.0 892 767 94.8 39 4.8 3 0.4 809 Total 35 2% 2017 98% 2052 1889 97% 56 3% 4 0.2% 1949 Paralomis anamerae G2 0 0.0 45 100.0 45 31 86.1 5 13.9 0 0 36 H 2 2.7 72 97.3 74 62 92.5 5 7.5 0 0 67 Total 2 2% 117 98% 119 93 90%. 10 10% 0 0% 103 Purves et al,: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 883 Table 7 Results of pairwise x'^ comparisons of the vitality of crabs on first being caught by different cruises (from Table 6). G2 = Argos Georgia cruise no. 2 and H = Argos Helena, F = Rformosa.S = P. spinosissima,A = P. anamerae. x'^ values are given, together with significance, df = 2 for all except i ,alicized results, when df was 1 beca use of the absence of anv dead crabs in the comparisons. NS = not significant. G2-F H-F G2-S H-S G2-A G2-F H-F 10.49, P<0.01 G2-S 2.9, NS — H-S — 6.99, P<. 05 20.7,P<0.001 G2-A 35.2. P<0.001 — 29.2,P<0.001 — H-A — 5.5, NS — 1.14, NS 0.49. NS P. spinoxissima mersion process. Out of the 35 lively control animals a similar proportion died dur- ing the experiment as in the re-immersed group (8%), but a lower proportion of the control group were lively following the experiment (63%). The lower proportion of lively animals in the control group may have been a result of interruptions in the supply of oxygenated water and the continual dis- turbance on deck due to the ship's motion. Consequently, controls were not performed in experiments 2 and 3. Experiments 2 and 3 took a random sample of lively and limp crabs and subjected them to re-immersion. The proportions of lively crabs at the beginning of the re-im- mersion experiments were lower than the proportions estimated for the population as a whole in Table 6, with the exception off! spinosissima in experiment 2. However, sample sizes were much smaller in the experiments and the crabs were subject to greater handling times than those as- sessed in Table 6; therefore the results presented in Table 6 are more likely to be representative of the condition of discarded crabs than results presented in Table 8. In experiment 1 on cruise Gl, no limp animals were subjected to re-immersion. In comparing the results of immersing lively animals, there was no significant differ- ence between the results for experiment 1 and 2 (cruises Gl and G2 on the Argos Georgia) for either P. formosa or P. spinosissima. Combining the results of the experiments for Gl and G2, there was a significantly lower vitality after re-immersion for both species on the Argos Helena compared with the Argos Georgia (Table 7). On the Argos Georgia there was no significant difference in vitality af- ter re-immersion between P. formosa and P. spinosissima, whereas on the Argos Helena vitality for P. spinosissima was significantly lower than vitality for P. formosa. These 2 It x; JO TO ii c _i 25 2 Ln {carapace length) Figure 8 Chela allometry in male (A) P. spinosissima and (B) P. formosa sampled at Shag Rocks in 2001, The intersections of fitted curves were used to determine size at maturity. results are similar to the initial assessment of vitality prior to re-immersion (Table 6), where significantly fewer P. spinosissima were lively in comparison with P. formosa on the Argos Helena , but there was no difTerence between the two species on the Argos Georgia. A single re-immersion experiment performed on 15 P. anamerae crabs (11 were "lively" and 4 were "limp"), on the Argos Helena in April 2001, resulted in a mortality rate of 73%. Only 27% of the crabs that were "lively" before re-immersion were still "lively" after the rehaul. Although more data need to be collected on the survival rate of this species, this high mortality rate, together with the higher incidence of individuals off! anamerae found to be "limp" during vitality assessments (Table 8; 10% compared to 3% of P. spinosissima and 1% of P. formosa), seems to indicate that this bycatch species might be particularly vulnerable to onboard handling and discarding. Crabs that were physically damaged (i.e. had missing legs or cracked carapaces) before being subjected to re-im- mersion were less hkely to survive than undamaged crabs. Of the 19 damaged P. spinosissima (13 of these were "lively" and 6 were "limp"), 58% did not survive re-immersion and only 32% of these were still "lively." The effect of damage 884 Fishery Bulletin 101(4) Table 8 Results of survival-rate experiments. For each experiment the number of re-immersions is given together with the total number of crabs in lively and imp condition that were re-immersed and their condition on rehauling the c •ab pots after re immersion. Number of Number of re-immersions crabs Initial condition Condition after re- mmersion P. formosa Experiment 1 Lively Limp Dead (cruise Gl) 2 30 Lively 20 6 4 0 Limp 0 0 0 Experiment 2 Lively Limp Dead (cruise G2) 6 98 (93%) Lively 81 11 6 7 Limp 4 1 2 Experiment 3 Lively Limp Dead (cruise H) 3 49 (91%) Lively 27 21 1 5 Limp 0 2 3 P. spinosissima Experiment 1 Lively Limp Dead (cruise Gl) 2 42 Lively 35 3 4 0 Limp 0 0 0 Experiment 2 Lively Limp Dead (cruise G2) 6 60(100%) Lively OLimp 55 0 5 0 0 0 Experiment 3 Lively Limp Dead (cruise H) 10 167 (88%) Lively 71 67 29 23 Limp 1 6 16 Table 9 Estimation of total survival rate for P. formosa and P. spinosissima based on results from re-immersion experiments. Because there was no significant difference between the responses of lively animals between experiments 1 and 2, these data were pooled to give a single estimate for the Ar^^os Georgia. On the Argos Georgia no Hnip P. spinosissima were encountered during the re-immersion exper- iments; therefore p(?imp,/iiie/y) was set equal to p(limp,liuely) from experiment 3. Three calculations are presented: the first according to the text description without the rehaul correction, the second including this rehaul, and the third where the original proportions of lively anima\s pdiuely J were used from the experimental data in Table 8 rather than the data from the larger sample in Table 6. pdiuely J from Table 6, without rehaul correction pdivelyj from Table 6, with rehaul correction pdivelyj from Table 8, without rehaul correction pdivelyj from Table 8, with rehaul correction P. formosa Argos Georgia 78.7% 79.8% 77.5% 83.7% Argos Helena 53.7% 56.2% 50.0% 58.4% P. spinosissima Argos Georgia 86.9% 88.5% 88.2% 88.2% Argos Helena 40.5% 45.4% 37.9% 48.5% was not so pronounced for P. formosa, although of the 12 crabs that were damaged prior to re-immersion (10 were "lively" and 2 were "limp"), only 58% were "lively" upon recovery. The mortality rate of 17% for damaged specimens was more than double the 8% overall mortality found dur- ing re-immersion experiments for this species. Most of the dead crabs examined after re-immersion had been attacked by isopods and amphipods and only the shell remained. It is possible that these organisms were in fact responsible for killing the crabs, particularly where damage to the shell allowed access to the softer tissues of the crab. Calculations of survival rate are given in Table 9 both with and without the re-haul correction. As discussed above, the more accurate estimate of lively ^^ is probably from Table 6 because of the additional handling stress associated with the experiment. However, Table 9 also presents results obtained from data in Table 8 to estimate this probability. The results suggest that the survival rate Purves et al.: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 885 of discarded crabs would be high on the Argos Georgia, be- tween 77% and 88% for both P. formosa and P. spinosissima (77-84% for the former, 87-88% for the latter). On the Ar- gos Helena discard survival rate was much lower, 50-58% for P. formosa and 38-49% for P. spinosissima. Discussion Crab species Three previously unreported or rarely reported crab spe- cies, Paralomis anamerae, NeoUthodes diomedeae (Bene- dict, 1894), and Lithodes murrayi (Henderson, 1888), were found in our study. The most abundant of these was P. ana- merae, found at mid-range depths. The only other record of P. anamerae is the original description of the species based on four specimens obtained from the continental shelf of Argentina at depths of 132-135 m (Macpherson, 1988). The specimens obtained from South Georgia, at depths of 530-1210 m, therefore represent a considerable southerly extension in the reported geographic distribution of this species, as well as a notable increase in its bathymetric range. Lopez Abellan and Balguerias (1994) reported both P. spinosissima and P. formosa from a 1986-87 trawl survey on the shelf, but no other species of crabs. A certain amount of confusion surrounds the identifica- tion of Paralomis species around South Georgia. Paralomis aculeata is found in the CCAMLR database, but almost cer- tainly because of its inclusion in the FAO species identifica- tion guide for the Southern Ocean CArnaud 1985, in Fischer and Hureau, 1985), attributed to Henderson (1888). This species is not mentioned in Macpherson (1988), even as a junior synonym. Conversely, none of the Paralomis spe- cies identified in the present paper appear in Fischer and Hureau (1985). It is not clear, therefore, which species of Paralomis CCAMLR scientific observers have been identi- fying as P. aculeata. Twenty-two specimens ofN. diomedeae were collected at depths ranging from 420 to 1294 m. This species has previously been recorded from South Georgia (Macpherson, 1988). Sixteen L. murrayi specimens were found on southeast Shag Rocks and west South Georgia, approximately 60 nmi apart, at depths of between 450 m and 605 m, Lithodes. murrayi is mainly reported from the southern Indian Ocean around Prince Edward, Crozet, and Possession Islands, as well as Macquarie Islands, Mozambique Chan- nel, and southern New Zealand at depths of 35-200 m (Hale, 1941; Yaldwyn and Dawson, 1970; Arnauld, 1976; Arnauld and Do-Chi, 1977; Kensley, 1977). However, it has been reported in small numbers in CCAMLR statistical catch records from 1993-94, 1997-98 and 1998-99 and by CCAMLR observers (CCAMLRi). We have confirmed the identification and the extension of the range of this species to South Georgia. Klages et al. (1995) reported on the distribution of L. murrayi off Peter Island, close to the Antarctic continent between 180 and 260 m depth, and a circum-Antarctic distribution has been claimed for the species (Macpherson, 1988). The present study therefore represents the greatest depth recorded for it. Distribution Catch rates of P. spinosissima encountered in our study were lower than experienced in the September 1995-Janu- ary 1996 crab fishery by the FV American Champion. Wat- ters (1997) reported that average catch rates of legal-size male P. spinosissima were between 14.2 and 28.4 males per pot in the Shag Rocks and northwest South Georgia areas (53.5-54°S, 37-40°W). Although the toothfish pots operated by the Argos Georgia from March and May 2000 produced catch rates between 0.5 and 4 crabs/pot, the pro- portion of legal-size males was very low (3%), resulting in legal (retained) crab catch rates of less than 1/pot (note that the retention rates given in Table 2 are lower than those calculated from the length-frequency sampling. Fig. 6). Retention rates for the 1992 FV Pro Surveyor cruise (July-August 1992) were 36% (Otto and Macintosh^). The retention rates on the Argos Helena's experimental crab fishery in 1999 were much lower than this (8% and 14% for P. spinosissima and P. formosa respectively) (Purves^). The low retention rate is most likely to be a consequence of the pot design used on vessels G and H, where the col- lapsible funnel entrances might have restricted the catch to smaller size crabs. A further feature of the American Champion crab fishery was the restriction of fising effort to depths of less than 500 m. The present trials were targeted at toothfish rather than crabs and were conducted according to an experimental plan that distributed fishing effort over time, area, and the full depth range of longlines used in the main toothfish fishery. Accordingly, fishing occurred over a much wider depth range than was used by the previous crab fisheries. Our very high catch rates of P formosa in deep water were therefore not reported by Watters (1997). However, even our catch rates did not result in high numbers of retained legal-size crabs on the Argos Georgia because the propor- tion legally permitted was only 10.5%. Interestingly, even in shallow water (400-800 m) P. formosa appeared to be more common than P. spinosissima. Only in waters less than 400 m deep did P. spinosissima become the dominant species. This finding confirms the results of Watters (1997) who found that P. spinosissima catch rates declined at depths deeper than 300 m. Catch rates of P. spinosissima were low even in these depths (5 crabs/pot). Only 9 of the total of 110 sets of the Argos Georgia were conducted in depths shallower than 400 m because the main target of the fishery was toothfish. The differences found in our study in the distribution by depth of the different sexes and sizes of crabs might ■^ Otto, R. S., and R. A. Macintosh. 1992. A preliminary report on research conducted during experimental crab fishing in the Antarctic during 1992 (CCAMLR Area 48). Document WG-FSA-92/29, 20 p. CCAMLR, RO. Box 213, North Hobart, Tasmania, Australia. * Purves. M. G. 1999. Report of the South African designated CCAMLR observer on board the British registered longliner "Argos Helena" in Statistical Subarea 48.3, 31 August to 23 September 1999, 13 p. CCAMLR, RO. Box 213, North Hobart, Tasmania, Australia. 886 Fishery Bulletin 101(4) indicate that recruitment of P. formosa takes place in deeper water. This conclusion is based on a decrease in size with increasing depths, the higher proportion of fe- males encountered in deeper water, and increasing crab densities at depth. For P spinosissima this trend was not so pronounced, although crabs of this species were also gener- ally of a smaller size in deeper water However, contrary to the case with P. formosa, females were more prevalent in shallower water. Very few females of P. spinosissima were encountered in deep water These rather unusual findings might suggest ecological partitioning of the benthic habi- tat, and warrant further investigation. Another unexpected result of our work was the discovery of a third species of Paralomis, P. anamerae, at intermedi- ate depths. This species was apparently not present in the American Champion or Pro Surveyor catches, presumably again because of the depth restriction in these cruises. Maturity For P. spinosissima our estimate of 67.3 mm carapace length at 50% male maturity is similar to the 66.4 mm carapace length found by Otto and Macintosh (1996) at Shag Rocks for this species. Unfortunately, as can be seen from Figure 8, relatively few small P. formosa males were encountered and size at maturity for this species (64.0 mm carapace length) is likely to have been poorly estimated in our analysis. However, if it is assumed that male P. formosa mature at the same size in relation to female P. formosa, as in the case of P. spinosissima, the female maturity data presented in Table 5 would suggest that male P. formosa would mature at 60.4 mm carapace length (64.3 mm cara- pace width) rather than the 64.0 mm shown in Figure 8. Watters and Hobday (1998) have also examined size at maturity for P. spinosissima and P. formosa, although their samples were taken from South Georgia rather than Shag Rocks. Using a method based on finding the maximum of the second derivative of smoothing spline fits to chela height and carapace width data, they found that size at morphometric maturity for P. spinosissima was 73 mm carapace length. This size is similar to that which Otto and Macintosh (1996) obtained for P. spinosissima at South Georgia using the same technique as we did. Watters and Hobday's (1998) results for P. formosa are, however, for a higher size at maturity (80 mm carapace length) than that for P. spinosissima, which would seem to be at variance with our results and the apparent relative sizes of the two species (see Fig. 6 and CCAMLR^). The minimum size limits for crabs at South Georgia were set by CCAMLR in 1992 but, in common with many crab stocks (Schmidt and Pengilly, 1993), these measures were not accompanied by rigorous analysis of the effectiveness of these measures in meeting management objectives. For P. spinosissima. Otto and Macintosh's^ male maturity re- sults for P. spinosissima were used, and allowing males at least one opportunity to breed and an assumed growth per moult of 15%, minimum size limits were calculated as 94 mm and 84 mm carapace width at South Georgia and Shag Rocks, respectively (CCAMLR^). These results are very similar to our own, but the CCAMLR limit of 102 mm width was based on the then-existing processing require- ments rather than on these calculations. Our results sug- gest, allowing at least one opportunity to breed, that the limit should be 83 mm for P. spinosissima. For P. formosa, taking our more conservative figure of 64.0 mm carapace length at 50% maturity, the catch size limit should be set at 78 mm carapace width (the less conservative figure, 60.4 mm carapace length, would suggest a size limit of 74 mm carapace width). Hoggarth (1991) reviewed minimum size limits for a number of stocks of lithodid crabs and found that mini- mum legal sizes were about 70% of the maximum size for males, which would suggest 85 mm and 84 mm carapace width for P. spinosissima and P. formosa, respectively. It should, however, also be taken into account that these estimates were probably biased because of the greater sampling effort made at Shag Rocks. Note that the length- frequency distribution for P. formosa in Figure 6B appears to indicate a lower maximum size for males of this species than for males of P. spinosissima. However, the largest P formosa actually encountered was 120 mm carapace width. Furthermore, Figure 6B seems to be truncated at the larger sizes, suggesting perhaps that a proportion of the large adult population was not encountered during fishing. Discard mortality Our results demonstrate that, although a high proportion of crabs is likely to survive the physical strain of being hauled to the surface from potentially great depths, some under- size individuals and nontarget females can be expected to die following discarding. The most significant factor affecting discard survivorship was handling on board the vessel. On the Argos Georgia, where crabs were unloaded from pots and sorted on a conveyor belt, survivorships were high, up to 88%, and P. spinosissima survived better than P. formosa. By contrast, on the Argos Helena, where crabs went down a chute prior to processing, survival rate was between 38% and 58% and P formosa survived consider- ably better than P spinosissima. In general, P. anamerae was the most vulnerable species, followed by P spinosis- sima, and the least vulnerable — P formosa. Studies of the discard mortality of lithodid crabs in North Pacific fisheries have produced a variety of results. Stevens ( 1990) found that crabs discarded from commercial sole trawls suffered high mortalities (47.3%), but Byersdor- fer and Watson'' and Zhou and Shirley ( 1995) both reported relatively low mortalities (<2%) resulting from handling when fishing with pots. Our results support these previous studies and extend them to the Antarctic, clearly indicating that where handling on a pot vessel is reduced, mortali- ties are relatively low (<15% mortality). When crabs are ' Byersdorfer, S., and L. J. Watson. 1992. A summary of bio- logical data collected during the 1991 Bristol Bay red king crab tagging study. Technical Fishery Report 92-14, 30 p. Alaska Department of Fish and Game, Division of Commercial Fisheries, P.O. Box 25526, Juneau, AK 99802-5526. Purves et al.: Distribution, demography, and discard mortality of crabs caught as bycatch in the South Atlantic 887 handled on a pot vessel, as they would be on a trawl vessel, mortalities are higher. Other factors, which could not be tested in the re-immersion experiment, may also affect the rate of crab survival. We re-immersed crabs in pots, whereas normally they would be simply dropped into the sea and would be subject to predation from birds and fish before they reached the bottom. The effect of discarding crabs away from their original habitat is unknown, but our results demonstrate a clear depth separation between the two species; therefore one would expect at least an energetic cost if crabs have to relocate. The crabs subjected to re-immersion experiments were sampled imme- diately before being discarded. They might suffer further damage through the actual discard process; for instance Stevens (1990) speculated that, while traveling through offal chutes, they could become entangled in machinery or suffer further damage upon impact with the surface of the water Ide- ally crabs should have been sampled after being through the full discarding procedure, but this was not practical. Finally, eggs often became dislodged during handling and this loss possibly impacted reproductive success. Zhou and Shirley (1995) presented results that indi- cate that there are no long-term effects of handling on crab survival, feeding rate, or crab condition; therefore we might reasonably expect that the survival rates seen in our experiments would also be the relevant long-term survival rates. However, even with relatively low discard mortality, the impacts of repeated catching and discarding of individuals will have a cumulative effect on crab popula- tions. Both retained and discarded bycatch should there- fore continue to be reported and be incorporated into crab population models. The presence of such a large discarded bycatch might provide the opportunity for the retention (and removal from the population) of parasitized crabs, as suggested by Basson (1994). Crabs are an inconvenience in a fishery targeting tooth- fish. In situ observations made during the AUDOS ex- periments on the UK's January 2000 survey confirm that toothfish seek to avoid direct contact with crabs (Yau et al., 2002), although crabs do form a component of their food (Pilling et al., 2001). An inverse relationship was found in the present study between toothfish numbers in pots and crab numbers in pots, suggesting toothfish avoid pots with large crab populations (Fig. 9). Therefore, conducting the toothfish pot fishery in an area of low crab abundance is sensible, and our data do suggest that, at intermedi- ate depths, the crab catch should be low and composed primarily of large P. formosa. Avoidance of areas of high crab bycatch will also reduce the mortality associated with discarding female and undersize male crabs. These discard levels are very high (>93%) — considerably higher than those in the Bering Sea (85%: Stevens, 1996). Such high discard levels could be reduced further by developing new pot designs to limit crab catches to larger, legal-size ani- mals— for instance designs with excluder panels (Stevens, 1996) — or perhaps by reducing the minimum size limit. 100 A 80 \ 60 \ 40 - \ 20 - n ■. ■iW^ 14 12 ro u 10 o D Frequency of number/pot toothfish ■ Number/pot P spinosissima 0-1 1-2 2-3 CPUE (number/pot) of toothfish Figure 9 The frequency distribution of the number of toothfish caught per pot compared with the CPUE's of Paralomis spinosissima and P. formosa (numbers/pot) and of toothfish {Dissostichus eleginoides) catches (numbers/pot), as found during the first cruise of the Argos Georgia during March to May 2000. Acknowledgments We wish to acknowledge the assistance of Argos Ltd. and the excellent cooperation of the captains of the two ves- sels. Captain Joaquin Abraldes Gonzalez and Captain Jose Andres Sampedro and their crew. Permission to publish these data was kindly granted by Argos Ltd. The Govern- ment of South Georgia and the South Sandwich Islands funded the research by MRAG Ltd. on South Georgia fisheries. Literature citations Agnew, D. J., T. M. Daw, M. G. Purves, and G. M. Pilling. 2001. Fishingfor toothfish using pots: results of trials under- taken around South Georgia, March-May 2000. CCAMLR (Commission for the Conservation of Antartic Marine Living Resources) Sci, 8:93-105. Agnew, D. J., L. Heaps, C. Jones, A. Watson, K. Berkieta, and J. Pearce. 1999. Depth distribution and spawning pattern of Dissosti- chus eleginoides at South Georgia. CCAMLR Sci. 6:19-36. Arnauld, P. M. 1976. Peches experimentales de Lithodes murrayi Hender- son, 1888 (Crustacea, Anomura) aux iles Crozet (SW de I'Oceanlndien). Thetys 3:167-172. Arnauld, R M., and T. Do-Chi. 1977. Donnees biologiques et biometriques sur les lithodes Lithodes murrayi (Crustacea: Decapoda: Anomura) des lies Crozet (SW de I'Ocean Indien). Mar Biol. 39:147-159. Basson, M. 1994. A preliminary investigation of the possible effects of rhizocephalan parasitism on the management of the crab fishery around South Georgia. CCAMLR Sci. 1:175-192. Benedict, J. E. 1894. Descriptions of new genera and species of crabs of the family Lithodidae, with notes on the young oi Lithodes 888 Fishery Bulletin 101(4) camtschaticus and Lithodes brevipes. Proc. U.S. Natl. Mus. 17:479-488. Fischer, W., and J. C. Hureau (eds.) 1985. FAO species identification sheets for fishery purposes, Southern Ocean, vol. l.,p. 232. FAO (Food and Agriculture Organization of the United Nations), Rome, Italy. Hale, H. M. 1941. Decapod Crustacea. Br. Aust. N.Z. Antarct. Res. Exped. 1929-1931. 48:257-285. Henderson, J. R. 1888. Report on the Anomura collected by H.M.S. Challenger during the years 1873-76. Rep. Sci. Res. Voy H.M.S. Chal- lenger, Zool. 27(I-vii): 1-221, plates 1-21. Hoggarth, D. D. 1991. An ecological and economic assessment of the Falk- land Islands inshore crab, Paralomis granulosa. Ph.D. diss., 312 p. Imperial College, Univ. London, London. UK. Kensley, B. 1977. The South African Museum's Meiring Naude cruises. Part 2. Crustacea, Decapoda, Anomura and Brachyura. Ann. S. Afr. Mus. 72:161-188. Klages, M., J. Gutt, A. Starmans, and T. Bruns. 1995. Stone crabs close to the Antarctic continent: Lithodes murrayi Henderson, 1888 (Crustacea; Decapoda; Ano- mura) off Peter I Island (68°51-S, 90°51-W). Polar Biol. 15:73-75. Lopez-Abellan, L, J., and E. Balguerias. 1994. On the presence of Paralomis spinosissima and Paralomis formosa in catches taken during the Spanish survey Antartida 8611. CCAMLR Sci. 1:165-173. Macpherson, E. 1988. Revision of the family Lithodidae Samouelle, 1819 (Crustacea, Decapoda, Anomura) in the Atlantic Ocean. Mongr. Zool. Mar 2:9-153. Otto, R. S., and R. A. Macintosh. 1996. Observations on the biology of the Lithodid crab Paralomis spinosissima from the Southern Ocean near South Georgia. In Proceedings of the international sympo- sium on biology, management and economics of crabs from high latitude waters. Anchorage, Alaska, October 1995, p. 627-647. Alaska Sea Grant, Fairbanks, AK. Pilling, G. M., M. G. Purves, T. M. Daw, D. J. Agnew, and J. C. Xavier. 2001. The stomach contents of Patagonian toothfish around South Georgia (South Atlantic). J. Fish Biol. 59: 1370-1384. Stevens, B. G. 1990. Survival of king and tanner crabs captured by com- mercial sole trawls. Fish. Bull. 88:731-744. 1996. Crab bycatch in pot fisheries: causes and solutions. In Proceedings of the solving bycatch workshop, September 25-27, 1995, Seattle, Washington (TWray,ed.), p. 151-158. Alaska Sea Grant, Fairbanks, AK. Schmidt, D. C, and D. Pengilly 1993. Review of harvest strategies used in the management of lithodid crab in Alaska. In Proceedings of the international symposium on management strategies for exploited fish pop- ulations: October 21-24, 1992, Anchorage, Alaska (G. Kruse, D. M. Eggers, R. J. Marasco, C. Pautzke, and T. J. Quinn, eds.), p. 385-407. Alaska Sea Grant, Fairbanks, AK. Somerton, D. A., and R. S. Otto. 1986. Distribution and reproductive biology of the golden king crab, Lithodes aequispina, in the eastern Bering Sea. Fish. Bull. 84:571-584. Watters, G. 1997. Preliminary analyses of data collected during experi- mental phases of the 1994/95 and 1995/96 antarctic crab fishing seasons. CCAMLR Sci. 4:141-159. Watters, G., and A. J. Hobday. 1998. A new method for estimating the morphometric size at maturity of crabs. Can. J. Fish. Aquat. Sci. 55:704-714. Yaldwyn, J. C, and E. W. Dawson. 1970. The stone crab Lithodes murrayi Henderson: the first New Zealand record. Rec. Dom. Mus. (Wellingt.) 6: 275-284. Yau, C, M. A. Collins, P M. Bagley, I. Everson, and I. G. Priede. 2002. Scavenging by megabenthos and demersal fish on the South Georgia slope. Antarct. Sci. 14:16-24. Zhou, S., and T C.Shirley 1995. Effects of handling on feeding, activity and survival of red king crabs Paralithodes camtschaticus (Tilesius, 1815). J. Shellfish Res. 14:173-177. 889 Abstract— Sea turtles are subjected to involuntary submergence and potential mortality due to incidental capture by the commercial shrimp fishing indus- try. Despite implementation of turtle excluder devices (TEDs) to reduce at- sea mortality, dead stranded turtles continue to be found in near-record numbers along the coasts of the west- ern Atlantic Ocean and northern Gulf of Mexico. Although this mortality may be due to an increase in the number of turtles available to strand, one alterna- tive explanation is that sea turtles are repetitively submerged (as one fishing vessel follows the path of another) in legal TEDs. In the present study, labo- ratory and field investigations were undertaken to examine the physiologi- cal effects of multiple submergence of loggerhead sea turtles iCaretta caretta ). Turtles in the laboratory study were confined during the submersion epi- sodes, whereas under field conditions, turtles were released directly into TED-equipped commercial fishing nets. Under laboratory and field condi- tions, pre- and postsubmergence blood samples were collected from turtles submerged three times at 7.5 min per episode with an in-water rest interval of 10, 42, or 180 min between submer- gences. Analyses of pre- and postsub- mergence blood samples revealed that the initial submergence produced a severe and pronounced metabolic and respiratory acidosis in all turtles. Suc- cessive submergences produced sig- nificant changes in blood pH, Pco„ and lactate, although the magnitude of the acid-base imbalance was substantially reduced as the number of submergences increased. In addition, increasing the interval between successive submer- gences permitted greater recovery of blood homeostasis. No turtles died during these studies. Taken together, these data suggest that repetitive sub- mergence of sea turtles in TEDs would not significantly affect their survival potential provided that the animal has an adequate rest interval at the surface between successive submergences. The physiological effects of multiple forced submergences in loggerhead sea turtles (.Caretta caretta) Erich K. Stabenau Kimberly R. N. Vietti Department of Biology Bradley University 1501 W Bradley Ave. Peoria, Illinois 61625 E-mail address (for E, K. Stabenau): eks@bradleyedu Manuscript approved for publication 25 March 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:889-899 (2003). The five sea turtle species inhabiting the waters of the U.S. Gulf of Mexico and Atlantic Ocean are considered to be threatened or endangered. One con- tributing factor to sea turtle mortality is incidental capture in the nets of com- mercial shrimping vessels. The National Research Council's Committee on Sea Turtle Conservation (1990) suggested that as many as 5500 to 55,000 log- gerhead (Caretta caretta) and Kemp's ridley (Lepidochelys kempi) sea turtles were killed annually during shrimp- ing-related activities. More recently, two independent studies statistically confirmed the relationship between shrimping activity and the appearance of stranded sea turtles in the U.S. Gulf of Mexico and the Atlantic Ocean (Cail- louet et al., 1991; Crowder et al., 1995). Because of the impact of trawl-related mortality on sea turtle populations, the U.S. government passed regulations in 1987 requiring that commercial shrimping vessels pull nets equipped with certified turtle excluder devices (TEDs). TEDs are designed to exclude any turtle that may enter into shrimp- ing nets, while not affecting the catch of the target species. Crowder et al. (1995) reported that the sea turtle population off the coast of South Carolina contin- ued to decline when TED regulations were implemented; however, the rate of decline decreased significantly after full-time TED use. In spite of the TED regulations, near-record numbers of dead stranded sea turtles have been found on U.S. Gulf of Mexico and Atlantic Ocean beaches (Shaver-Miller'). Although there may be other man-related or natural causes for this continued sea turtle mortality, there are two plausible reasons for the increased mortality during shrimping activities. First, commercial shrimp fishermen generally do not carry le- gally certified TEDs in their trawl nets and the TEDs that are used are often installed incorrectly or purposely sewn shut. Second, the shrimp fisher- men may pull legal TEDs; however, the turtles are repetitively submerged as they are caught in the TEDs of ves- sels that follow each other These suc- cessive submergences may exacerbate the physiological effects experienced by sea turtles during a forced submer- sion, and thus, may limit their survival potential. Sea turtles spend approximately 99% of their time under the surface of the water. During the brief period at the surface, the turtle will exhale and inhale a solitary breath and then dive under the surface (Jackson, 1985). In fact, multiple breaths by sea turtles are generally seen only after prolonged dives. Minimal information is available on the physiological effects of forced submergences of sea turtles. It has been suggested that voluntary dives by sea turtles are aerobic in nature (Wood et al., 1984), whereby oxygen availability minimizes the metabolic production of lactic acid. The turtles may accumulate carbon dioxide, resulting in a respira- Shaver-Miller, D. 2002. Personal com- mun. Texas coordinator, Sea Turtle Stranding and Salvage Network, USGS, Corpus Christi, Texas 78406. 890 Fishery Bulletin 101(4) tory acidosis that is ameliorated by hyperventilation at the surface. Therefore, voluntary diving in the absence of any other external stressor does not limit sea turtle survival potential. In contrast, forced submergence of Kemp's ridley and log- gerhead sea turtles produces significant blood respiratory and metabolic derangements. Stabenau et al. ( 1991) report- ed that forced submergence of Kemp's ridley sea turtles for less than 7.5 min in shrimp nets equipped with TEDs re- sulted in significant increases in blood lactic acid and PcOj, and decreases in blood pH. Moreover, several hours were required for these turtles to fully recover blood homeostasis (National Marine Fisheries Service, unpubl. data^). How- ever, the study by Stabenau et al. ( 1991) did not address the physiological effects of multiple forced submergences of sea turtles. It is plausible that repeated submergence induces progressive, significant blood acid-base disturbances, and limits sea turtle survival potential. Therefore, the present study examined the physiological effects of multiple forced submergences on loggerhead sea turtles. This investigation was divided into two phases. First, a laboratory component was conducted to examine the feasibility of a multiple submergence study. This phase of the research permitted characterization of the magnitude of the acid-base disturbance under controlled conditions. Second, a field investigation was conducted to expose tur- tles to TED-equipped commercial fishing nets. Data from these studies may offer greater insight into potential sea turtle mortality caused by multiple capture in commercial shrimping nets carrying legal TEDs. Materials and methods Laboratory study Thirty-nine headstarted 2-year-old loggerhead sea turtles reared in captivity at the National Marine Fisheries Ser- vice (NMFS) Galveston Laboratory were used in this phase of the study. Each turtle was randomly placed into experi- mental (submerged, 37.0 ±0.2 cm, 6.51 ±0.06 kg, n=21) or control (nonsubmerged, 36.9 ±0.2 cm, 6.45 ±0.10 kg, n=18) treatments. All turtles were of comparable size and weight and therefore any alterations in blood parameters between experimental and control turtles represented treatment effects rather than size effects. It should be noted that the turtles used in our study were representative of the aver- age size of dead stranded turtles and those animals used in annual TED certification tests. The study was initiated by collecting presubmergence blood samples from the experimental turtles immediately prior to their individual confinement in a weighted canvas bag. Each turtle was then submerged for 7.5 min in sea- water filled tanks. Postsubmergence blood samples were collected within 30 s of bringing the turtle out of the water to minimize blood acid-base changes. Following an in-water 2 National Marine Fisheries Service. 1994. Unpubl. data. (Available from E. K. Stabenau, Bradley University, 1501 W. Bradley Ave., Peoria, IL 61625] rest interval of 10 (treatment 1), 42 (treatment 2), or 180 (treatment 3) min, a presubmergence blood sample was collected and the turtle was submerged a second time. A postsubmergence blood sample was then collected imme- diately upon surfacing. The turtle was then submerged a third time, following the same rest interval between the first and second submergence episodes, and pre- and post- submergence blood samples were collected as described above. The seventh serial blood sample was collected 180 min after the final submergence in all turtles. Blood sam- ples were also collected from control turtles over the same time intervals to ensure that repetitive handling and blood sampling did not alter blood homeostasis. All blood samples were collected into heparinized vacutainers from the dorsal cervical sinus as described by Owens and Ruiz (1980). No more than 4-6% of blood volume was collected during the serial sampling to minimize potential physiological effects associated with blood volume depletion. Field study Thirty-six headstarted 2-year-old loggerhead sea turtles reared in captivity from the NMFS Galveston Laboratory were used in this phase of the study. The turtles were trans- ported from Galveston, TX, to Panama City, FL, where they were placed into large pens in St. Andrews Bay. The sub- mergence study was initiated after a minimum of 21 days of natural conditioning in the in-water pens. Each turtle was randomly placed into experimental (submerged, 35.9 ±0.2 cm, 6.77 ±0.09 kg, n=24) or control (nonsubmerged, 35.4 ±0.3 cm, 6.46 ±0.12 kg, n=12) treatments. As in the laboratory study, all experimental and control turtles were of comparable size and weight. The study was initiated by collecting presubmergence blood samples from the experimental turtles immediately prior to their individual confinement in a weighted mesh bag. Each turtle was then submerged using the standard protocol for TED certification tests. Briefly, the mesh bag containing a turtle was placed onto a line connecting the trawl vessel to the headrope on the shrimp net. Divers then released the turtle (without handling the animal) into the mouth of the trawl. Often, turtles were observed vigorously swimming in the trawl until being overcome by the net. Although the shrimp net was equipped with a TED, divers held the escape door closed for 5 min. The turtle was then permitted to leave the trawl and surface. Thus, the total submergence time was approximately 7.5 min, including the time for the weighted mesh bag containing the turtle to reach the headrope for release into the trawl, the 5 min within the trawl, and the time for the turtle to surface. Turtles were immediately captured at the surface and returned to the trawl vessel for postsubmergence blood sampling. Typically, postsubmergence blood samples were collected within 1-2 min of the turtle surfacing. Following a rest interval of 10 (treatment 4), 42 (treatment 5), or 180 (treatment 6) min in water-filled containers on the trawl vessel, a presubmergence blood sample was collected and the turtle was submerged a second time. A postsubmer- gence blood sample was then collected immediately upon surfacing. The turtle was then submerged a third time, Stabenau and Vietti: The physiological effects of multiple forced submergences of Caretta caretta 891 following the same rest interval between the first and second submergence episodes, and pre- and postsubmer- gence blood samples were collected as described above. A seventh serial blood sample was collected 180 min after the final submergence in all turtles. Blood samples were also collected from nonsubmerged control turtles over the same time intervals to ensure that repetitive handling and blood sampling did not alter blood homeostasis. The blood sampling technique and volume collected was identical to that described for the laboratory component of the study. Blood and plasma analyses In the laboratory study, blood PcOj and pH were analyzed immediately following collection by using a clinical blood gas analyzer with electrodes thermostatted at 37°C. Both variables were corrected to turtle cloacal temperature using requisite correction factors for sea turtle blood and plasma (Stabenau and Heming, 1994). In the field study, blood gases (Po., and PcOr,) and pH were analyzed on the trawl vessel immediately following collection using a blood gas analyzer with electrodes thermostatted to turtle body temperature (27-28. 5°C). The remaining analyses were comparable for both the laboratory and field compo- nents of the submergence study. Packed red cell volume (hematocrit) was determined by following centrifugation of heparinized microcapillary tubes. Two hundred micro- liters of whole blood were then added to 10% trichloro- acetic acid for lactate analysis. The deproteinized samples were centrifuged, and the supernatant removed and stored at -70°C. Lactate was determined spectrophotometrically by using standard enzymatic techniques (Sigma, kit 826- B, Saint Louis, MO). The remaining whole blood was then centrifuged, the plasma removed and stored at -70°C. Plasma Na+ and K* were measured with flame photometry ( Jenway, model PFP7, Essex, England), and plasma CL was determined with electrometric titration (Haake-Bucher, model 4425000, Saddle Brook, NJ). Plasma glucose was measured spectrophotometrically (Sigma, kit 16-20), and plasma osmolality was determined with a vapor pressure osmometer (Wescor, model 5500, Logan, UT). For the labo- ratory study, plasma norepinephrine was analyzed with HPLC (BAS, model LC-300, West Lafayette, IN). All data are expressed as means ±SE. Where appropri- ate, the data was analyzed with one-way ANOVA. Post-hoc comparisons between means were analyzed with Tukey's multiple comparison test. A fiduciary level of P<0.05 was regarded as significant. Results Blood pH, PcOj, and lactate The initial submergence of loggerhead sea turtles under laboratory and field conditions produced a dramatic and severe acidosis in all experimental turtles. Blood pH fell an average of 0.54 ±0.03 (range 0.49 to 0.59 pH units) and 0.63 ±0.06 (range 0.53 to 0.73 pH units) m labora- tory turtles and field turtles, respectively, following ini- tial submergence (Figs. lA and 2A). The blood acidosis was derived from respiratory and metabolic components as evident from a positive proton-lactate deficit (Buffer capacityx4pH-/l[Iactatel), and from significant increases in blood PcOj and lactate (Figs. 1 and 2). The initial sub- mergence also produced significant decreases in blood Pog and increases in plasma norepinephine (P<0.05, n=24 for Po., and n = ll for norepinephrine). In contrast, minimal changes in blood pH, PcOg, and lactate were observed fol- lowing collection of the first two blood samples in nonsub- merged control turtles (Figs. 1 and 2). Recovery of the respiratory and metabolic derangements in submerged turtles was dependent on the interval be- tween successive submergences. A 10-min in-water rest interval between the first and second submergence (treat- ment-! and -4 turtles) permitted partial recovery of blood pH (Figs. lA and 2A) and Pco,, (Figs. IB and 2B), but blood pH remained significantly different from presubmergence values. Washout of additional lactate was also detected in these animals, whereby blood lactate concentration in- creased higher than the postsubmergence value (Figs. IC and 2C ). Turtles with a 42-min surface interval (treatment- 2 and -5 turtles) between the first and second submergence had partial to complete recovery of blood pH (Figs. lA and 2A), complete recovery of blood Pco.^ (Figs. IB and 2B), and slight recovery of blood lactate (Figs. IC and 2C). Only the blood lactate remained significantly different from the initial presubmergence value after the 42-min rest in- terval. Turtles with a 180-min in-water recovery interval (treatment-3 and -6 turtles) showed complete recovery of blood pH and PcOr,, although the lactate concentration was slightly higher than baseline levels (Figs. 1 and 2). Blood Pog and plasma norepinephrine recovered completely re- gardless of the surface interval (P>0.05, n=24 and n = ll for Poj and norepinephrine, respectively). Nonsubmerged control turtles in the laboratory and the field exhibited few significant changes in blood pH, Pcoo, or lactate, whether the interval between the second and third serial blood sample was 10, 42, or 180 min (Figs. 1 and 2). The second 7.5-min submergence produced a drop in blood pH and an increase in Pco,^ (Figs. 1 and 2) in all of the experimental animals, and significant differences occurred in treatment 2-6 turtles. It is noteworthy, however, that the severity of the acid-base imbalance was not as drastic as the acidosis measured following the first submergence. The mean pH difference (4pH) between the second pre- and postsubmergence ranged from 0.11 and 0.16 in treatment-1 and -4 turtles (animals with a 10-min interval between submergences), respectively, to 0.50 in treatment-3 turtles and 0.66 in treatment-6 turtles (animals with a 180-min interval between submergences). The acidosis in treat- ment-1 and -4 turtles resulted, in part, from the continual elevation in blood lactate. In contrast, the longer surface interval between the two submergence episodes resulted in enhanced recovery of acid-base variables. Therefore, the turtles with a surface interval of 42 or 180 min had in- creased production of CO2 and lactate in relation to turtles with a brief surface interval (Figs. 1 and 2). Comparable changes in blood Poj and norepinephrine were measured following the second submergence (P<0.05, n=24 and n=9 892 Fishery Bulletin 101(4) Blood collection Figure 1 Blood pH (A), PcOj (B), and lactate (C) measured prior to and after three succes- sive forced submergence episodes in loggerhead sea turtles in the laboratory. Blood collection 1, 3, and 5 are presubmergence samples, whereas blood collection 2, 4, and 6 are postsubmergence samples. Blood collection 7 was taken 180 min after the final submergence. The surface interval between the submergences was 10 min (T), 42 min (•), or 180 min (■). Data from control sea turtles (A) are shown for comparison. for Po.2 and norepinephrine, respectively). Collection of the fourth sample from nonsubmerged control turtles revealed no significant changes in blood pH, PCO2, or lactate when compared to the third sample (Figs. 1 and 2). The remaining serial blood samples revealed comparable patterns in the blood pH, Pco.,, Po,, lactate, and norepi- nephrine. Turtles given a longer rest interval at the surface (after the second submergence^ had enhanced recovery of Stabenau and Vietti The physiological effects of multiple forced submergences of Caretto caretta 893 Blood collection Figure 2 Blood pH (A), PcOj (B), and lactate (C) measured prior to and afler three succes- sive forced submergence episodes in loggerhead sea turtles in TED-equipped nets. Blood collection 1, 3, and 5 are presubmergence samples, whereas blood collection 2, 4, and 6 are postsubmergence samples. Blood collection 7 was taken 180 min after the final submergence. The surface interval between the submergences was 10 min (T), 42 min (•), or 180 min (■). Data from control turtles (A) are shown for comparison. blood acid-base variables, whereas a brief surface interval permitted minimal recovery of blood homeostasis (Figs. 1 and 2). Submersion of experimental turtles a third time resulted in similar changes in blood pH, PCO2, and lactate to that measured following the second submersion, and the length of the at-surface rest interval affected the mag- nitude of recovery of blood acid-base status. The seventh serial sample collected 180 min after the final postsubmer- 894 Fishery Bulletin 101(4) Table 1 Mean (±SE) plasma Na*, K*, and plasma osmotic pressure (OP) prior to an of sea turtles with a 10-min, 42-min, or 180-min rest interval. Serial blood methods" section. Significant differences between samples 1 and 2, 3 and 4, significant differences of samples from the initial blood sample (serial sampl d following laboratory multiple forced submergences sampling regime is described in the "Materials and and 5 and 6 are indicated by an asterisk (*), whereas e 1) are denoted by a pound sign (#). Treatment 10 min 42 min 180 min Na+ (mM) (mM) ( OP mosm/kg) Na* (mM) (mM) OP (mosm/kg) Na* (mM) (mM) OP (mosm/kg) Control 153 ±3 4.0 ±0.2 319 ±3 152 ±4 3.9 ±0.3 305 ±2 158 ±4 4.3 ±0.4 322 ±4 Serial sample 1 144 ±5 4.5 ±0.3 319 ±6 158 ±6 3.9 ±0.2 314 ±11 162 ±2 4.1 ±0.1 296 ±3 2 159 ±6 5.9 ±0.6 341 ±4 163 ±3 6.1 ±0.6" 364 ±10** 187 ±2*« 6.9±0.3*« 342 ±5*" 3 145 ±3 4.9 ±0.2 330 ±5 156 ±2 4.1 ±0.3 336 ±8 160 ±4 4.4 ±0.1 308 ±4 4 166 ±7« 6.2 ±0.3" 351 ±14" 160 ±6 5.5 ±0.2" 342 ±15 179 ±4 6.7±0.5*» 339 ±4*» 5 158 ±6 5.1 ±0.1 335 ±11 147 ±6 4.1 ±0.3 334 ±12 158 ±6 3.9 ±0.2 305 ±5 6 154 ±5 6.1 ±0.3« 340 ±12 157 ±9 4.8 ±0.3 345±12« 181 ±2 5.7 ±0.5* 323 ±8" 7 139 ±3 4.8 ±0.4 323 ±8 149 ±6 4.4 ±0.3 331 ±7 158 ±10 4.4 ±0.5 305 ±3 gence sample revealed that blood pH, Pcog, and lactate recovered completely for all experimental turtles (Figs. 1 and 2). Minimal changes in blood pH, PcOj and lactate were detected in laboratory and field control turtles during col- lection of the 5-7 serial blood samples (Figs. 1 and 2). Ions, glucose, and osmotic pressure Postsubmergence blood samples from laboratory turtles revealed elevations in plasma Na*, K+, and osmotic pres- sure when compared to the corresponding presubmergence values (Table 1). Significant increases in the plasma Na*, K+, and osmotic pressure were observed more frequently in turtles with a longer in-water rest interval between suc- cessive submergences (Table 1). In contrast, the plasma ion concentrations and osmotic pressure of control turtles did not substantially change (P>0.05, n=9) during serial blood sample collection. In addition, no significant differences in plasma glucose and CI" (P>0.05, n = 10) were measured in any of the experimental turtles. Although most of the postsubmergence changes in the blood parameters in experimental turtles were not significant (Table 1), and minimal alterations in blood chemistry were observed in control turtles, the results suggested that there was a relationship between blood acid-base status and plasma osmolality and ion concentration. Therefore, correlation analyses were used to determine the interdependence of these variables. Figure 3 shows the results of the correlation analyses, where pH is plotted versus ion concentration (i.e. Na*, K*, and CI" concentration), osmolality, or hematocrit. Nonsubmerged control turtles had a significant correla- tion between blood pH and plasma chloride, and pH and hematocrit (Fig. 3). As pH declined, there were slight, yet significant, increases in the [CI"] and hematocrit. How- ever, no correlation was detected between pH and plasma [Na*], [K+], or osmolality in these animals. In contrast, a significant correlation was detected between blood pH and plasma [Na*], [K+], [CI"], osmolality, and hematocrit in experimentally submerged turtles (Fig. 3). In each case, a decrease in blood pH led to an increase in the correlated variable. These data are consistent with significant water movement into and out of the red blood cells during and after forced submersion. Brief forced submergence of loggerhead turtles in trawl- equipped fishing nets had a profound effect on the plasma ionic status (Table 2). Plasma [K*] increased significantly immediately following submergence in all experimen- tal turtles. Significant increases were also observed in the plasma [Na*] and osmotic pressure, although these changes did not occur in turtles from all of the experimen- tal treatments (Table 2). Turtles partially to completely recovered from the ionic imbalances, although subsequent submergences caused significant increases in plasma K"^ and nonsignificant increases in plasma Na* and osmolality in most experimental turtles (Table 2). Ionic homeostasis in forcibly submerged turtles was achieved within 180 min of the final submergence, whereby plasma ion concentra- tions were comparable to the initial presubmergence val- ues (Table 2). The plasma ion concentrations and osmotic pressure in nonsubmerged control turtles were unaffected by serial blood sampling. Thus, ionic changes in experimen- tal turtles resulted from the forced submergence and not from handling and repetitive blood sampling. Discussion Acid-Base status Multiple submergences of 2-year-old loggerhead sea tur- tles under laboratory and field conditions produced sig- Stabenau and Vietti: The physiological effects of multiple forced submergences of Caretta carelta 895 7Sn Control Treatment 7 6- 7 4- •'i'iH.; = ' 7 2- 7 0- 66- r^<00001 P=0 9717 7 76- 74- 7 2 7 0- 68 Submergence Treatment r =0 1965 P=0 0001 [Na*]. mM r' =00165 P=0,3147 7 8' 7 6- 74 72- 70- 68 66 7 6- 74- 72- 7 0 6 8-] r''=0 0640 P=0,0454 78 76 74 72 70- 6 8-1 r'=0 6301 P<0 0001 SV. [K*], mM 7 8-| 7 6- 74- \ v^^V. 72- 70- . a • .\ fi«- r' =0 1946 •• ...X P= =0,0002 . 66- , , [Cr], mM r=0.0434 P=0 1014 ■ :c^^^ ■ " B . r^=0 4724 •'.. V P<0 0001 ■ Osmolality, mosmoles/kg I Q. •^ft*^. r=0 0975 P=0 0123 7.6 7.4' 7.2 7.0 68 66 r=0 2968 P<0 0001 Hematocrit (%) Figure 3 Relationship between blood pH and plasma |Na*|, (K*], [C!"], osmolality, and hematocrit in control (left column) and submerged (right column) loggerhead sea turtles. The lines are best fits to the data. Significance of the correlated variables is noted on each figure. 896 Fishery Bulletin 101(4) Table 2 Mean l±SE) plasma Na*, K*, and plasma osmotic pressure (OP) prior to and following multiple forced submergences of sea turtles in TED-equipped nets with a 10-min, 42-min, or 180-min rest interval. Serial blood sampling regime is described in the "Materials and methods" section. Significant differences between samples 1 and 2, 3 and 4, and 5 and 6 are indicated by an asterisk (*), whereas significant differences of samples from the initial blood sample (serial sample 1) are denoted by a pound sign (#). Treatment 10 min 42 min 180 min Na* (mM) K* (mM) OP (mosm/kg) Na* (mM) K* (mM) OP (mosm/kg) Na* (mM) K* (mM) OP (mosm/kg) Control 150+3 3.0 ±0.2 313 ±8 139 ±6 3.4 ±0.2 321 ±7 151 ±1 3.1 ±0.1 310+5 Serial sample 1 153 ±2 3,3 ±0.3 318 ±4 160 ±4 3.1 ±0.2 331 ±12 164 ±2 4.5 ±0.7 325 ±9 2 171 ±8 5.5 ±0.3*» 345 ±4*» 186 ±8* 5.0 ±0.4*» 368 ±10 188 ±4 7.0 ±0.6*" 355 ±3*» 3 156+6 4.3 ±0.0" 332 ±4 163 ±3 2.8 ±0.1 338 ±11 163 ±10 3.6 ±0.3 314 ±3 4 171 ±8 5.3±0.1*» 349 ±1" 181 ±3 4.9 ±0.3*'* 361 ±13 176+10 6.2+0.3*» 352 ±9* 5 166 ±4 4.3 ±0.1" 334 ±2 160 ±8 2.9 ±0.2 332 ±9 173 ±10 4.0 ±0.2 323 ±3 6 166 ±12 5.1±0.1« 335 ±11 185 ±4* 4.5 ±0.4*« 343 ±14 175 ±18 5.3 ±0.0 333 ±11 7 157 ±4 3.7 ±0.1 325 ±2 161 ±6 2.6 ±0.2 326 ±9 159 ±11 3.6 ±0.6 320 ±4 nificant blood metabolic and respiratory disturbances. The most dramatic changes in blood pH, Pco.^, and lactate occurred following the first of the three forced submer- gences in all of the experimental turtles (Table 3). Under laboratory conditions, the turtles exhibited an average pH change of 0.54 U following the first submergence, whereas initial submergence of 2-year-old loggerhead sea turtles in TED-equipped commercial fishing nets induced a pH decrease of 0.63 U. The initial acid-base disturbances mea- sured in our study were comparable in magnitude to those measured in Kemp's ridley and loggerhead sea turtles in standard TED certification trials (Table 3). The second and third submergences of 2-year-old log- gerheads sea turtles did not result in similar changes in blood pH, PcOj, and lactate, as was measured following the initial submergence (Table 3). To our knowledge, no infor- mation is available in the literature on the physiological effects of multiple submergences in sea turtles for compari- son. Obviously, the interval between the submergence epi- sodes directly influenced the magnitude of the blood acid- base imbalance during successive submergences. A longer time interval at the surface led to enhanced recovery of blood pH, Pcog, and lactate. Lutz and Dunbar-Cooper ( 1987) reported that loggerhead sea turtles captured during trawl- ing at Cape Canaveral, Florida, exhibited a 16.8% decline in lactate 180 min following submergence. Those authors proposed that the rate of lactate decline was dependent on the magnitude of the lactate concentration, so that 10 mM of lactate would decline at a rate of 1.25 niM lactate/h. However, in the present study, the rate of lactate decline was considerably higher than that suggested by Lutz and Dunbar-Cooper (1987). Lactate declined 70.0% and 79.6% within 180 min of the submergence episodes in treatment 3 turtles, whereas no decline was measured in treatment 1 turtles (10 min interval) between submergences. In fact, it was apparent that lactate continued to washout into the bloodstream during the 10-min recovery phases in these turtles (Fig. 1, Table 3). Thus, turtles with a brief period between the submergence episodes would have a limited ability to release the CO2 retained during submersion or to break down lactic acid produced during the course of the forced dive. Lactate declined 15.2%- and 18.7% during the 42-min interval between submergences in treatment-2 turtles. Blood lactate declined 80.9%, 76.0%, and 82.5% in treatment-1, -2, and -3 turtles, respectively, during the final 180-min recovery period. Thus, the overall rate of lactate decline in the final 180 minutes of the laboratory study was 2.6 ±0.2 mM/h. Finally, the elevated lactate concentration in sea turtles during the 180-min postsubmergence recov- ery time interval suggests that the samples were collected too soon to permit complete recovery of blood lactate. Comparable rates of lactate clearance measured in the laboratory submergence study were detected following forced submergences of loggerhead sea turtles in TED- equipped fishing nets. Substantial retention of CO, and additional washout of lactate occurred during the 10-min postsubmergence recovery interval in treatment-4 turtles. Treatment-5 turtles exhibited a 6% drop in the blood lactate concentration during the first 42-min postsubmer- gence recovery interval and a 17.5% decrease in the blood lactate during the second recovery interval. Thus, the 42- min postsubmersion recovery interval permitted recovery of blood gases, but was inadequate to clear the blood lactate (Fig. 2, Table 3). Lactate declined 80.4% and 83.8%, respec- tively, during the first two 180-min postsubmergence re- covery intervals in treatment-6 turtles. As was the case for laboratory submerged sea turtles, a longer surface interval ultimately resulted in an increased ability to recover from the submersion episodes. In fact, lactate declined 82.7%, 82.8%, and 87.9%, respectively, in treatment-4, -5, and -6 turtles 180 minutes after the final submersion episode (Fig. 2, Table 3). Stabenau and Vietti: The physiological effects of multiple forced submergences of Caretta caretta 897 Table 3 Effects of forced submergence on blood pH, Pco.„ and lactate in Kemp's ridley (LK) and loggerhead (CC) sea turtles. Data are | expressed as the mean difference (J) between post- and presubme rgence values. Data from this study are provided from the three submergence episodes of treatment 1-3 turtles under the laboratory protocol and treatment 4-6 turtles in the field protocol, ND = | not determined. Turtle size Submergence dPco^ iilactate Species (kg) duration (min) 4pH ( mm Hg) (mMl Reference LK 5-16.5 <7.3 0.37 12.8 8.5 Stabenau etal. (1991) 5-6 <7.3 0.31 24.5 15.1 TED certification tests' CC 5-6 4.3 0.33 ND 13.4 TED certification tests' 5-6 12.5 0.52 ND 17.2 CC 6.5-7.0 7.5 treatment 1 1)0.49 61.1 7.6 Laboratory study 2)0.11 16.3 -0.1 3)0.10 15.3 1.1 treatment 2 1)0.57 70.8 9.3 2)0.20 20.9 2.3 3)0.23 21.1 1.1 treatment 3 1)0.59 98.7 9.6 2)0.50 68.6 7.2 3)0.46 67.3 5.9 treatment 4 1)0.63 45.8 10.2 Field study 2)0.16 24.5 1.9 3)0.11 9.3 0.9 treatment 5 1)0.53 36.3 9.1 2) 0.38 19.9 3.5 3) 0.28 17.5 3.0 treatment 6 1)0.73 54.2 11.2 2)0.66 31.3 9.2 3)0.65 27.5 9.3 ' Data were collected by one of the authors (EKS) during standard TED certification tests in 1993-94. S amples were collected from the cervical sinus of Kemp's ridley and loggerhead sea turtles prior to and following forced submergences in a commercial shrimp net equipped with a TED, Turtles | m these studies were permitted to exit the TED-equipped net. It must be noted that any discussion on lactate produc- tion and recovery following submersion is applicable to environmental conditions comparable to those reported in this study. For example, lactate formation and recovery rates of lactate build-up would be significantly influenced by water temperature. Longer recovery rates may take place in cold water, whereas warmer waters may lead to additional lactate production thereby influencing the rate of lactate elimination. In addition, the blood lactate con- centrations measured in this study may underestimate the true lactate burden. Lactate has been shown to partition into other tissues, including the shell, following submer- sion of freshwater turtles (Jackson et al., 1999). Finally, sea turtle size could potentially alter lactate production and elimination. Results from submersion experiments conducted in our laboratory indicate that smaller animals exhibit a significant acidosis and lactate build-up in com- parison to larger sea turtles. Whether less acidosis and lactate build-up is due to additional lactate buffering by the larger sea turtles warrants further investigation. Ions, osmolality, and hematocrit There are three primary mechanisms for recovery of blood pH following an acid-base disturbance: cellular buffer- ing, and respiratory and renal compensation. Cellular responses occur immediately following the disturbance, whereas respiratory and renal adjustments occur within minutes to hours, respectively. Previously, Stabenau et al. (1991) reported that Kemp's ridley sea turtles exhibited a significant increase in plasma [K+] following trawl sub- mergences. However, those authors reported that trawl stress had no effect on plasma [Cli, [Na+], or hematocrit. In the present study, a cellular response to the severe acid-base disturbance caused by the multiple forced submergences was suggested by alterations in plasma ion concentrations, osmolality, and hematocrit during the blood acidosis. As shown in Figure 3, decreases in blood pH were correlated with increases in [K*], [Na*l, [CI"], osmolality, and hematocrit. Hematocrit (percent packed red blood cells) changes may result from washout of additional red blood cells into the bloodstream, from areas such as the spleen, in order to pro- vide more red blood cells during the hypoxic phases of the forced submergence. This explanation, however, is unlikely given that substantial fluctuations in hematocrit were ob- served during the course of the submergence experiments and that a normal hematocrit was measured in the final se- rial blood sample. A more plausible explanation is that there was an osmotically obliged influx of water into the red blood 898 Fishery Bulletin 101(4) cells, swelling the cells, and leading to increases in hemato- crit, and in plasma ion concentration and osmotic pressure. Red cell volume is regulated in animals through transport of intracellular and extracellular solutes. Although there is minimal information available in the literature concerning regulatory volume transport in reptiles, the mechanisms of regulatory volume increase (RVI) and regulatory volume decrease (RVD) are known in other lower vertebrates. For example, Cala (1983) reported that in Amphiuma (am- phiuma [common name]) red cells, the mechanism of RVD is K+„^j/H+jj^ counter-transport coupled with Chg^j/HCOg'jj, exchange (where the subscripts in and out represent transport into and out of the cell, respectively), whereas RVI is accomplished by Na^j^/H+^j^^j transport coupled with Cl-,^/HC03-„,jt exchange (Cala, 1983). Other studies have suggested that red cell RVD occurs because of electroneu- tral KCl cotransport out of the cell and RVI occurs because of electroneutral NaK2Cl or NaCl cotransport into the cell (Haussinger and Lang, 1991). It is impossible to determine which of these mechanisms, if any, were involved in regu- lating red cell volume in sea turtles during and following forced submergence. These transporters, however, have been shown to be sensitive to cellular hypoxia (i.e. low Pog) and low blood pH (Cossins and Gibson, 1997) — conditions present in the experimental turtles following submergence. In addition, hypoxic and acidotic conditions were absent in nonsubmerged control turtles which did not experience substantial shifts in plasma ion concentrations, osmotic pressure, or hematocrit. Effects of handling Significant changes in blood pH, Pco.2, and lactate were occasionally detected in nonsubmerged control turtles. However, it is impossible to determine if these changes resulted from repetitive handling during blood sampling or from increased activity while free-swimming in a large cir- cular tank following blood collection. Nevertheless, control turtle blood lactate concentration was substantially less than the lactate measured following forced submergence in experimental turtles (Figs. 1 and 2). In addition, the blood pH remained fairly constant in the control turtles during collection of the seven serial samples. Laboratory versus field experimentation It should be noted that conducting the study under labo- ratory and field conditions provided unique benefits for analyzing the physiological effects of submersion. For example, the laboratory conditions permitted collection of blood samples immediately upon termination of the sub- mersion period, whereas in the field, sea turtles had to be transported back to the trawl vessel for postsubmersion blood sampling. Turtles forcibly submerged under labora- tory or field conditions hyperventilated upon surfacing. Stabenau et al. (1991) reported a 9- to 10-fold increase in the breathing frequency of trawled Kemp's ridley sea turtles. Comparable breathing rates were observed in the present study after submersion and, thus, it is plausible that the blood PcOg measured in turtles under field condi- tions underestimated the actual buildup in blood COg (see Table 3 for a comparison of the blood Pco, under labora- tory and field conditions). In contrast, the field experiment permitted examining the physiological stress of semiwild turtles in TED-equipped commercial fishing nets following a minimum of 21 days of in-water conditioning. The greater acidosis measured in forcibly submerged turtles resulted from increased swimming activity during the forced sub- mergence. This is confirmed by a postsubmergence increase in blood lactate of 10.1 niM under trawling conditions versus 8.8 mM following laboratory submergence. Under laboratory and field conditions, the behavior of the turtles following submergence was monitored up to their release. It is unclear, however, if the acid-base and ionic imbal- ance caused by forced submersions would alter long-term normal physiology and behavior. It is plausible that repeti- tive alteration of blood pH by the magnitude measured in the present study may have pathological consequences. For example, no information is available on whether turtles resume normal diving and feeding behavior following pro- longed or multiple forced submersions, or whether turtles become more susceptible to repeated submersions in TED- equipped nets. Use of turtles reared in captivity Two-year-old loggerhead sea turtles reared in captivity were used for all of the submergence experiments. It was assumed that these animals were adequate surrogates for wild sea turtles. In fact, similar-size animals from the NMFS Galveston Laboratory are used in annual TED certification trials. Nevertheless, there may be differences in the physiology of captive and wild turtles subjected to forced submergences. For example, it is possible that wild sea turtles would be exposed to forced submergences fol- lowing lengthy, voluntary dives. No information is available in the literature on the acid-base and ionic status of wild sea turtles following prolonged voluntary dives or forced multiple submergences. If dives are anaerobic, then sub- jecting wild sea turtles to multiple forced submergences may adversely affect survival potential. Conclusions The data suggest that forced submergences of 2-year-old loggerhead sea turtles reared in captivity produce signifi- cant blood metabolic and respiratory acidosis. Repetitive submergences did not augment the acidosis, rather subse- quent submergences resulted in less severe acid-base dis- turbances. Under trawl conditions, the turtle must recover from any physiological acid-base disturbance when it is freed from a TED-equipped net. Recovery is accomplished, in part, by the turtle immediately surfacing and hyper- ventilating (Jackson, 1985; Stabenau et al., 1991). This behavior was observed following each submergence epi- sode. Turtles would then resume normal voluntary diving behavior, presumably after partial-to-eomplete recovery from the acid-base disturbance. These data suggest that repetitive submergences of sea turtles in TED-equipped Stabenau and Vietti: The physiological effects of multiple forced submergences of Caretta caretta 899 nets would not significantly affect their survival potential, provided that the turtles have a recovery interval between successive submergences. However, it should be noted that the latter statement is based on comparable-size turtles that may be submerged in shrimp nets equipped with legally certified and installed turtle excluder devices. Poor installation or lack of use of legal TEDs would result in aug- menting the acid-base imbalance in the turtles. Increasing the magnitude of the blood acid-base and ionic disturbance during each submersion would increase the length of time necessary to achieve partial or complete recovery. Acknowledgments Grateful appreciation is expressed to personnel from the National Marine Fisheries Service Galveston and Pasca- goula Laboratories for their assistance in turtle husbandry and in conducting the submersion protocol under labora- tory and very difficult field conditions. These studies were conducted under appropriate threatened and endangered species permits issued by the U.S. Fish and Wildlife Ser- vice, Texas Park and Wildlife Department, and the Florida Department of Natural Resources. Literature cited Caillouet, C. W., M. J. Duronslet, A. M. Landry, D. B. Revera, D. J. Shaver, K. M. Stanley, R. W. Heinly, and E. K. Stabenau. 199L Sea turtle strandings and shrimp fishing effort in the Northwestern Gulf of Mexico, 1986-89. Fish. Bull. 89:712-718. Cala, R M. 1983. Volume regulation by red blood cells: mechanisms of ion transport. Mol. Physiol. 4:33-52. Cossins, A., and J. Gibson. 1997. Volume-sensitive transport systems and volume homeostasis in vertebrate red blood cells. J. Exp. Biol. 200:343-352. Crowder, L. B., S. R. Hopkins-Murphy, and J, A. Royle. 1995. Effects of turtle excluder devices (TEDs) on loggerhead sea turtle strandings with implications for conservation. Copeia 1995:773-779. Haussinger, D., and F. Lang. 1991. The mutual interaction between cell volume and cell function: a new principle of metabolic regulation. Biochem. Cell Biol. 69:1-4. Jackson, D. C. 1985. Respiration and respiratory control in the green turtle, Chelonia mydas. Copeia 1985:664-671. Jackson, D. C, Z. Goldberger, S. Visuri, and R. N. Armstrong. 1999. Ionic exchanges of turtle shell in vitro and their rel- evance to shell function in the anoxic turtle. J. Exp. Biol. 202(part5):513-.520. Lutz, R L., and A. Dunbar-Cooper 1987. Variations in blood chemistry of the loggerhead sea turtle Caretta caretta. Fish. Bull. 85:37-44. National Research Council. 1990. Decline of sea turtles: causes and prevention, 259 p. National Academy Press, Washington, DC. Owens D. W., and G. J. Ruiz. 1980. New methods of obtaining blood and cerebrospinal fluid from marine turtles. Herpetologica 36:17-20. Stabenau, E. K., and T A. Heming. 1994. The in vitro respiratory and acid-base properties of blood and tissue from the Kemp's ridley sea turtle, Lepido- chelys kempi. Can. J. Zool. 72:1403-1408. Stabenau, E. K., T. A. Heming, and J. A. Mitchell. 1991. Respiratory, acid-base and ionic status of Kemp's ridley sea turtles (Lepidochelys kempi) subjected to trawling. Comp. Biochem. Physiol. 99A:107-111. Wood, S. C, R. N. Gatz, and M. L. Glass. 1984. Oxygen transport in the green sea turtle. J. Comp. Physiol. B. 154:275-280. 900 Abstract— The sectioned otoliths of four fish species from a tropical demer- sal trawl fishery in Western Australia revealed a series of alternating trans- lucent and opaque zones in reflected light. The translucent zones, referred to as growth rings, were counted to determine fish ages. The width of the opaque zone on the periphery of the otolith section as a proportion of the width of the previous opaque zone (index of completion) was used to de- termine the periodicity of growth-ring formation. This article describes a method for modeling changes in the index of ring completion over time, from which a parameter for the most probable time of growth-ring formation (with confi- dence intervals) can be determined. The parameter estimate for the timing of new growth-ring formation for Leth- rinus sp. 3 was from mid July to mid September, for Lutjanus vitta from early July to the end of August, for Nemipterus furcosus from mid July to late September, and for Lutjanus sebae from mid July to mid November. The confidence intervals for the timing of formation of growth rings was variable between species, being smallest for L. vitta, and variable between fish of the same species with different numbers of gi'owth rings. The stock assessments of these commercially important species relies on aging information for all the age classes used in the assessment. This study demonstrated that growth rings on sectioned otoliths were laid down annually, irrespective of the number of growth rings, and also demonstrated that the timing of ring formation for these tropical species can be deter- mined quantitatively (with confidence intervals). Quantitative determination of the timing of otolith ring formation from marginal increments in four marine teleost species from northwestern Australia Peter C. Stephenson Western Australian Marine Research Laboratories West Coast Drive (off Elvire St) Waterman, Western Australia, 6020, Australia E-mail address: pstephensongifish.wa.govau Norm G. Hall School of Biological and Environmental Sciences Murdoch University Murdoch, Western Australia, 6150, Australia Manuscript approved for publication 3 June 2003 by Scientific Editor. Manuscript received 26 June 2003 at NMFS Scientific Publications Office. Fish. Bull 101:900-909 (2003). The Pilbara fish trawl fishery, operat- ing on the North West Shelf of Western Australia, has (developed rapi(ily in the last ten years and is now the most valuable commercial scalefish fishery in Western Australia. Catch from this fish- ery was valued at $7 million (wholesale value) in 2001. In this multispecies fish- ery, Lutjanus vitta (Quoy and Gaimard, 1824) (brownstripe red snapper), Nemipterus furcosus (Valenciennes, 1830) (fork-tailed threadfin bream, also known as rosy threadfin bream), Leth- rinus sp. 3 (Carpenter and Niem, 2001) (lesser spangled emperor, known locally as blue-spot emperor) made up 8%, 10%, and 20% respectively of the total sca- lefish trawl catch in 2000. The highly prized species, Lutjanus sebae (Cuvier, 1828) (red emperor), although compris- ing only 4% of the catch, is important because of its high market value. In 1993 a research project was com- menced to determine the fishing effort required for optimal level of catches in the Pilbara trawl fishery (Stephenson and DunkM. The project relied on vali- dated age composition data for L. vitta, L. sp. 3, N. furcosus, and L. sebae. The growth rings on otoliths have been shown to be formed annually for only one to three growth rings for N. furcosus (Sainsbury and Whitelaw, 1984), and for two to three growth rings for L. sebae (McPherson and Squire, 1992). After pooling of all age classes, Davis and West (1992) showed that growth rings of L. vitta were formed annually. Determining age composition in- volves counting growth rings on hard parts of fish (otoliths, scales, spines, bones) and determining the timing of growth-ring formation. Sagittal otoliths are commonly used for aging teleost fishes and recent studies (Hyndes et al., 1992; Milton et al., 1995; Newman et al., 1996) have indicated that for some species sectioned otoliths give more reliable age estimates than whole, or broken-and-burnt otoliths. The peri- odicity of ring formation is commonly determined by the mark-recapture method in which fish are injected with chemical markers and the number of rings created between injection and recapture are compared (Ferreira and Russ, 1992; Francis et al., 1992; New- man et al., 1996). An alternative to mark-recapture is marginal increment analysis in which the distance from the growth ring to the edge of the otolith, for a sample offish, is tracked over time (Campana, 2001) and a sharp drop in this marginal increment, once a year, is taken as an indication of annual ring formation. The analysis is often performed on Stephenson, P. C, and I. Dunk. 1996. Re- lating fishing mortality to fish trawl effort on the North West Slope of Western Australia. Final report of project 93/25 to the Fisheries Research and Development Corporation, 1995, 44 p. Western Austra- lia Marine Research Laboratories, PO Box 20, North Beach, Western Australia 6092, Australia. Stephenson and Hall; Timing of otolith ring formation in marine teleosts from northwestern Australia 901 pooled age classes (Barger, 1985; Manickchand-Heileman and Kenny, 1990; Murphy and Taylor, 1990; Ross et al., 1995; Pearson, 1996; Morales-Nin and Moranta, 1997; Van der Walt and Beekley, 1997) or on a restricted number of age classes (Sainsbury and Whitelaw, 1984; McPherson and Squire, 1992). Analysis with pooled data has limited value because there may be different patterns of gi'owth-ring formation at different life stages (e.g. at sexual maturity) and pooled data may have interage differences masked by dominant age groups (Beamish and McFarlane, 1983; Hyndes et al. , 1992). Studies in which data were pooled only for young and old fish, due to low fish numbers, reduced these prob- lems and improved the credibility of the results (Hyndes et al., 1992; Fletcher and Blight, 1996; Hasp et al., 2002). Accounts of statistical analysis of the marginal incre- ment data are rare. Davis and West ( 1992) used AN OVA to show that there were differences in the marginal increment of urohyal bones of L. vitta with time of year. As this sea- sonal pattern was the same for age classes 1 to 6, Davis and West ( 1992) pooled the data and used a graphical represen- tation to show the time of formation of the annual rings. This article describes a method for modeling changes in the index of completion of an otolith growth increment over time. This method enables quantitative determination of the most probable time of growth-ring formation (with confidence intervals) and is illustrated for the species L. vitta, L. sp. 3, N. furcosus, and L. sebae, from the Pilbara fish trawl fishery. Materials and methods Between October and November 1993 and between October and November 1994, samples of 30 fish of each species were randomly selected each month from fishery-independent trawl surveys. For the other months between January 1994 and March 1995, samples of 30 fish of each species were randomly selected each month from commercial catches. The samples came from an area between 115°30'E longi- tude and 120°E longitude; between the 50 meter and 100 meter depth isobaths. The sagittal otoliths were extracted from each sampled fish and the right otolith was embedded in epoxy resin and then sectioned transversely through the otolith core to a thickness of 0.4 mm. A Gemmaster high speed saw with a 100 mm by 0.1 mm diamond tipped saw blade was used for sectioning. The otolith sections were set on 76 mm by 50 mm glass slides with casting resin and covered with cover slips. The sections were viewed with a dissecting microscope with an attached color video camera connected to a per- sonal computer and a color monitor. Transmitted light revealed alternating wide opaque and narrow translucent zones. The translucent zones, referred to in the present study as growth rings, were counted to determine fish ages. The distance from the outer extremity of the last wide, dark band to the otolith edge, u',, is referred to as the mar- ginal increment and the distance between the outer edges of the second to last and the last dark band is denoted by «',_[. The distances w^ and w^^ on the portion of the otolith ventral to the sulcus towards the proximal margin were measured on the computer screen. The index of comple- tion, c,, was determined by using the formula of Tanaka etal. (1981) (1) and written to a file by using a computer program written in the programming language "HiSoft Basic" (version 2.0. MichTron, Auburn Hills, MI). The index of completion, c,, we expect to increase over time, and then decrease abruptly when a new growth ring is formed. The timing of formation of a new ring would occur at the same time for a fish species with the same number of rings, but there would be considerable variability in timing and detection between species and individuals (Fig. 1). The increase in the index of completion over time, t, is modeled as a strictly increasing function /■(<, a, b, d) with the following parameters: maximum value, a, rate of increase, 6, and horizontal translation, d. For our study, data were collected over a period of 18 months (October 1993 to March 1995) and the relation between the index of completion and time was expressed as two functions, denoted Fj and Fg Fj : f-j = fit, a. b, c/,) and F., : i .-. = f\t, a, b, d.J, (2) where c ^ and t ,, = the estimates of the index of comple- tion; t = the time in months from t = 0 il Octo- ber 1993) to < = 18 (31 March 1995); and d^, d.2 = the translation parameters for func- tions Fj and F.2 respectively. Ifthe point (c,, t) is associated with function Fj, the value of the normal probability density function of the observed deviation from Fj, evaluated at observation, i, is given by A,. = — p=exp and similarly the value of the normal probability density function of the observed deviation from F.^, evaluated at observation, i, is given by '^., =• a\!lK ^exp (c-t\)- where a- is the variance of the residuals when Fj is fitted to the data and where it is assumed to be equal to the vari- ance of the residuals when the function F^ is fitted. To ensure the tractability of the subsequent analysis, we assume that the probability, P,, of a point with index of completion c,, at time t, being represented by Fj, is given by the logistic function P.= 1 1 +exp ln(i9)^'^-') (S-R)_ (3) 902 Fishery Bulletin 101(4) Q. E Month Figure 1 Values of index of completion at time t, (c,. t) forL. vitta with five growth rings during the monthly time intervals from 1 October 1993 (t=Q) to 31 March 1995 (t=\S). Also illustrated are the functions Fj and Fj fitted to (C,, t) and the function P, representing the probability that (C,, t) is most hkely represented by function Fj. The likelihood functions, Aj and A.^ are shown for the values of index of completion when t = 10.4. where R, S, and R-iS-R) are the values off corresponding to the 50'*^ , 95*, and S"' percentiles of the logistic function. The probability that the index of completion is associated with Fj, rather than Fj, is calculated as 1-P,. Figure 1 illustrates typical values of the index of comple- tion at time t and the functions Fj and Fj representing these points before and after new growth-ring detection. The most likely time at which a new growth ring is detected is given by the value oft when P,=0.5. The likelihood func- tions Aj and Ag are illustrated for t = 10.4 months. When X, j is high, (Cj, t) is likely to lie closest to Fj and when A^ , is high, (c,, t) is likely to lie closest to F2. As a point (c,, t) will be associated with either Fj or Fg (but not both), it follows that the likelihood function K is given by the probability of the observed deviation from Fj, X given that the point is associated with F, That is, probability that the point is represented hyF, the probability of the observed deviation from Fji given that the point is associated withF,. X probability that the point is represented byF,. K,, = KA + ^^-^-P,^- The overall log-likelihood associated with all the observed points (c,, t), , for j = 1 to n in a particular age class is given by n n ^ln(^,,) = ^ln[A,,/^ + A,,(l-P,)]. The parameters of the functions Fj and Fj (i.e. a, b, d^, and 1^2' as well as a, R, and S) were estimated separately for each value of the number of rings by maximizing the log-likelihood. The value oft = R corresponds to the month where a val- ue of the index of completion is equally likely to be on either Fj or F2; that is, the point where the drop in the index of completion occurs. The value of S and R-iS-R) correspond to the 95'*^ and the 5'^ percentiles for the time at which a new growth ring is likely to be detected, indicating reliabil- ity of the estimate of the time of ring formation t = R. Results The plots of index of completion versus time reflected this growth pattern in the four species we studied, with the growth rate decreasing as the time of new growth-ring detection approached. The temporal pattern of growth of the otolith suggests the index of completion could be modeled with a logistic function F{t.a.b,d) = - 1 -i-exp lnl9 (cl't) with the maximum value o = 1, phase shift d, and rate of increase b. A characteristic of otolith growth is that the distance between growth rings decreases each year, thus, the rate of increase in the marginal increments will be greater for fish with few rings and less for fish with many rings. On the other hand, the index of completion, being the ratio of Stephenson and Hall: Timing of otolith ring formation in marine teleosts from northwestern Australia 903 ? 08 0.6 0.4- 02 • ^ 4 rings • • X • 1 • • • •v.— — — T" • • 1 . u- 1 • \ • ! 1 . . • 1 • • 1 \ * * A \ 0123456789 101112131415161718 0123456789 1011121314 15161718 ONDJ FMAMJJASONDJFM ONDJ FMAMJJASONDJFM Month Figure 2 Index of completion at time<,(c„ t), for L. sp. 3 for growth-ring categories 2,3,4,5,6, 7,8, and 9-11 sampled between the months 1 October 1993 {t=0} to 31 March 1995 it=\8). The sohd circles (•) represent those (c,, t) most likely represented by c^ and lying closest to the function Fj and the solid triangles (A) are those most likely represented by c,, and lying closest to the function Fj. The logistic function, P,, indicates the probability that points (c,, t) are most likely represented by function F,. the marginal increment to the width of the previous band, would be expected to be constant at the same time of year for a particular species, regardless of the ring count. Thus, during the maximization of the objective function n the rate of increase parameter b is assumed constant for all ring counts for a species, but the other parameters (dj , d^, R, and S) are estimated for each number of rings for each species. Figure 2 shows the pattern of changes in the index of completion for L. sp. 3 for otoliths with two to eleven growth rings from 1 October 1993 («=0) to 31 March 1995 (<=18). The data were pooled for the older age classes (nine or more growth rings) because of the small numbers of old fish in the samples. The points indicating the values of in- dex of completion, c^, are represented with different symbols according to whether they are most likely to be represented by Cj (lying closest to the function Fj) or most likely to be represented by c^ dj^ng closest to the lower line F^). The function P, represents the probability of points being repre- sented by line Fj. The time where P,=0.5 is the most likely time of detection of the formation of a new growth ring. In the other three species, L. vitta, N. furcosus, and L. sebae, the points representing index of completion, (c,, t) and functions, F^, Fj, and P, are illustrated in Figures 3, 4, 904 Fishery Bulletin 101(4) 3 rings ^ 0.8- • • • Ar -♦- • 0.6- • • 1 ^i : 1 • • A J»r • * A *^ r 0.4- o % (B (1^- ^^ — ■% ' ^ CL \ A t nn- 1-0 5 rings • • * • 1 0.8- • ■ • V • 0.6- 0,4- • • • • • • • • A • A P 0 2- nn- • • . A A ^ d: i*-i r- ( 0123456789 101 1 12 13 14 15 1617 1£ ONDJ FMAMJ J ASONDJFM 0 12 3 4 5 6 7 ONDJ FMAM Month Figure 3 Index of completion at time t , (c,, t), for L. vitta for growth-ring categories 2 ,3, 4, 5, 6, 7, 8, and 9-12 sampled between the months 1 October 1993 (;=0) to 31 March 1995 (?=18). The solid circles (•) represent those (c,, t) most likely represented by t-j and lying closest to the function Fj and the solid triangles (A) are those most likely represented by c, and lying closest to the function F.,. The logistic function, P, , indicates the probability that points (c,, I) are most likely represented by function F,. and 5. Because of the smaller numbers offish, the data was pooled for L. vitta with 9-12 growth rings, for N. furcosus with 7-9 growth rings, and for L. sebae with 13-14 and 15-19 growth rings. The estimates of the parameters b, d^, d, (the rate of in- crease and phase shift of Fj and F^), R, the time of formation of a new growth ring, with the range ±(S-R), and the stan- dard deviation (cj) for the four species are listed in Table 1. The estimated time of formation of a new ring varied between age classes and occurred for L. sp. 3 from mid July to mid September, for L. vitta from early July to early September, for N. furcosus from mid July to late September, and for L. sebae from mid July to mid November. The confidence intervals are small for L. sp. 3 and L. vitta and generally larger for A^. furcosus and L. sebae where read- ing increments was more difficult. The confidence interval was smaller for L. sebae with 12-19 growth rings because the clarity of the rings generally improved for older fish. The difference between the phase shift for functions f j and F^, that is d^-di, was between 11 to 13 months. This difference indicated an annual cycle of growth-ring forma- tion. The standard deviation, a, of the values of index of completion was lowest for forL. vitta (0.1-0.14) and highest fori, sebae (0.11-0.21) and L. sp. 3 (0.11-0.24). Discussion With different starting values for the parameters, the method we described found that the time of detection of growth-ring formation for the four species was consistent. Although the timing varied considerably for different num- bers of growth rings, the estimate was generally similar for each species. Other marginal increment studies on these species pro- duced estimates of the times for growth-ring formation Stephenson and Hall: Timing of otolith ring formation in marine teleosts from northwestern Australia 905 0123456789 1011121314 15161718 0123456789 1011121314 15161718 ONDJ FMAMJ J ASONDJFM ONDJ FMAMJ J ASONDJFM Month Figure 4 Index of completion at time t, (c,, t), for N. ftircostis for growth-ring categories 2 ,3, 4, 5, 6, and 7-9 sampled between the months 1 October 1993 (^=0) to 31 March 1995 (/=18). The solid circles (•) represent those (c,, t), most likely represented by c"j and lying closest to the function F, and the solid triangles (A) are those most likely represented by c"., and lying closest to the function Fj. The logistic function, P,, indicates the probability that (c,, t) are most likely represented by function Fj. that are consistent with those determined in the present study (Table 2). Davis and West ( 1992) found that the time of formation of a new translucent growth-ring on urohyal bones for L. vitta was October, later than the timing found in our study. Sainsbury and Whitelaw (1984) found the marginal increment values on whole otoHths from N. fur- casus had low values in July 1979 and in May 1980 (earlier than observed in the present study for sectioned otoliths) but the sampling reported by Sainsbury and Whitelaw (1984) was very sparse: four sampling times in 1979 and two in 1980. McPherson and Squire (1992) reported that the mean monthly marginal increment of the first two age classes for L. sebae appeared to have a minimum between July and September that is consistent with the present study. In our study, the growth zones on L. vitta were generally clearly defined; the opaque zone was easily distinguished from the translucent zone, and there were few discontinui- ties (areas of dissimilar structure or optical density within the growth zone). This clear definition was especially no- ticeable for the outer growth zones of the otolith for older fish which often had very clear dark zones. This finding is consistent with the small confidence intervals for this spe- cies, especially in the fish with a greater number of growth rings in their otoliths. The growth rings on L. sp. 3 and N. furcosus had poor contrast and had many discontinuities which made the analysis difficult. For young L. sebae, there were many discontinuities within the growth zones which made locating the translucent zone difficult. For fish with two to four growth rings, low values of the index of comple- tion occurred when t = 7 and also when t = 12 (Fig. 5). For the older L. sebae (where the number of rings is greater than or equal to 12) the wide zone was very dark and by in- creasing the magnification, the marginal increment could be measured relatively easily. The narrow confidence limits for the timing of growth-ring formation for L. sebae are consistent with this explanation but the small number of data points results in less reliable measures in the timing of new ring formation. The time of formation of the new growth ring was slightly earlier for L. vitta than for L. sp.3, N. furcosus, or L. sebae. The calculation of an earlier growth-ring forma- tion may be attributed to the more clearly defined trans- lucent zone in L. vitta which may be detectable earlier in 906 Fishery Bulletin 101(4) 0,8- 5 rings • : 1 • • ( • • U.b- • • . r^^T « . '>'"^* O.4. • • ^^ -r"*^ -— ' i 0.?- ' A * A nn. A 1— i. . 10 rings • N.«-i-^ ■ • • • -^ A • >^f^ \ * , ^- * • • ^v — -"^ • * A \i 0.8- 8 rings t • • • ■*. t .1 • • • • 0-6- 0.4- 02- I I. • • • i « V nn. * 0123456789 101112131415161718 0123456789 1011121314 15161718 ONDJFMAMJJASONDJFM ONDJFMAMJJASONDJFM Month Figure 5 Index of completion at time t, (c,, t). for L. sebae for growth-ring categories 2-3, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13-14, and 15-19 sampled between the months 1 October 1993 «=0) to 31 March 1995 (t=18). The solid circles (•) represent those (c,, t) most likely represented by i", and lying closest to the function Fj and the solid triangles (A) are those most likely represented by c'j and lying closest to the function F^. The logistic function, P, , indicates the probability that (c,, t) are most likely represented by function F,. Stephenson and Hall: Timing of otolith ring formation in marine teleosts from northwestern Australia 907 Table 1 Parameter estimates for the fit of two functions to the index of completion data. The parameters are the phase shifts rfj and d.,, the time of new growth-rmg detection t = R with the confidence interval (5"' and 9£ standard deviation, a, of the observed values of index of completion for points fitted to function Fj and F^. rate of increase, b, the "> percentiles), and the rings ;i h rf; d.. R Month a Lutjanus sp. 3 2 36 20.3 3.6 15.3 9.5 ±0.1 Jul 0.14 3 134 20.3 2.0 15.4 9.9 ±2.0 Jul 0.13 4 146 20.3 1.7 16.2 10.2 ±2.1 Aug 0.13 5 72 20.3 2.3 14.8 10.5 ±0.1 Aug 0.24 6 38 20.3 1.7 14.9 11.5 ±0.2 Sep 0 11 7 22 20.3 0.6 13.7 10.7 ±0.1 Aug 0.12 8 22 20.3 1.1 14.0 10.2 ±0.1 Aug 0 12 9-11 29 20.3 0.9 14.2 10.9 ±1.3 Aug 0.15 Lutjanus vitta 2 31 15.8 3.0 14.3 10.9 ±0.1 Aug 0.13 3 62 15.8 1.5 18.3 10.5 ±0.2 Aug 0.14 4 96 15.8 3.4 15.9 10.5 ±0.2 Aug 0 10 5 127 15.8 2,7 16.3 10.2 ±2.0 Aug 0.12 6 66 15.8 2.5 16.5 10.0 ±1.3 Aug 0.12 7 56 15.8 2.0 15.4 10.3 ±1.7 Aug 0.11 8 24 15.8 3.3 15.8 9.0 ±0.2 Jul 0.10 9-12 39 15.8 3.1 16.2 11.1 ±0.2 Sep 0.10 Nemipterus furcosus 2 85 11.9 2.0 13.1 10.6 ±2.7 Aug 0.18 3 95 11.9 3.9 14.2 9.5 ±2.2 Jul 0.17 4 116 11.9 4.0 15.0 9.5 ±4.3 Jul 0.14 5 82 11.9 3.3 16.4 11.8 ±2.6 Sep Oil 6 53 11.9 4.0 16.4 11.8 ±4.2 Sep 0.10 7-9 39 11.9 4.5 15.8 11.4 ±3.0 Sep 0.10 Lutjanus sebae 2-3 34 17.4 1.9 14.4 7.8 ±2.5 May 0.17 4 62 17.4 5.0 15.9 11.9 ±1.2 Sep 0.21 5 43 17.4 5.1 13.4 11.6 ±6.7 Sep 0.14 6 60 17.4 3.3 17.0 9.6 ±3.7 Jul 0.13 7 63 17.4 2.6 17.0 10.6 ±4.9 Aug 0.16 8 60 17.4 2.8 15.8 13.5 ±7.6 Nov 0.15 9 52 17.4 3.2 15.9 11.0 ±3.3 Sep 0.19 10 32 17.4 3.3 15.9 10.6 ±3.4 Aug 0.15 11 34 17.4 3.1 14.9 11.6 ±2.6 Sep 0.17 12 22 17.4 2.1 13.9 9.6 ±0.1 Jul 0.12 13-14 23 17.4 4.7 14.6 13.3 ±0.4 Nov 0.21 15-19 21 17.4 4.0 17.6 11.5 ±0.2 Sep 0.11 Table 2 The timing of growth-ring formation for the four species in the present study and for comparative studies Lutjanus sp.3 L. vitta Nemipterus furcosus L. sebae Present study July-September July-September July-September July-November Davis and West (1992) October Sainsbury and Whitelaw ( 1984) May-July McPherson and Squire (1992) July-September 908 Fishery Bulletin 101(4) the year than it is in the other three species. Similarly, the apparent earlier timing of new ring creation in our study, compared to the findings of Davis and West (1992) may have been due to the fact that the translucent zone can probably be detected closer to the time of formation in sectioned otoliths than in urohyals (Reshetnikov and Claro, 1976). In summary, modeling the change in index of completion over time enabled estimates to be made of the time of the formation of a new growth ring (with confidence intervals) for four tropical species. Although the index of completion was modeled with a logistic function in the present study, alternative functions (e.g. sine or linear), gave very similar results. The technique is a useful addition to marginal in- crement analysis because it can be used in place of previous subjective methods to determine quantitatively the timing of new ring formation. Acknowledgments This project was financed by the Fisheries Research and Development Corporation (FRDC) (project 94/25) and the Department of Fisheries, Western Australia where the main author was employed for the duration of the project. The authors thank Mike Moran (Department of Fisher- ies, Western Australia) for obtaining FRDC funding and providing critical advice and encouragement. The senior author also thanks Robert Black (University of Western Australia) for his valued suggestions and encouragement. Stephen Newman (Department of Fisheries, Western Australia) provided advice on reading and interpretation of otolith bands, Iain Dunk (Department of Fisheries, Western Australia) collected samples, sectioned otoliths, and acted as the second reader for fish aging and mar- ginal increment analysis. Tony Paust, Ken Bryers, Justin Chidlow, Daryn Payne (Department of Fisheries, Western Australia) assisted in sample collection. I also acknowledge the assistance of M. G. Kailis, Kraus Fishing Company, and Westmore Seafoods whose vessels were used for sample collection, and two anonymous referees for their construc- tive 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. 112:735-743. Barger, L. E. 1985. Age and growth of Atlantic Croakers in the Northern Gulf of Mexico based on sectioned otoliths. Trans. Am. Fish. Soc. 114:847-850. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish Biol. 59:197-242. Carpenter, K. E., and V. H. Niem. 2001. FAG species identification guide for fishery purposes. The living marine resources of the Western Central Pacific, vol. 5. Bony fishes, part 3 (Menidae to Pomacentridae), p. 2791-3380. FAG (Food and Agriculture Organization of the United Nations), Rome, Italy. Davis, T. L. G., and G. J. West. 1992. Growth and mortality of Lutjanus vittus (Quoy and Giamard) from the North West Shelf of Australia. Fish. Bull. 90:395-404. Ferreira, B. P., and G. R. Russ. 1992. Age, growth and mortality of the inshore coral trout Plectropornus maculatus (Pisces: Serranidae) from the Cen- tral Great Barrier Reef Australia. Aust. J. Mar Freshw. Res. 43:1301-1312. Fletcher, W. J., and S. J. Bhght. 1996. Validity of using translucent zones of otoliths to age the pilchard Sardinops sagax neopilchardus from Albany, Western Australia. Mar Freshw. Res. 47:617-624. Francis, R. I. C. C, L. J. Paul, and K. P Mulligan. 1992. Ageing of adult snapper (Pagrus Auratus), from otolith annual ring counts: validation by tagging and oxytetacy- cline injection. Aust. J. Mar Freshw. Res. 43:1069-1089. Hesp, S. A., I. A. Potter, and N. G. Hall. 2002. Age and size composition, growth rate, reproductive biology, and habitats of the West Australian dhufish, Glau- cosoma hebraicum, and their relevance to management of this species. Fish. Bull. 100:214-227. Hyndes, G. A., N. R. Loneragan, and I. C. Potter 1992. Influence of sectioned otoliths on marginal increment trends and age and growth estimates for the flathead Platy- cephalus speculator. Fish. Bull. 90:276-284. Manickchand-Heileman, S. C, and J. S. Kenny. 1990. Reproduction, age, and growth of the whitemouth croaker Micropogonias furnieri (Desmarest 1823) in Trini- dad waters. Fish. Bull. 88:523-529. McPherson, G. R., and L. Squire. 1992. Age and growth of three dominant Lutjanus species of the Great Barrier Reef inter-reef fishery. Asian Fish. Sci. 5:25-36. Milton, D. A., S. A. Short, M. F O'Neill, and S. J. Blaber 1995. Ageing of three species of tropical snapper (Lutjani- dae) from the Gulf of Carpentaria, Australia, using radiom- etry and otolith counts. Fish. Bull. 93:103-115. Morales-Nin, B., and J. Moranta. 1997. Life history and fishery of the common dentex (Dentex dentex) in Mallorca (Balearic Islands, western Mediterranean). Fish. Res. 30:67-76. Murphy, M. D., and R. G. Taylor 1990. Reproduction, growth and mortality of red drum Sciaenops ocellatus in Florida waters. Fish. Bull. 88: 531-542. Newman, S. J., D. M. Williams, and G. R. Russ. 1996. Age validation, growth and mortality rates of the tropi- cal snappers (Pisces: Lutjanidae ) Lu(/a/?!zs adetii (Castelnau, 1973) andL. quinquelineatiis (Bloch, 1790) from the Central Great Barrier Reef Aust. J. Mar Freshw. Res. 47:575-584. Pearson, D. E. 1996. Timing of hyline-zone formation as related to sex, loca- tion, and year of capture in otoliths of the widow rockfish, Sebastes entomelas. Fish. Bull. 94:190-197. Reshetnikov, Y. S., and R. M. Claro. 1976. Cycles of biological processes in tropical fishes with reference to Lutjanus synagris. J. Ichthyol. 16:711-733. Ross, L. J.,T. M. Stevens, and D. S. Vaughan. 1995. Age, growth, mortality, and reproductive biology of red drums in North Carolina waters. Trans. Am. Fish. Soc. 124:37-54. Sainsbury, K. J., and A. W. Whitelaw. 1984. Biology of Peron's threadfin bream Nemipterus pero- nii (Valenciennes), from the North West Shelf of Australia. Aust. J. Mar Freshw. Res. 35:167-185. Stephenson and Hall: Timing of otolith ring formation In marine teleosts from northwestern Australia 909 Tanaka, K.. Y. Mugiya, and J. Yamada. Van der Walt, B.A., and L. E. Beckley. 1981. Effects of photoperiod and feeding on daily growth 1997. Age and growth of Sarpa salpa (Pisces: Sparidae) off patterns in otoliths of juvenile Ttlapia nilotica. Fish. the coast of South Africa. Fish. Res. 31:241-248. Bull. 79:459-465. 910 Application of DNA-based techniques for the identification of whaler sharks (Carcharhinus spp.) caught in protective beach meshing and by recreational fisheries off the coast of New South Wales Ricky W. K. Chan School of Biological, Earth and Environmental Sciences The University of New South Wales, UNSW Sydney, New South Wales 2052, Australia Present address: Educational Testing Centre The University of New South Wales ULD 3 East Parcel Centre Rosebery, New South Wales 2018, Australia E-mail address sharkman@etc.unsw,edu au Patricia I. Dixon Centre for Manne and Coastal Studies The University of New South Wales, UNSW Sydney, New South Wales 2052, Australia Julian G. Pepperell Pepperell Research and Consulting PO Box 1475 Noosaville DC, Queensland 4566, Australia vessels may experience difficulties in identification if distinguishing parts of a shark are discarded prior to confir- mation of species (Stevens and Wayte^). Similarly, observers of protective beach meshing may find species identifica- tion difficult on severely decomposed sharks. Without proper identification, the exact number of individual species inhabiting NSW waters and the num- ber of each species being landed cannot be determined (Chan, 2001). The rise of molecular biological tech- niques in marine forensic science has facilitated the development of accu- rate taxonomic identification of shark species by sampling biological tissue (Martin, 1991; Lavery 1992; Heist and Gold, 1999). DNA techniques require only muscle tissue, allowing biopsy tissue to be taken from specimens that can be released, rather than having to sacrifice the shark to obtain liver and heart tissue for allozyme analysis (Godfrey, 1997). Methods of taxonomic identification include PCR-based restriction fragment length poly- Dennis D. Reid New South Wales Fishenes PO Box 21 Cronulla, New South Wales 2230, Australia The International Union for the Conser- vation of Nature's (lUCN) development of the Shark Specialist Group is indica- tive of the increasing environmental awareness of sharks' crucial ecological role as apex predators and that they are being threatened by human activi- ties. Although the conservation status of certain carcharhinid species (Car- charhinus limbatus, C. obscurus, and C. plumbeus) are presently considered at low risk or near threatened accord- ing to the lUCN's threatened species categories,' species from the genus Carcharhinus are known to inhabit the waters of New South Wales (NSW), Australia (Stevens, 1984; Last and Stevens, 1994); however their conser- vation status has not been determined. Known as whaler or "requiem" sharks, they are also commonly caught off the coast of New South Wales in com- mercial fisheries (Stevens and Wayte^; Tanner and Liggins^), recreational fish- eries (Pepperell, 1992; Gartside et al., 1999; Steffe et al.*) and by protective beach meshing (Reid and Krogh, 1992; Dudley 1997). Because of morphological similari- ties between a number of shark spe- cies in the genus Carcharhinus (Last and Stevens, 1994; Naylor and Marcus, 1994), taxonomic identification to spe- cies level has been difficult or inac- curate (or both)(Stevens and Wayte^). Historical catches of certain species of sharks in NSW commercial fisheries, recreational fisheries, and protective beach meshing have been recorded to genus level only (Pepperell, 1992; Reid and Krogh, 1992; Tanner and Liggins'^). Formally trained Australian Fisheries Management Authority (AFMA) ob- servers aboard commercial longlining 1 Musick, J., and S. Fowler. 2000. Car- charhinus limbatus, C. obscurus and C. plumbeus In lUCN 2002. 2002 lUCN red list of threatened species, http: //www. iucn.org/redlist/2000index. html. [Accessed 1 October 2002.1 2 Stevens,J. D..andS. E.Wayte. 1998. A review of Australia's pelagic sharks resources. Fisheries Research and Devel- opment Corporation project 98/107, 64 p. CSIRO Marine Research, GPO Box 1538, Hobart, Tasmania 7001 Australia. 3 Tanner, M., and G. W. Liggins. 2000. New South Wales commercial fisheries statistics 1993/94 to 1996/98, 82 p. New South Wales Fisheries, PO Box 21, Cronulla, NSW 2230 Australia. ■* Steffe, A. S., J. J. Murphy, D. J. Chapman, B. E. Tarlinton, G. N. G. Gordon, and A. Grinberg. 1996. An assessment of the impact of offshore recreational fishing in New South Wales waters on the manage- ment of commercial fisheries. Fisheries Research and Development Corporation Project 94/053, 139 p. New South Wales Fisheries, PO Box 21, Cronulla, NSW 2230 Australia. Manuscript approved for publication 24 June 2003 by Scientific Editor Manuscript received 13 July at NMFS Scientific Publications Office. Fish. Bull. 101:910-914 (2003). NOTE Chan et al.: Identification of Carcharhinus spp, by DNA-based techiniques 911 morphism (PCR-RFLP; Martin, 1991), DNA sequencing techniques (Heist and Gold, 1999), isoelectric focusing of muscle proteins (Renon et al., 2001; Smith and Benson, 2001) and direct multiplex PCR amplification (Shivji et al., 2002). These techniques use the differences in the se- quences of nucleotide bases within a DNA strand among species. DNA techniques have high sensitivity, are easily reproduced, and allow the development of a unique "DNA fingerprint" for each species (Martin, 1991; Innes et al., 1998; Pepperell and Grewe, 1999). It was the aim of this project to initiate a shark DNA database for species iden- tification of pelagic sharks (beginning with species from the genus Carcharhinus) in New South Wales by using PCR-RFLP techniques. Materials and methods Sharks of the genus Carcharhinus landed by recreational fisheries and caught in NSW beach meshing were identified to species level by using morphometric taxonomic guides and dentition identification (Cliff and Wilson, 1994; Last and Stevens, 1994; Naylor and Marcus, 1994). Positively identified sharks were retained as voucher specimens (see "Acknowledgments" section). A tissue biopsy (5-10 g) from the dorsal region on either side of all voucher specimens and unidentified sharks was taken and stored in 75% etha- nol prior to DNA extraction. Mitochondrial DNA (mtDNA) of specimens from six species of the genus Carcharhinus (C. brachyurus, C. brevipinna, C. falciformis, C. leucas, C. limbatus, and C ohscurus; see Table 1 for sample sizes) was extracted by using a Fastprep DNA Extraction kit (BIOlOl, Integrated Sciences, Sydney, New South Wales) following the manufacturer's instructions. Approximately 200-300 mg of tissue was placed in a sterilized 1.5-mL eppendorf tube and after the addition of 1 mL Fastprep lysis buffer, incubated at 56°C for three hours prior to the extraction stage (Chan, 2001). Following the extraction procedure of the Fastprep protocol, quality and quantity of DNA was measured by using a GeneQuant DNA/RNA calculator ( Amersham Biosciences, Sydney, New South Wales). The polymerase chain reaction (PCR) was used to am- plify the 1146 nucleotide base-pair (bp) cytochrome b (cyt h) region of the mtDNA (Martin and Palumbi, 1993; Kita- mura et al., 1996). For each 50 pL PCR reaction, 100-200 ng of template mtDNA was used, 1.5 niM MgCU, IX PCR buffer, 2 mM of dNTP, 5 mM of each external primer (5'-TGACTTGAARAACCAYCGTTG-3' and 3'-CTCCAG- TCTTCGRCTTACAAG-5') and two units of DyNAzyme EXT DNA polymerase (Finnzymes, GeneWorks, Adelaide, South Australia) were added to a sterilized 200 pL PCR tube. The PCR was undertaken in a MJR MiniCycler (MJR, GeneWorks) with a heated bonnet on a cycle of 94°C for three minutes, followed by 35 cycles of 94°C denaturing for 45 seconds, 48°C annealing for 30 seconds, 72°C exten- sion for 90 seconds, and a final 10-minute extension time at 72°C (Chan, 2001). To determine if the cyt 6 region was successfully ampli- fied, 10 pL of PCR product was added to 2 pL loading dye (25% bromophenol blue, 40% sucrose in distilled water) and loaded into wells of a 1.5% w/v agarose gel submerged in 0.5X TBE (Tris-borate-EDTA, pH 8) buffer, with 1 pg of a 100-bp DNA ladder (Sigma-Aldrich, Sydney, New South Wales) added to 5 pL distilled water added to each side- end well. The gel was subject to electrophoresis at 125 V for 45-60 minutes and then stained in ethidium bromide for 10 minutes, de-stained in fresh distilled water for 20 minutes prior to illumination under ultraviolet (UV) light to determine the success and yield of the amplification. To obtain species-specific profiles, restriction fragment length polymorphism (RFLP) was used on the entire 1146 bp cyt b fragment (Martin, 1991; Chan, 2001). The suc- cessful amplified reaction products had the primers, taq polymerase, and buffer chemicals removed by using a BRESAspin PCR purification kit (GeneWorks). For each RFLP reaction, 30 pL of purified PCR-amplified cyt b mtDNA (300-1000 ng; 30 pL of distilled water was used for control reactions), 5 pL of lOX buffer, one unit of a re- striction enzyme (Alu l,Hae III, Ps? I, Taq I,andXAo I) and distilled water up to 50 pL volume was added to a steril- ized 200 pL PCR reaction tube and incubated at 37°C for 2 hours in the MJR minicycler (with heated bonnet), with the exception of Taq I (incubated at 65°C for two hours in a water bath). After the allotted digestion time, 10 pL of loading dye was added to each tube prior to loading into a 1.5% w/v agarose gel submerged in 0.5XTBE buffer In both end wells, 5 pL of distilled water -i- 1 pg of a 100 bp DNA ladder was added. The gel was subject to electrophoresis at 125 V for 60-90 minutes, stained in ethidium bromide for 10 minutes, and destained in fresh distilled water for 20 minutes prior to illumination under UV light. Enzyme- digested DNA fragments >100 bp were then "scored" to the nearest 25 bp based upon migration of the DNA fragment (the smaller the fragment, the faster the migration) and recorded for each enzyme and sample (Martin, 1991) by using the 100-bp DNA ladder as a standard measuring guide for size estimation. Results and discussion PCR-RFLP profiles were successfully developed for six spe- cies of the genus Carcharhinus; distinct and discrete pat- terns were observed for each species with five restriction enzymes (Table 1, and Chan, 2001). The only intraspecific polymorphism observed was for two specimens of C. bre- vipinna with the Xho I restriction enzyme. Increasing the sample size of all species may identify more intraspecific polymorphisms. Because of the relatively small sample sizes, no statistical analyses were undertaken. Other restriction enzymes were tested (Chan, 2001), and with the possible inclusion of other species from the genus Car- charhinus into this database in the future, these restriction enzymes may be required in order to discern the additional species. Because some of the fragment sizes were rounded to the nearest 25 bp, the sum of the fragments for a restric- tion enzyme of a species may be more than 1146 bp, the size of the C3rt b uncleaved region for sharks (Martin and Palumbi, 1993). Fragments <100 bp were not recorded because the DNA ladder had a lower limit of 100 bp. 912 Fishery Bulletin 101(4) Table 1 Summary of PCR-RFLP banding patterns for the cytochrome b (cyt 6) region n Carch arhinus spp. Fragment sizes are given in number of base pairs (bp) and have been rounded to the nearest 25 bp. Where the enzyme appeared not to have cleaved the cyt h | region, it was scored ' 1146 " n = denotes sample size. / = denotes fragment size present Carcharhinus species C. brevipinna C brevipinna Fragment C.l imhatus C. brachvurus C. leucas C. obscurus C. falciformis Haplotype 1 Haplotype 2 Enzyme size (n=9) (;i=12) (n=3) ( /!=29) (I 1 = 12) (;!=6) (n=2) Alu I 1000 700 600 500 450 350 300 200 / / • • / / / / / • • Hae III 1100 975 750 225 / / / • / / / / 175 / / / / / Pst\ 1146 975 175 / / • / / / / / Taql 1146 1100 850 650 325 300 / / / / / / / / / Xho I 1146 850 325 / / / / • / / These techniques can be used to complement morphomet- ric identification (Chff and Wilson, 1994; Last and Stevens, 1994; Naylor and Marcus, 1994) or can be used to identify "cryptic" species when morphological identification cannot be done. Other "cryptic" species caught in beach meshing and by recreational fisheries can be added to the DNA da- tabase, such as hammerhead sharks (Sphyrna spp.) which are commonly caught and are recorded in catch records to genus level only (Pepperell, 1992; Reid and Krogh, 1992; Chan, 2001). Although this project positively identified six species from the genus Carcharhinus, other species of this genus are known to inhabit the NSW coastline (Ste- vens, 1984; Last and Stevens, 1994). During the warmer months, when the northern currents extend farther south to the Sydney region, transient tropical Carcharhinus spp. may appear off the coast. In the northern regions of NSW, there have been recorded catches of the blacktip reef shark (C melanopterus) by shore-based anglers (Gartside et al., 1999). Although transient tropical whaler sharks may not have permanent stocks in NSW waters, it is important to discern them from resident Carcharhinus spp. prior to any species-specific stock assessment. Given the number of shark species and difference in life histories (Last and Stevens, 1994; Smith et al., 1998), identification to species level is crucial. The use of genetic techniques allows, for the first time, accurate identification of species of whaler sharks that were landed by recreational fisheries and caught in pro- tective beach meshing in NSW and that have been his- torically recorded to genus level. Continual sampling and formal identification are required for comparison of catches between species oi Carcharhinus. Genetic techniques have the potential to be used for all other shark species and fisheries within the Australian Fishing Zone (AFZ). The use of genetic techniques has been employed in the field of law enforcement to prevent the selling of protected fish species at local fish markets where the majority of the carcass is not retained (Ward et al., 1999). This use could NOTE Chan et a\. Identification of Carcharhinus spp by DNA-based techniques 913 be extended to ensure that protected shark species such as the grey nurse shark (Carcharias taurus), white shark (Carcharodon carcharias), and the smalltooth sand tiger shark (Odontaspis ferox) are not sold. This project is the first time that Carcharhinus spp. have been formally identified to species level in the 60-year his- tory of NSW protective beach meshing and only the second time in NSW recreational fisheries after Stevens (1984). The depositing of voucher specimens and all DNA biopsies at the Australian Museum ensures that these valuable and irreplaceable biological samples can be used in future re- search. It is evident that DNA techniques can be used to taxonomically identify "cryptic" specimens, especially Car- charhinus spp., and Sphyrna spp. to species level that were once recorded to genus level only in many fisheries based in NSW (Pepperell, 1992; Reid and Krogh, 1992; Chan, 2001; Tanner and Liggins^). It is important that sharks that are caught be recorded to the lowest taxonomic level for management and conservation strategies. Long-term routine sampling and recording to species level will provide useful data on which conservation management strategies can be developed as part of the Australian national plan of action for the conservation and management of sharks. Acknowledgments The authors would like to acknowledge the assistance of the NSW Game Fishing Association and all NSW recre- ational gamefishing clubs, their officials and their anglers who cooperated with the research, NSW protective beach meshing contractors and observers, NSW Fisheries staff, and the numerous volunteers (Joanne Bennett, Tanya Compton, Rikke Dano, Paul Godfrey, Gary Henry, Andrew Hodges, Alex Irwin, Jeff Murphy, Julie Needham, Milena Rantala, and Clint Wilson) who helped collect samples. We thank Ed Heist and Andrew Martin for their comments on the manuscript and specially thank Bill Sherwin (UNSW) and Marie Roseline Yardin for their assistance in this project. This project was funded by NSW Fisheries and the National Heritage Trust Coast and Clean Seas' Marine Species Protection Program (CCS Project no. 9856). Voucher shark specimens were retained at the Australian Museum, Sydney, NSW, Australia (Collection Manager, Fish Section) and NSW Fisheries, NSW, Australia (Dennis Reid). Literature cited Chan, R.W. K. 2001. Biological studies on sharks caught off New South Wales. Ph.D. diss., 314 p. School of Biological Science, Univ. New South Wales, Sydney, Australia. Cliff, G., and R. B. Wilson. 1994. Natal Sharks Board's field guide to sharks and other marine animals, 57 p. Group Editors, Durban, South Africa. Dudley, S. R J. 1997. A comparison of the shark control programs of New South Wales and Queensland (Australia) and KwaZulu- Natal (South Africa). Ocean Coast. Manag. 34:1-27. Gartside, D. F., B. Harrison, and B. L. Ryan. 1999. An evaluation of the use of fishing club records in the management of marine recreational fisheries. Fish. Res. 41:47-61. Godfrey, P. 1997. Identification of sharks caught in NSW waters using allozyme electrophoresis. B.S. (Hons.) thesis, 54 p. Cen- tre for Marine and Coastal Studies, Univ. New South Wales, Sydney, Australia. Heist, E. J., and J. R, Gold. 1999. Genetic identification of sharks in the U.S. Atlantic large coastal shark fishery. Fish, Bull. 97:53-61. Innes, B. H., P. M. Grewe, and R. D. Ward. 1998. PCR-based genetic identification of marlin and other billfish. Mar. Freshw. Res. 49:383-388. Kitamura, T., A. Takemura, S. Watabe, T. Taniuchi, and M. Shimizu. 1996. Mitochondrial DNA analysis for the cytochrome 6 gene and D-loop region from the bull shark Carcharhinus leucas. Fish. Sci. 62:22-27. Last, P. R., and J. D. Stevens. 1994. Sharks and rays of Australia, 513 p. CSIRO (Com- monwealth Scientific and Industrial Research Organisa- tion) Publishing, Melbourne, Australia. Lavery, S, 1992. Electrophoretic analysis of phylogenetic relationships among Australian carcharhinid sharks. In Sharks: biology and fisheries (J. G. Pepperell, ed.), p. 97-108. Aust. J. Mar Freshw. Res. 43. Martin, A. P. 1991. Application of mitochondrial DNA sequence analysis to the problem of species identification of sharks. In Con- servation biology of elasmobranchs (S. Branstetter, ed.), p. 53-59. NOAA Tech. Rep. NMFS 1 15. Martin, A. R, and S. R. Palumbi. 1993. Protein evolution in different cellular environments: cjftochrome b in sharks and mammals. Mol. Biol. Evol. 10: 873-891. Naylor, G. J. P., and L. F. Marcus. 1994. Identifying isolated shark teeth of the genus Carcha- rhinus to species: relevance for tracking phyletic change through the fossil record. Am. Mus. Novit. 94:1-53. Pepperell, J. G. 1992. Trends in the distribution, species composition and size of sharks caught by gamefish anglers off South-eastern Australia, 1961-90. In Sharks: biology and fisheries (J. G. Pepperell, ed.), p. 213-225. Aust. J. Mar Freshw. Res. 43. Pepperell, J. G., and P. M. Grewe. 1999. A field guide to Indo-Pacific billfishes, 16 p. CSIRO Publishing, Melbourne, Australia. Reid, D. D., and M. Krogh. 1992. Assessment of catches from protective shark mesh- ing off New South Wales beaches between 1950 and 1990. In Sharks: biology and fisheries (J. G. Pepperell, ed.), p. 283-296. Aust. J. Mar Freshw. Res. 43. Renon, P., M. M. Colombo, F. Colombo, R. Malandra, and PA. Biondi. 2001. Computer-assisted evaluation of isoelectric focusing patterns in electrophoretic gels: identification of smooth- hounds (Mustelus mustelus, Mustelus asterias) and com- parison with lower value shark species. Electrophoresis 22:1534-1538. Shivji, M., S. Clarke, M. Pank, L. Natanson, N. Kohler, and M. Stanhope. 2002. Genetic identification of pelagic shark body parts for 914 Fishery Bulletin 101(4) conservation and trade monitoring. Conserv. Biol. 16: 1036-1047. Smith, P. J., and P. G. Benson. 2001. Biochemical identification of shark fins and fillets from the coastal fisheries in New Zealand. Fish. Bull. 99; 351-355. Smith, S. E., D. W. Au, and C. Show. 1998. Intrinsic rebound potentials of 26 species of Pacific sharks. Mar. Freshw. Res. 49:663-678. Stevens, J. D. 1984. Biological observations on sharks caught by sport fishermen off New South Wales. Aust. J. Mar Freshw. Res. 35:573-590. Ward, R. D., R. K. Daley, J. Andrew, and G. K. Yearsley 1999. Protein fingerprinting. In Australian seafood handbook: an identification guide to domestic species (G. K. Yearsley P. R. Last, and R. D. Ward, eds.), chap. 9, p. 358-392. CSIRO Marine Research, Hobart, Australia. 915 Red sea urchins (Strongylocentrotus franciscanus) can live over 100 years: confirmation with A-bomb carbon Thomas A. Ebert Department of Zoology Oregon State University Corvallis, Oregon 97331-2914 E mail address eberttfn'sciences oregonstate edu John R. Southon Center for Accelerator Mass Spectrometry Lawrence Livermore National Laboratory Livermore, California 94551-9900 Red sea urchins (Strongylocentrotus franciscanus) along the west coast of North America, like most large sea urchins in temperate waters world- wide, are the focus of a commercially important fishery. In a review of bio- logical data for purposes of fishery management, the life span of red sea urchins was suggested to be 7-10 years (Sloan, 1986) and they have been included with much shorter-lived species for illustrating complex popula- tion dynamics (Hastings and Higgins, 1994). Recent work with tetracycline and calcein tagging (Ebert, 1998; Ebert et al., 1999), however, has shown that individuals continue to grow through- out life, although at a very slow rate, and large individuals are estimated to be in excess of 100 years old. A poten- tial problem with the studies using tetracycline and calcein is that one- year time intervals were used between tagging and recapture and therefore it is possible that occasionally there may have been very good years for growth that were missed. If occasional growth spurts occurred, largest sizes would have been attained in much less than 100 years. The potential problem of missed good years for growth could be resolved with a marker that captures a longer period of time. The accuracy of age estimates has consequences for resource management where size limits may need adjustment in order to protect older individuals (Hilborn and Walters, 1992; Congdon et al., 1994; Ebert, 1998). There is also the need to understand the evolution of life histo- ries of species where long life tends to be an indicator of uncertainty in indi- vidual reproductive success (Murphy, 1968; Roff 1992; Stearns, 1992). Enhanced radiocarbon in the oceans due to atmospheric testing of nuclear weapons that began in the 1950s (Ny- dal and Lovseth, 1983; Broecker et al., 1985, Duffy et al., 1995) provides a permanent marker in carbonate- based skeletal elements that are not reworked by resorption and deposition during growth and hence has a long time period between mark and recov- ery. The enhanced radiocarbon marker has been used in various studies to validate the periodic (usually annual) nature of growth zones in fish (Kalish, 1993, 1995; Campana, 1997; Campana et al., 2002) and invertebrates (Tureki- an et al., 1982; Witbaard et al.. 1994; Peck and Brey, 1996) where validation by chemical tags such as tetracycline has been impractical. Red sea urchins lack interpretable growth zones (Breen and Adkins, 1976) and therefore there is no natural feature to serve as a cross check for studies using chemical tags. In the present study we present a test and confirmation of age in red sea urchins estimated from tetracycline tagging using an enhanced ^*C signal in the ocean from nuclear weapons testing. Materials and methods Red sea urchins were tagged with tet- racycline from 1989 to 1992 in northern California, Oregon, and Washington and collected after time intervals of approximately one year (details pre- sented in Ebert et al., 1999). It is not possible to determine whether a live sea urchin has a tetracycline mark and therefore large collections had to be made. Skeletal elements were cleaned with sodium hypochlorite bleach to remove all organic material not bound in the calcite of the skeleton, and then skeletal ossicles were exam- ined by using UV illumination to detect the tetracycline marks, which fluoresce yellow. Growth increments were mea- sured in jaws of Aristotle's lantern of 1582 tagged-recovered red sea urchins and used to estimate growth parame- ters. Jaw ossicles, the demipyramids of Aristotle's lantern, are internal skeletal elements that grow around all surfaces but not equally in all directions so that a change in jaw length, AJ , is mostly at the end closest to the esophagus and there is little growth closest to the mouth, the labial end, where the teeth extend from the jaw. The Tanaka function (Eq. 1) was used to describe growth (Tanaka, 1982, 1988) because it can model data that show an initial lag, an exponential phase with a maximum, and can include continuing growth throughout life. This function is described in greater detail else- where (Tanaka, 1982, 1988; Ebert et al., 1999). The usual formulation of the Tanaka model is zisize as a function of size at time t and At is assumed to be fixed for all individuals in the sample, usually at zit = 1 year (Tanaka, 1982, 1988; Ebert, 1998; Ebert et al., 1999) and not included explicitly in the equa- tion. In the present study we estimated the amount of jaw that would have to be removed to represent the time span from the time of collection in the 1990s with relatively high i*C levels to the time before atmospheric testing of atomic bombs (relatively low '■'O and Manuscript approved for publication 10 July 200.3 by Scientific Editor. Manuscript received 13 July 2003 at NMFS Scientific Publications Office. Fish. Bull. 101(4):915-922 (2003). 916 Fishery Bulletin 101(4) therefore the Tanaka model was modified from previous uses to make AJ a function ofJf^^f the size on the date of recapture rather the date of marking, which is the usual way of estimating growth parameters. Also, At was explic- itly included as a variable (Eq. 2), A/ = y„ V7 In lG + 2^G-+fa where and G = EIA-falE-fAt E = exp{4fU„^-d}) (1) (2) (3) The three parameters of the Tanaka function, a, d, and f, have biological meaning: "a" is related to maximum growth rate, which is approximately l^/a; "d" shifts the size at which growth is maximum; and "/" is a measure of the rate of change of the growth rate. A graphical presentation of how changes in these parameters change the growth curve is given in Ebert et al. (1999). Explicit use of At and making AJ a function of Jf+^t required a modification of the usual presentation of the Tanaka function. In Ebert et al. (1999) Equation 2 was written as G = E/4-fa/E+f (4) with no At and with "+ /". Equation 3 was written as E = exp[4f{J,-d)). (5) with Jf, rather than J^^^^. Tetracycline tagging for a period of one year, At = 1, provides the Tanaka parameter esti- mates and these parameters were used to estimate a Ajaw size that would cover the time from the date of collection to a time. At, before A-bomb testing; At is time run backwards from the date of collection, which is the reason for the sign change from Equation 4 to Equation 2. The samples of red sea urchins that were selected for radiocarbon analysis were part of the tagging study at Halftide Rocks off San Juan Island, Washington (Ebert et al., 1999). Individuals were tagged with tetracycline on 26 October 1991 and collected again on 21 October 1992. The recaptured tagged individuals {n=365) are part of the 1582 tagged sea urchins from northern California, Oregon, and Washington that were used to estimate Tanaka param- eters. For ''*C analysis, specimens were selected from the Halftide Rocks collection that did not show fluorescence in the skeleton and therefore probably had not been handled in 1991. The use of untagged individuals for radiocarbon analyses avoids any possible contamination from handling and tagging in 1991. Cleaned jaws for ''*C analysis were cemented to alumi- num blocks with a two-part epoxy cement and aligned so that the esophageal margin was approximately parallel with the block base. The block was held on the stage of a small milling machine and the stage tilted so that the jaw was as parallel as possible with the milling bit. Approxi- mately 0.5 mm of the jaw surface was removed and sides were milled to remove recently deposited calcite and to ex- pose the underlying older skeleton. The jaw was measured and successive samples were milled from the esophageal edge to a depth of 0.5 mm, which produced samples larger than 1 mg of carbonate in most cases. Samples were placed in individual reaction chambers, evacuated, acidified with orthophosphoric acid, and heated. The evolved CO2 was converted to graphite by reduction with an excess of hydro- gen in individual reactors with iron powder as a catalyst (Vogel et al., 1987). Analysis of "C in the graphite targets was done at the Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, and reported as A^'^C7cc (Stuiver and Polach, 1977), which includes a correc- tion for a S^^C of -3 based on stable isotope analyses. Mean precision (1 standard deviation) of radiocarbon measure- ments was 4.2%o (range: 3.0-7.9). Results Of the 1582 tag recoveries from all sites, 739 jaws showed a growth increment, AJ, of <0.02 cm and of these only 13 had a labial measurement >0, which is at the end of the jaw at the mouth opening. The smallest nonzero measurements were 0.001 cm and therefore growth less that this was recorded as 0; 54 sea urchins in the sample had clear tetra- cycline marks but 0 measurable growth. For large jaws, the measured labial component was too small to be measured and therefore all of the calculated AJ since the late 1950s was milled from the esophageal end of the jaw only. Tetracycline tagging indicated that annual jaw growth (Fig. 1) was very slow for large sea urchins and many in- dividuals showed annual increments of less than 0.01 cm. The resulting growth curve of jaw length as a function of age (Fig. 2A) showed that at least some large individuals would be expected to have ages in excess of 100 years. If this age estimate is correct, a drop in ^^C should be found in successive small slices removed from large jaws, which would first show current ^^C levels and then drop to pre- bomb levels. Because the Halftide Rocks samples were col- lected in 1992 we used At = 35 years, which would go back to 1957. Using Equations 1-3, growth parameters given in Fig. 1, and At = 35 years, we estimated the increment to be between 1 and 2 mm for jaws between 2.5 and 2.6 cm (Fig. 2B). Successive milled samples from the esophageal ends of large jaws (Fig. 3, A-D) showed a precipitous drop in radiocarbon to prebomb levels over 1-2 millimeters, in agreement with predictions. Variations across replicates and samples probably are the result of differences in the width of milled samples and an inability to remove all re- cently deposited calcite or to follow the exact growing edge of the jaw with the milling machine. Smaller jaws (Fig. 3, E-G) were not expected to show a prebomb signature, and indeed they did not. They do, however, indicate the ^*C level to be expected in recent skeletal material and emphasize the rapid change in radiocarbon shown in large jaws. Changes in ^•*C in successive milled samples in jaws NOTE Ebert and Southon: Confirmation of longevity for Strongy/ocentrotus franascanus with '""carbon 917 0.06 ■ 0 04' ? = 0 03- t B 0 02- i^ □o » " ° o fejj^ ft •¥ » % ■(. 001 ■ ^ ^m ^\ 0 nil ■ ^?vS Jf^^f^trr — 2.5 3.0 Figure 1 Jaw growth increments, AJ, for tagged red sea urchins iStrongylocen- trotus franciscanus) from northern California, Oregon, and Washing- ton; fitted Hne is the Tanaka function (Tanaka, 1982, 1988; Ebert et al., 1999) v/ithf= 10.95650 ±0.35064 SElstandard error), d = 0.04937 ±0.01664, and a = 8.63029 ±0. 16659; M is approximately 1 year for all samples; (A) entire data set; n = 1582; (B) restricted scale to show just jaws larger than 2.0 cm and degree of scatter; n = 336; conversion to 1 981 7 body diameter, D, from jaw length, J, is D =4.8951J ; therefore jaw lengths of 2.0 and 2.7 cm would have expected test diameters of 11.9 and 17.9 cm, respectively. E-G are similar to changes shown in coral samples from the Galapagos (Guilderson and Schrag, 1998) and may in- dicate that '■^C levels in surface waters in regions of strong upwelling were still rising when sea urchin were collected in 1992. The conclusion is that '''C analysis supports the age estimates based on tetracycline tagging and use of the Tanaka function: large red sea urchins are old and may have ages of 100 years or more Discussion The largest reported red sea urchins, with body diam- eters over 19 cm, are from British Columbia, Canada, (Bureau, 1996) and with estimated jaw lengths of about 2.8 cm would be expected to be around 200 years old (Fig. 2A), Age estimates of lOO-i- years far exceed estimates of life span for other sea urchins (Table 1) based on growth lines in ossicles. Natural growth lines, however, tend to underestimate ages of old individuals because very small increments will have alternations of dark and light areas that are difficult or impossible to resolve and hence counts underestimate age (Ebert, 1988). For example, the maxi- mum age estimate for Strongylocentrotus droebachiensis, the commercial species of the U.S. east coast, is 25 years by counts of growth lines (Robinson and Maclntyre, 1997) but at least twice this if tagging and size structure (Russell et al., 1998) are used. Similarly, tagging and size structure of Evechinus chloroticus (Lamare and Mladenov, 2000) have indicated survival rates similar to S. franciscanus but the maximum number of growth lines reported was only 10 (Dix, 1972). Survival rate, however, is not a fixed parameter for a species and there is local variation, as well as geographic patterns, evident in the survival rate for S. franciscanus (Ebert et al. 1999). Estimates of annual survival rates based on growth pa- rameters and mean size for red sea urchins from southern California to Alaska (Ebert et al., 1999) indicate that very old individuals would not be expected in southern Califor- nia where few individuals attain ages of 50 years. At more northern locations, the probability of long life increases (Fig. 4) and ages of lOO-i- are expected, particularly in Washington and Alaska. The mechanism causing the lati- tudinal pattern are unclear. Latitudinal differences in sur- vival may be due to increased disease outbreaks associated with higher temperatures in the south (Ebert et al. 1999) or the presence of more predator species in the south (Tegner, 2001). Physiological senescence related to temperature is unlikely because there is no pattern to growth differences associated with latitude (Ebert et al., 1999) and no evidence for physiological decline in relative gonad size in the south (Tegner and Levin, 1983) or north (Kramer and Nordin, 1975). The largest individuals continue to develop gonad masses in accord with the same allometric relationships as smaller individuals. It is reasonable to conclude that senescence does not occur in red sea urchins. 918 Fishery Bulletin 101(4) 28 24 ? 2° S 16- ^ 12 S en -> 0,8 i, 0.4 00 20 40 60 80 100 Age (yr) 120 140 160 2.4 2.5 Jaw length (cm) 2.7 Figure 2 (A) Growth curve for jaws of red sea urchins from northern California, Oregon, and Washington usingTanaka parameters and an initial jaw size of 0.44 cm, the approximate jaw size for a 1-year-old red sea urchin; (B) change in jaw size, Ajaw, with Atime = 35 years starting with a final jaw length; Ajaw is the estimate of how much of the jaw would have to be removed to expose prebomb calcite for sea urchins collected in 1992. Table 1 Maximum age estimates for sea urchins based on growth zones in skeletal ossicles. Species Years Reference Lytechinus variegatus 4 Beddingfield and McCUntock (2000) Strongylocentrotus nudus 6 Kawamura (1966) Psammechinus miliaris 7 Jensen (1969) Paracentrotus lividus 8 Crapp and Willis (1975) Sphaerechinus granulans 9 Lumingas and Guillou (1994) Evechmus chloroticus 10 Dix(1972) Psammechinus miliaris 10 Gage (1991) Strongylocentrotus intermedlus 10 Agatsuma(2001) Echinus acutus var. norvegicus 11 Gage etal. (1986) Loxechinus albus 11 Gebauer and Moreno (1995) Echinus esculentus 12 Nichols etal. (1985) Allocentrotus fragilis 15 Sumich and McCauley (1973) Echinus elegans 21 Gage etal. (1986) Strongylocentrotus droebachiensis 24 Robinson and Maclntyre (1997) Echinus affinis 28 Gage and Tyler (1985) Sterechinus neumayeri 40 Brey etal. (1995) Sterechinus antarcticus 75 Brey (1991) Red sea urchins larvae spend at least two months in the plankton (Strathmann, 1978) during which time they can be carried far along the coast or out to sea. There is year- to-year variation in settlement and recruitment and years of zero success and greater variation at northern sites (Ber- nard and Miller, 1973, Low, 1975,Tegner and Dayton, 1981, Duggins, 1983, Pearse and Hines, 1987, Sloan et al., 1987, Ebert et al., 1994). An important point, however, is that these authors reported some recruitment at study sites and so extreme longevity would at first seem unnecessary for species survival. The important issue for evolution of life histories, however, is not whether some individuals recruit to the population but how successful an individual is each year in leaving offspring. The long life of adult red sea urchins emphasizes the difficulties individuals have in successfully having offspring that settle in suitable habitat and survive to reproductive age. Many annual reproductive episodes appear to be required to succeed and therefore red sea urchins are classic bet hedgers that use resources to promote annual survival of adults as well as to reproduce (Stearns, 1992), Attributes of a long life span have consequences for resource management. The implications for management of the red sea urchin resource have been explored by us- NOTE Ebert and Southon: Confirmation of longevity for Stmngylocentrotus franciscanus with '''carbon 919 -130 130 0.0 0-5 1.0 1.5 2.0 2,5 0.0 0.5 1.0 1.5 2.0 2.5 Jaw length (cm) Figure 3 Change in J C %c with successive milled samples from the esophageal end of red sea urchin jaws collected at Halftide Rocks, Washington (48°28.8'N, 1220°59.8'W), 21 Oct. 1992. Step width in lines is the width of a milled sample. Jaws of three sea urchins are drawn but a total of 7 were sampled; jaw lengths are A = 2.58 cm, B = 2.54, C = 2.53, D = 2.51, E = 2.29, F = 2.01, G = 1.80 cm; primes represent replicate jaws from the same sea urchin; jaw C is labeled just at the final milled sample (initial sample is just below the initial sample for jaw Al; bands labeled 1 through 4 on jaw G show how samples were milled from the jaw after the surface layer had been removed by approximately 0.5 mm. Shaded areas of the jaw, which were not sampled, indicate an area below the top surface where a shelf exists for articulation with the epiphysis, another ossicle of Aristotle's lantern. ing elasticity analysis (de Kroon et al., 1986) of a matrix model and have shown that small changes in survival of individuals larger than 9 cm would have a greater effect on population maintenance than survival of smaller sea urchins (Ebert, 1998). The conclusion from matrix analysis, which is supported by the '''C test of growth and age of Strongylocentrotus franciscanus we present in our study, is that the preservation of large individuals must be included in long-term management plans for this species as well as for other long-lived sea urchins in developing fisher- ies such as that for Evechinus chloroticus (Barker, 2001). Finally, our work strongly suggests that life spans of other exploited sea urchm species should be explored in greater detail in developing management plans because preserva- tion of large and old individuals may be very important for the long-term viability of these fisheries. Acknowledgments Tagging and processing sea urchins were done in collabora- tion with S. Schroeter and J. Dixon, with support from the Pacific States Fishery Commission, a self-imposed landing tax of sea urchin fishermen administered by the Calif Dept. Fish & Game, Oregon Sea Urchin Community Com- mission, and Ore. State Univ. Sea Grant; field work was facilitated by resource managers in California, Oregon, and Washington (P. Kalvass, N. Richmond, A. Bradbury). Stable isotope analysis was done at the University of California at Davis and radiocarbon analysis was funded by a Center for Accelerator Mass Spectrometry minigrant at Lawrence Livermore National Laboratory and was carried out under the auspices of the US DOE. The manuscript benefited from a critical reading by G. Fox. 920 Fishery Bulletin 101(4) 1 1 ,01 1 Q. .001 1 .0001 .00001 50 80 Age (yr) 100 Figure 4 Probabilities of attaining ages of 50. 80, and 100 years for red sea urchins (Strongylocentrotus franciscanus) at sites from Alaska to northern California; estimates were based on Tanaka growth parameters and size- frequency distributions (Ebert et al., 1999); numbers of sites are shown at the tops of bars and there are two samples at each site; error bars are 2SEs. Literature cited Agatsuma, Y. 2001. Eco\ogy o{ Strongylocentrotus intermedius. In Ed- ible sea urchins: biology and ecology (J. M, Lawrence, ed.), p. 333-346. Developments in aquaculture and fisheries science, no. 32. Elsevier Science B. V., Amsterdam, Neth- erlands. Barker, M. F. 2001. The ecology of Evechinus chloroticus. In Edible sea urchins: biology and ecology (J. M. Lawrence, ed.), p. 245-260. Developments in aquaculture and fisheries science, no. 32. Elsevier Science B. V., Amsterdam, Neth- erlands. Beddingfield, S. 0., and J. B. McClintock. 2000. Demographic characteristics of Lytechinus variega- tiis (Echinoidea: Echinodermata) from three habitats in a north Florida bay Gulf of Mexico. Mar. Ecol. 21:17-40. Bernard, E R., and D. C. Miller. 1973. Preliminary investigation on the red sea urchin resources of British Columbia [Strongylocentrotus fran- ciscanus (Agassiz)). Fish. Res. Board Can. Tech. Rep. 400: 1-37. Breen, P. A., and B. E. Adkins. 1976. Growth rings and age in the red sea urchin, Strongy- locentrotus franciscanus. Fish. Res. Board Can. Manuscr. Rep. Ser 1413. Brey T 1991. Population dynamics o{ Sterechinus antarcticus (Echinodermata: Echinoidea) on the Weddell Sea shelf and slope, Antarct. Antarctic Sci. 3:251-256. Brey.T., J. Pearse, L. Basch, J. McClintock, and M. Slattery 1995. Growth and production of Sterechinus neumayeri (Echinoidea: Echinodermata) in McMurdo Sound, Antarct. Mar Biol. 124:279-292. Broecker, W. S.. T.-H. Peng, G. Ostlund, and M. Stuiver 1985. The distribution of bomb radiocarbon in the ocean. J. Geophys. Res. 90:6953-6970. Bureau, D. 1996. Relationship between feeding, reproductive condition, jaw size and density in the red sea urchin, Strongylocentro- tus franciscanus. M.S. thesis, 90 p. Simon Eraser Univ., Burnaby, Canada. Campana, S. E. 1997. Use of radiocarbon from nuclear fallout as a dated marker in the otoliths of haddock Melanogrammus aeglifinus. Mar Ecol. Prog. Ser 150:49-56. Campana, S. E., L. J. Natanson, and S. MyklevoU. 2002. Bomb dating and age determination of large pelagic sharks. Can. J. Fish. Aquat. Sci. 59:450-455. Congdon, J. D., A. E. Dunham, and R. C. van Loben Sels. 1994. Demographics of common snapping turtles iChelydra serpentiana): implications for conservation and manage- ment of long-lived organisms. Am. Zool. 34:397-408. Crapp, G. B., and M. E. Willis. 1975. Age determination in the sea urchin Paracentrotus lividus (Lamarck), with notes on the reproductive cycle. J. Exp. Mar Biol. Ecol. 20:157-178. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67: 1427-1431. Dix, T G. 1972. Biology of Evechinus chloroticus (Echinoidea: Echi- nometridae) from different localities. 4. Age, growth, and size. N. Z. J. Mar Freshw. Res. 6:48-68. Duffy, P. B., D. E. Eliason, A. J. Bourgeois, and C. C. Covey. 1995. Simulation of bomb radiocarbon in two global ocean general circulation models. J. Geophys. Res. 100:22545- 22563. NOTE Ebert and Southon: Confirmation of longevity for Strongylocentrotus franclscanus witfi '''carbon 921 Duggins, D. O. 1983. Starfish predation and the creation of mosaic patterns in a kelp-dominated community. Ecology 64:1610-1619. Ebert, T. A. 1988. Calibration of natural growth lines in ossicles of two sea urchins, Strongylocentrotus purpuratus and Echinome- tra mathaei, using tetracycline. In Echinoderm biology: proceedings of the sixth international echinoderm confer- ence (R. D. Burke, P. V. Mladenov, P. Lambert, and R. L. Parsley, eds.), p. 435-443. A. A. Balkema, Rotterdam, Netherlands. 1998. An analysis of the importance of Allee effects in management of the red sea urchin Strongylocentrotus franciscanus. In Echinoderms: San Francisco. Proceed- ings, 9th international echinoderm conference ( R. Mooi and M. Telford, eds.), p. 619-627. A. A. Balkema, Rotterdam, Netherlands. Ebert, T. A., J. D. Dixon, S. C. Schroeter, P. E. Kalvass, N. T. Richmond, W. A. Bradbury, and D. A. Woodby 1999. Growth and mortality of red sea urchins Strongylo- centrotus franciscanus across a latitudinal gradient. Mar Ecol. Prog. Ser 190:189-209. Ebert, T. A., S. C. Schroeter, J. D. Dixon, and P. E. Kalvass. 1994. Settlement patterns of red and purple sea urchins {Strongylocentrotus franciscanus and S. purpuratus) in Cahfornia, USA. Mar Ecol. Prog. Ser 111:41-52. Gage, J. D. 1991. Skeletal growth zones as age-markers in the sea urchin Psammechinus miliaris. Mar Biol. 110:217-228. Gage, J. D., and PA. Tyler 1985. Growth and recruitment of the deep-sea urchin Echi- nus affinis. Mar Biol. 90:41-53. Gage, J. D., P A. Tyler, and D. Nichols. 1986. Reproduction and growth of Echinus acutus var nor- vegicus Duben & Koren and E. elegans Duben cfe Koren on the continental slope off Scotland. J. Exp. Mar Biol. Ecol. 101:61-83. Gebauer, P., and C. A. Moreno. 1995. Experimental validation of the growth rings of i-o.r- echmus albus (Molina, 1782) in southern Chile (Echinoder- mata: Echinoidea). Fish. Res. 21:423-435. Guilderson, T P. and D. P Schrag. 1998. Abrupt shift in subsurface temperatures in the tropi- cal Pacific associated with changes in El Nino. Science 281:240-243. Hastings, A., and K. Higgins. 1994. Persistence of transients in spatially structured eco- logical models. Science 263:1133-1136. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics, and uncertainty, 570 p. Chapman & Hall. New York, NY. Jensen, M. 1969. Breeding and growth of Psammechinus miliaris (Gmehn). Ophelia 7:65-78. Kalish,J. M. 1993. Pre- and post-bomb radiocarbon in fish otoliths. Earth Planet. Sci. Lett. 114:549-5.54. 1995. Radiocarbon and fish biology. In Recent develop- ments in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds.), p. 637-653. Univ. South Carolina Press, Columbia, SC. Kawamura, K. 1966. On the age determining character and growth of a sea urchin, Strongylocentrotus nudus. Sci. Rep. Hokkaido Fish. Exp. Station 6:56-61. Kramer, D. E., and D. M. A. Nordin. 1975. Physical data from a study of size, weight and gonad quality for the red sea urchin (Strongylocentrotus francis- canus (Agassiz)) over a one-year period. Fish. Res. Board Can. Manuscr Rep. Ser 1372:68. Lamare, M. D., and P. V. Mladenov. 2000. Modelling somatic growth in the sea urchin Evechins chloroticus (Echinoidea: Echinometridae). J. Exp. Mar Biol. Ecol. 243:17-43. Low, C. J. 1975. The effect of grouping of Strongylocentrotus francis- canus, the giant red sea urchin, on its population biology. Ph.D. diss., 157 p. Univ. British Columbia, Vancouver, Canada. Lumingas, L. J. L., and M. Guillou. 1994. Growth zones and back-calculation for the sea urchin, Sphaerechinus granulans, from the Bay of Brest, France. J. Mar Biol. Assoc. U K. 74:671-686. Murphy, G. L 1968. Pattern in life history and the environment. Am. Nat. 102:391-411. Nichols, D., A. A. T. Sime, and G. M. Bishop. 1985. Growth in populations of the sea-urchin Echinus esculentus L. (Echinodermata: Echinoidea) from the Eng- lish Channel and Firth of Clyde. J. Exp. Mar. Biol. Ecol. 86:219-228. Nydal, R., and K. Lovseth. 1983. Tracing bomb "C in the atmosphere 1962-1980. J. Geophys. Res. 88:3621-3621. Pearse, J. S., and A. H. Hines. 1987. Long-term population dynamics of sea urchins in a central California kelp forest: rare recruitment and rapid decline. Mar Ecol. Prog. Ser 39:275-283. Peck, L.S., and TBrey 1996. Bomb signals in old Antarctic brachiopods. Nature 380:207-208. Robinson, S. M. C, and A. D. Maclntyre. 1997. Aging and growth of the green sea urchin Bull. Aquacul. Assoc. Can. 91:56-60. Roff, D. A. 1992. The evolution of life histories, 535 p. Chapman & Hall, New York, NY. Russell, M. P, T A. Ebert, and P S. Petraitis. 1998. Field estimates of growth and mortality of the green sea urchin, Strongylocentrotus droebachiensis. Ophelia 48:137-153. Sloan, N. A. 1986. Red sea urchin. Underwater world. DFO/2322 UW/ 53, 4 p. Fisheries and Oceans Canada, Ottawa, Canada. Sloan, N. A., C. P. Lauridsen, and R. M. Harbo. 1987. Recruitment characteristics of the commercially har- vested red sea urchin Strongylocentrotus franciscanus in southern British Columbia, Canada. Fish. Res. 5:55-69. Stearns, S. C. 1992. The evolution of life histories, 249 p. Oxford Univ Press, New York, NY. Strathmann, R. 1978. Length of pelagic period in echinoderms with feeding larvae from the northwest Pacific. J. Exp. Mar Biol. Ecol. 34:23-27. Stuiver, M., and H. A. Polach 1977. Discussion: reporting of "C data. Radiocarbon 19: 355-363. Sumich, J. L., and J. E. McCauley 1973. Growth of a sea urchin, Allocentrotus fragilis, off the Oregon Coast. Pac. Sci. 27:156-167. 922 Fishery Bulletin 101(4) Tanaka, M. 1982. A new growth curve which expresses infinite increase. Publ. Amakusa Mar Biol. Lab. 6:167-177. 1988. Eco-physiological meaning of parameters of ALOG growth curve. Publ. Amakusa Mar Biol. Lab. 9:103-106. Tegner, M. J. 2001. The ecology of Strongylocentrotus franciscanus and Strongylocentrotus purpuratus. In Edible sea urchins: biol- ogy and ecology (J. M. Lawrence, ed.), p. 307-332. Devel- opments in aquaculture and fisheries science, no. 32. Elsevier Science B. V., Amsterdam, Netherlands. Tegner, M. J., and P. K. Dayton. 1981. Population structure, recruitment and mortality of two sea urchins (Strongylocentrotus franciscanus and S. purpuratus) in a kelp forest. Mar Ecol. Prog. Ser 5: 255-268. Tegner, M. J., and L. A. Levin. 1983. Spiny lobsters and sea urchins: analysis of a predator- prey interaction. J. Exp. Mar Biol. Ecol, 73:125-150. Turekian, K. K., J. K. Cochran,. Y. Nozaki, L Thompson, and D. S. Jones. 1982. Determination of shell deposition rates of Arctica islandica from the New York Bight using natural '--'Ra and 228'pi5 gfjjj bomb produced '■*€. Limnol. Oceanogr. 27: 737-741. Vogel, J. S., J. R. Southon, and D. E. Nelson. 1987. Catalyst and binder effects in the use of filamentous graphite for AMS. Nucl. Instru. Methods Phys. Res. Sec. B 29:50-56. Witbaard, R., M. L Jenness, K. van der Borg, and G. Ganssen. 1994. Verification of annual growth increments in Arctica islandica L. from the North Sea by means of oxygen and carbon isotopes. Neth. J. Sea. Res. 33:91-101. 923 Abundance and distribution of cetaceans in outer continental shelf waters of the U.S. Gulf of Mexico Gregory L. Fulling Keith D. Mullin Carrie W. Hubard Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 3209 Frederic Street Pascagoula, Mississippi 39567 E-mail address (for G. L Fulling) Greg Fulling(a)noaa, gov The U.S. Marine Mammal Protection Act (MMPA) requires that stocks of marine mammal species in U.S. waters be maintained at or above their opti- mum sustainable population (OSP) level, defined as the number of animals that results in the maximum net pro- ductivity. To meet this requirement for each stock, the U.S. National Marine Fisheries Service (NMFS) estimates annual human-caused mortality and potential biological removal (PBR), the maximum number of animals that may be removed from a stock due to human activities (e.g. fisheries bycatch) while allowing the stock to reach or maintain its OSP PBR is calculated by follow- ing specific criteria and using the estimated abundance of the stock, its maximum net productivity rate (theo- retical or estimated), and a recovery factor (Barlow et al., 1995; Wade and Angliss, 1997). The NMFS is required to prepare an annual stock assessment report (SAR) for each stock to update abundance, stock structure, maximum net productivity, human-caused mor- tality, PBR, and status (e.g. Waring et al.,2001). Cetaceans in the U.S. Gulf of Mexico (U.S. GOM) occur in two species assem- blages that overlap in upper continen- tal slope waters (-200-1000 m). The oceanic waters (>200 m) are routinely inhabited by 20 species that, in most cases, inhabit deep warm-temperate to tropical waters throughout the world. Bottlenose dolphins iTursiops trim- catus) and Atlantic spotted dolphins [Stenella frontalis) are the only two species commonly found in continen- tal shelf waters (<200 m) (Mulhn and Hansen, 1999). In the U.S. GOM the distribution of T. fruncatiis ranges from inshore waters to deep waters of the continental slope (Blaylock and Hoggard, 1994; Hansen et al.'; Mullin and Hoggard'^). In the U.S. GOM, the NMFS divides T. trun- catus into 38 stocks: 33 inshore stocks (bays, sounds, and estuaries); 3 coastal stocks (western, northern, and eastern) from shore to 9 km seaward of the 18-m (10-fm) isobath; 1 outer continental shelf (OCS) stock from the coastal boundary to 9 km seaward of the 183-m (100-fm) isobath; and 1 continental shelf edge and slope stock from the OCS boundary out to the U.S. Exclu- sive Economic Zone (FEZ) (Waring et al., 2001). The abundance estimate for the OCS T. truncatus stock is 50,247 dolphins (CV=0.18) and is based on aerial surveys conducted during fall which covered all the U.S. GOM shelf waters over 3 years in sections, west, central, and east, in 1992, 1993, and 1994, respectively (Blaylock and Hog- gard, 1994; Waring et al., 2001). One U.S. GOM S. frontalis stock is recognized, and the abundance, 3213 dolphins (CV=0.44), is estimated from ship surveys of shelf edge and oceanic waters >100 m deep conducted from 1991-94 (Hansen et al.M. Abundance estimates for S. frontalis for the U.S. GOM OCS were not made from the 1992-94 aerial surveys although S. frontalis groups were sighted (War- ing et al., 2001). The majority of S. frontalis are thought to inhabit the shelf-edge region. However, data from opportunistic sightings (e.g. Mills and Rademacher, 1996) and a summer 1994 ship survey of the eastern GOM (Hof- stetter, 2002) have indicated that they are common throughout eastern GOM shelf waters >10 m deep, and in oceanic waters <500 m. The NMFS Southeast Fisheries Sci- ence Center (SEFSC) conducts annual spring and fall ichthyoplankton sur- veys in the U.S. GOM. The spring sur- vey targets the entire oceanic portion of the U.S. GOM, and the fall survey focuses on shelf waters from the U.S.- Mexico border to southern Florida. Since 1991, abundance estimates of oceanic cetacean species in the U.S. GOM have been based primarily on data collected during annual spring surveys (Hansen et al.'; Mullin and Hoggard'^; Mullin and Fulling^). Be- cause of the lack of current assessment information on and the uncertainty of abundance estimates for T. truncatus and S. frontalis in OCS waters, ceta- cean surveys were conducted during the fall ichthyoplankton surveys from 1998 to 2001. From these surveys, we report the abundance and distribution of cetaceans in OCS waters (20-200 m deep) ofthe U.S. GOM. 1 Hansen, L. J., K. D. Mullin, and C. L. Roden. 1995. Unpublished report. Es- timates of cetacean abundance in the northern Gulf of Mexico from vessel sur- veys, 20 p. Southeast Fisheries Science Center, 3209 Frederic St., Pascagoula, MS .39.567. 2 Mullin, K.D., and W. Hoggard. 2000. Vi- sual surveys of cetaceans and sea turtles from aircraft and ships. In Cetaceans, sea turtles and seabirds in the northern Gulf of Mexico: Distribution, abundance and habitat associations. Volume II: Tech- nical report (R.W. Davis, W. E. Evans, and D. Wursig, eds.), p. 111-172. OCS Study MMS 96-0027. Minerals Management Service, Gulf of Mexico OCS Region, New Orleans, LA. 70123. 3 Mullin, K.D., and G.L. Fulling. 2003. Un- published report. Abundance of ceta- ceans in the oceanic northern Gulf of Mexico, 1996-2001, 35 p. Southeast Fisheries Science Center, 3209 Frederic Street, Pascagoula, MS 39567. Manuscript approved for publication 10 July 2003 by Scientific Editor. Manuscript received 20 July 2003 at NMFS Scientific Publications Office. Fish. Bull. 101:923-932 (2003). 924 Fishery Bulletin 101(4) i 28- Figure 1 Survey effort in Beaufort sea state of <3 (dark lines), east (1342 km) and west (2202 km) of Mobile Bay, Alabama (bgld vertical line), in the northern U.S. Gulf of Mexico outer continental shelf (20-200 m) during fall 1998-2001. The 20- and 200-m isobaths (thin lines) are shown. Methods Study area The study area (245,800 km^) included continental shelf waters of the U.S. GOM between the U.S-Mexico border and Key West, Florida, between the 20- and 200-m isobaths ( Fig. 1). However, survey effort did not extend south of 26.0°N in the southeastern GOM and therefore abundance estimates were extrapolated for this region. The shelf is wide (up to 200 km) off the Florida peninsula and off northern Texas and Louisiana, and narrower off the Florida Panhandle near DeSoto Canyon, the Mississippi River Delta, and southern Texas. The continental slope is a steep escarp- ment from 1000 to 2000 m in the eastern GOM. Survey design and data collection Surveys were conducted from the 68-m NOAA Ship Gordon Gunter (1998, 1999, and 2001) and the 52-m NOAA Ship Oregon II (2000). The four surveys ranged from 28 to 32 days between 28 August and 2 October and were divided into two legs of 12 to 19 days. Standard ship-based, line- transect survey methods for cetaceans, similar to those used in the Pacific Ocean and U.S. GOM, were used (e.g. Barlow, 1995; Hansen et al.'*). Surveys were conducted 24 hours a day along a predetermined trackline between plankton stations uniformly spaced 30 nmi apart. The trackline uniformly covered the shelf waters roughly 10-200 m deep in 1998-2001 (Fig. 1). ■• Hansen, L. J., K. D. MuUin, T. A. Jefferson, and G. P. Scott. 1996. Visual surveys aboard ships and aircraft. In Distribu- tion and abundance of marine mammals in the north-central and western Gulf of Mexico: Final report. Volume II: Techni- cal report (R.W. Davis and G.S. Fargion, eds.), p. 55-132. OCS Study MMS 96-0027. Minerals Management Service, Gulf of Mexico OCS Region, New Orleans, LA. 70123. Data were collected by two teams of three observer — one team positioned on the flying bridge 14.5 m above the wa- terline (Gunter) and the other team positioned 9.2 m above the waterline (Oregon II) during daylight hours while the vessels moved between plankton stations, weather permit- ting (i.e. no rain, Beaufort sea state <6). Each team had at least two members experienced in ship-based, line-transect methods and in identification of tropical cetaceans. The left- and right-side observers searched to the horizon in the arc from 10° right and left of the ship's bow to the left and right beams (90°), respectively, using 25x binoculars. The third observer searched, using unaided eye or 7x hand-held binoculars, and recorded data. Observers changed position every 30-40 minutes, and the two teams alternated 2-h watches throughout daylight hours. Survey speed was usu- ally 18 km/h (-10 knots) but varied with sea conditions. Data were recorded on a computer interfaced with a glob- al positioning system (GPS) by an in-house BASIC data acquisition program (Southeast Fisheries Science Center, NMFS, Pascagoula, MS). For each cetacean sighting, time, position, bearing and reticle (a measure of radial distance) of the sighting, species, group-size, behavior, bottom depth, sea surface temperature, and associated animals (e.g. sea- birds, fish) were recorded. The bearing and radial distance for groups sighted without 25x binoculars and close to the ship were estimated. Survey effort data were automatically recorded every 2 minutes and included the ship's position and direction, effort status, observer positions, and envi- ronmental conditions that could affect the observers' abil- ity to sight animals (e.g. Beaufort sea state, sun position). Typically, if a sighting was within a 5.5-km strip on either side of the ship, the ship was diverted from the trackline to approach the group to allow the observers to identify species and estimate group-size by consensus. Cetaceans were identified to the lowest taxonomic level possible. Observers' ability to make identifications depended on weather and animal behavior. Differences between T. truncatus and S. frontalis could not always be NOTE Fulling et al : Abundance and distribution of cetaceans in the U.S. Gulf of Mexico 925 distinguished at long distances and were therefore some- times recorded as "T. truncatus + S. frontalis." Analytical techniques Survey effort that was parallel to the bathymetry gradi- ents, occurred in waters outside the OCS study area, or occurred in a Beaufort sea state >4 was excluded from analyses (Fig. 1). Survey effort used in analyses is sum- marized in Table 1. Survey effort was not uniformly dis- tributed throughout the study area due to poor survey conditions, particularly in the eastern GOM, during two of the four years. Because S. frontalis sightings were clearly more numerous in the east, we delineated the study area into west (106,186 km^) and east (139,614 km^) regions at 88°1.5.0'W (ca. Mobile Bay, Alabama) and estimated abun- dances separately for each region. A combination of line- transect and strip-transect methods were used to make abundance estimates. Line-transect methods were used for sightings detected with 25( binoculars, which constituted the majority of sightings (129/140). Strip-transect methods were used for the 11 sightings that were made without the 25x binoculars (naked-eye sightings) and that were observed by the primary team. Line-transect estimates For each species or species group (i) [i.e. T. truncatus, S. frontalis, rough-toothed dolphins iSteno bredanensis) and T. truncatus+S. frontalis] detected by 25x binoculars, and for each region (J) (east and west), abundance estimates were made with line-transect methods (Af^, j) by using the software program DISTANCE (Colorado Coop. Fish and Wildlife Research Unit, Colorado State Univ., Fort Collins, CO) (Laake et al., 1993; Buckland et al., 2001 ) and by mcor- porating data into the following equation: A',, 2L-g{0) (1) where A^ = area of region j; n^ = number of group sightings of species; in region 7; ^Li I ~ niean group size of species i in region j; f (0) = sighting probability density function at per- pendicular distance zero for species i; L = total length of transect line in region j; and g(0) = probability of seeing a group on the transect line. The parameter g(0) was not estimated; g(0) = 1 was used for each abundance estimate. Abundances were negatively biased because observers usually miss some groups at the surface on the transect line, and some groups were under the surface while in the observation area, therefore g(0) <1 (see "Discussion" section). The log-normal 95% confidence interval was computed (Buckland et al., 2001) for each abundance estimate because it was a product of estimates and tended to have a skewed distribution. The variance of ^Li ^^^ estimated by using Table 1 Total survey efTort (km) during 1998- -2001 in waters 20-200 m and under Beaufort sea state conditions <3. Year West East Total 1998 174 67 241 1999 477 120 597 2000 281 0 281 2001 448 629 1077 Total 1380 816 2196 var(A',,^) = A'-,^ var(/i,, ) var(S,, ^l. ) van - + [/(O)] f,(or (2) The sampling unit was the length of the transect completed on-effort each day with Beaufort sea state <3 in a region. The formula used to estimate each component of the vari- ance is given in Buckland et al. (2001). Varin^ , ^) was length-weighted and based on the variation in the number of on-effort group sightings between sampling units that ranged up to 191 km/d. Estimation of f(0) The perpendicular distance (y ) was estimated by using bear- ing and reticle measurements. The reticle readings were converted to radial sighting distances {R) by the method of Lerczak and Hobbs ( 1998;y=R sin(6), where 6=angle between the sighting and the transect line). Because of the difference in observer height (5.3 m) between the Oregon II and Gunter, each ship could potentially yield a different sighting function, g(x). However, only seven sightings were made in sea states <3 from the Oregon II during the one year it was used; there- fore data from both ships were pooled. Estimates of/'/O) were made by using a hazard-rate, uniform, or half-normal model with exact perpendicular sighting distances and no adjust- ments. Model selection was determined by using Akaike's Information Criterion (AIC; Buckland et al., 2001). The number of S. bredanensis groups and the number of T. truncatus+S. frontalis groups sighted was insufficient to estimate /ID) for each. Because the S. bredanensis group and T. truncatus+S. frontalis group had similar sighting characteristics (e.g. body size, group-size, surface behavior), we pooled them with sightings of T. truncatus to estimate f'iO). Total number of sightings for both T. truncatus and S. frontalis was sufficient to estimate fiO) for each without pooling with other species. Truncation for T. truncatus, S. bredanensis, and T. truncatus + S. frontalis was 3300 m, and was 5000 m for S. frontalis. Each estimate of /"/O) was based on pooled sightings from the east and west regions. Estimation of mean group-size Group-sizes tend to be related toy, because in many cases larger groups are easier to see than small groups with 926 Fishery Bulletin 101(4) increasing^'. In general, the arithmetic mean of group-size may be an overestimate of the true mean group-size and could lead to positively-biased abundance estimates. There- fore, a regression of group-size by j' was used to estimate an "expected mean group-size" (program DISTANCE) and it was used if the regression was significant (P<0. 15). VarCS^^) was the analytical variance for mean group-sizes based on arithmetic means or was estimated as in Buckland et al. (2001:74) for expected mean group-sizes. Strip-transect estimates One requirement for unbiased line-transect estimates of abundance is that the cetacean group should not move in response to the ship before it is sighted (Buckland et al., 2001). If cetaceans are not sighted before they respond to the ship, in cases of attraction to the ship, /TO) and abundance will be overestimated. During previous U.S. GOM surveys, groups of T. truncatus or S. frontalis were consistently attracted to ride the bow waves as the ship approached (Wiirsig et al., 1998). Therefore, the abundance and variance of groups sighted by naked eye (Ng) were estimated by N., 2-L-w, and var(/V„J = N=.^ var(/i^,J ^ ^ar(-^5,.;) (3) (4) where w^ = l//",(0) which was treated as a constant, i.e. strip width, w^ , was equal to the line-transect effec- tive strip half-width [l//](0)] with vardu^) = 0. For each region, species total abundance (Nj.^ ) was the line-transect and strip-transect estimates added, iV^.^ = Nj^, j + Ns,,j. Total U.S. GOM OCS abundance for each species was Nj,^ = 27V j., ^. The coefficient of variation (CV) for each abundance was estimated as CViN) = [vartAOl'^^W and the CV for each summed abundance as CV(/V,„„) ■(I CV'{N)N r/1 N. (5) Results Abundance estimates were based on 2196 km of effort and 140 sightings (Figs. 1 and 2). For east and west regions, there was 816 km of effort and 73 sightings, and 1380 km of effort and 67 sightings, respectively (Tables 1 and 2). Only three cetacean species were encountered. Groups of J! trun- catus (30 east region, 45 west region) and S. frontalis (34 east, 12 west) were the most frequently encountered (Fig. 2, Table 2) and S. bredanensis groups ( 1 east, 2 west) were also sighted. Tiirsiops truncatus and S. frontalis were esti- mated to have flO) of 0.6238/km (CV=0.12) and 0.4101/km (CV=0.11), and an effective strip half-width of 1603 and 2438 m, respectively (Figs. 3 and 4). Steno bredanensis and T. truncatus+S. fron talis abundances were based on an /! 0 ) = 0.6059/km(CV=0.11) and an effective strip half-width of 1650 m. Mean group-sizes (from 25x binocular sightings) of T. truncatus for east (9.8, 0.25) and west (10.0, 0.18) regions were similar (Table 2), and had an overall range of 1-68 animals. The mean group size of S. frontalis was larger in the east (24.3, 0.19) than the west (15.6, 0.21) with an overall range of 1-267 animals. Group-sizes of S. bredanen- sis were 8, 11, and 20 animals. The east mean group-size for both T. truncatus and S. frontalis is the size-biased or expected mean group-size because the expected mean was significantly smaller that the arithmetic mean, 10.9 (P=0.14) and 31.9 (P=0.08), respectively The most abundant species (number of individuals; CV) found in U.S. GOM OCS waters was S. frontalis (30,772; 0.27); the vast majority (91%) occurring in the east (27,997; 0.29). The density of S. frontalis was about eight times greater in the east compared to the west (20.1 and 2.6 dolphins/100 km^, respectively). The abundance of T. truncatus was 25,320 (0.26); there was greater abundance in the east (15,198; 0.34) than m the west (10,122; 0.29) but with similar densities (10.9 and 9.5 dolphins/100 km^, respectively). The total OCS abundance of S. bredanensis was 1238 (0.65), and that of 7: truncatus+S. frontalis, 1868 (0.37). Discussion Both T. truncatus and S. frontalis occur in northern GOM waters outside the OCS (i.e. waters <20 m or >200 m). About 23,000 T. truncatus inhabit inshore and coastal waters (<20 m) (Waring et al., 2001) and nearly 3000 occur in oceanic waters (Mullin and Fulling-). Both the "coastal" and "offshore" ecotypes of T! truncatus (Hersh and Duffield, 1990) occur in the northern GOM (LeDuc and Curry 1998). How these ecotypes are distributed in the northern GOM and western North Atlantic is being investigated from skin biopsy samples collected, in part, during the 1998-2001 OCS surveys. Using mitochondrial DNA, obtained from biopsy samples collected during a U.S. Atlantic ship survey, Torres et al. (2003) reported no offshore form was sampled within 6 km of shore and no coastal from was sampled beyond 39 km from shore or in waters >34 m deep. Forty- seven percent (35/75) of the GOM OCS T. truncatus groups were in waters >34 m deep. Ship surveys of northern GOM waters indicate that very few S. frontalis (<500 animals) occur in oceanic waters, and those that do are usually found close to the shelf edge in waters <500 m deep (Davis et al, 1998; Mullin and Full- ing2). The smaller "offshore" or "Gulfstream" S. frontalis that occurs in parts of the oceanic Atlantic (Perrin, 2002) has not been recorded for the northern GOM. During the 1998-2001 surveys, S. frontalis was sighted in waters <20 m deep. However, because sampling was not perpendicular to bathymetry, abundance estimates were not calculated. This species is not known to occur in U.S. GOM inshore waters (Mullin and Hansen, 1999). NOTE Fulling et al.: Abundance and distribution of cetaceans in the US Gulf of Mexico 927 3 28- 30- 28- 26- 98 96 Figure 2 Locations of all on-effort sightings of Turswps truncatus (top), Stenelta frontalis (center), Steno bredanensis, and T. truncatus+S. frontalis (bottom) in the northern U.S. Gulf of Mexico outer continental shelf (20-200 m) during fall 1998-2001. Numbers of sightings shown are prior to truncation. The 20- and 200-m isobaths (thin lines) are shown (AL=Alabama, FL=Florida, LA=Louisiana, MS=Mississippi, TX=Texas). Abundance The abundance estimates for cetaceans reported in the present study are the first ship-based estimates for the U.S. GOM OCS. Abimdance estimates for T. truncatus on the OCS (25,320; 0.26) are half the estimate in the pre-2002 SARs (50,247; 0.18) (e.g. Waring et al, 2001), that were based on aerial surveys conducted during fall 1992-94 (Blaylock et al, 1994; Waring et al, 2001). The abundance estimate for S. frontalis for the entire U.S. GOM in SARs prior to 2002 (3,213; 0.44) was based on data from ship surveys of OCS and oceanic waters >100 m deep (Waring et al., 2001; Hansen et al.'). Our current abundance estimate of S. frontalis (30,772; 0.27) for the OCS is almost an order of magnitude larger During the 1991-94 aerial surveys, there were 13 sight- ings of S. frontalis groups and 10 sightings that were iden- tified as T. truncatus+S. frontalis in OCS waters (SEFSC, NMFS, Pascagoula, MS, unpubl. data). Using these sight- ings and 139 T. truncatus sightings to estimate /(O), we esti- mated the abundance of S. frontalis from the aerial survey data to be 14,866 ( 0.37) for the U.S. GOM OCS [west, 3,526 (0.86); east, 11,340(0.40)]. There are several potential reasons for the differences in abundances of the two species from ship and aerial surveys. The U.S. GOM OCS east of 85.5°W makes up about 44% of the U.S. GOM OCS. Aerial survey abundance estimates in this area were based on a small number of transect lines grouped in two places and most of the area was not 928 Fishery Bulletin 101(4) Table 2 Group-size, density and abundance estimates of cetaceans in the northern U.S. Gulf of Mexico outer continental shelf (waters 20-200 m deep) during fall 1998- -2001 (n = number of group sightings, 5= :mean group-size, Z)=animals/100 km- Niabundance esti- mate, CV=coefficient of variation . R=reticl e sightings and K=naked eye sightings . Species and stratum ;; S CV(S) D N cv^N) 95"r CI T\irsiops truncatus East-R 27 9.8 0.25 10.1 14,132 0.40 6426-31,082 East-K 3 6.7 0.70 0.8 1066 0.85 139-8182 East total 30 10.9 15,198 0.38 7442-31,035 West-R 41 10.0 0.18 9.2 9786 0.30 5502-17,406 West-K 4 3.5 0.34 0.3 336 0.67 94-1201 West total 45 9.5 10,122 0.29 5790-17,696 OCS total 75 10.3 25,320 0.26 15,457-41.478 Stenella frontalis East-R 32 24.3 0.19 19.5 27.226 0.30 15,09.3-49,113 East-K 2 11.0 0.09 0.6 771 0.55 252-2358 East total 34 20.1 27,997 0.29 15,978-49,057 West-R 11 15.6 0.21 2.6 2712 0.42 1192-6169 West-K 1 4 — 0.6 63 1.17 9-433 West total 12 2.6 2775 0.41 1279-6023 OCS total 46 12.5 30.772 0.27 18,418-51,412 Steno bredanensis East-R 0 — — — 0 — — East-K 1 11 — 0.4 586 0.85 118-2902 East total 1 0.4 586 West-R 2 14 0.43 0.6 652 0.98 115-3715 West-K 0 — — — 0 — — West total 2 0.6 652 OCS total 3 0.5 1238 0.65 384-3990 T. truncatiis+S. frontalis East-R 8 2.4 0.22 0.7 983 0,57 324-2983 East-K 0 — — — 0 — — East total 8 0.7 983 West-R 8 4.8 0.28 0.8 885 0.47 355-2207 West-K 0 — — — 0 — — West total 8 0.8 885 OCS Total 16 0.8 1868 0.37 920-3793 surveyed (see Fig.l in Baumgartner [1997]). Complete coverage would have certainly led to more S. frontalis sightings and it is possible the lines that were surveyed were in areas with more T. triuicatus. Blaylock and Hog- gard (1994) estimated from aerial surveys that about 31% of the T. truncatus in OCS waters west of Mobile Bay were in rather a small area from the Mississippi River Delta west to about 90.5°W. Our ship survey effort in this area was small and resulted in only one sighting of T! truncatus (Fig. 2). Therefore, our ship-based estimates may have un- derestimated the abundance ofT. truncatus in the western OCS. Aerial abundances were based on survey lines that extended from 9.3 km past the 18 m (10 fm) curve to 9.3 km past 183 m ( 100 fm) curve; therefore the area surveyed was somewhat different than our 20-200 m OCS study area for ship surveys. Aerial survey effort in waters >200 m may have resulted in more sightings ofT. truncatus than S. frontalis because the deeper waters are not the common habitat of S. frontalis (Mullin and Fulling'^) and sightings in waters <20 m would have also been biased toward T. truncatus. Stenella frontalis and T. truncatus are similar in length and shape. Stenella frontalis are born without spots and become progressively more spotted with age, but young ani- mals look very similar to T. truncatus (see Herzing, 1997). Therefore, depending on the composition of the group, from a distance S. frontalis are not always easily distinguished from T. truncatus: therefore it is possible that some groups were misidentified as T. truncatus during aerial sui-veys, leading to bias in the relative abundance of each species. The annual PBR for the OCS stock of T. truncatus was 432 dolphins, and for the U.S. GOM stock of S. frontalis, NOTE Fulling et a\ Abundance and distribution of cetaceans in the US Gulf of Mexico 929 1 27713 1 14942 1 0217 Tui slops truiicatus B 089399 CD n 2 0766277 Q. c 9 0 638564 o OJ ■S 0 510851 Q \ . 0 383138 0 255426 0 127713 \. 0 500 1000 1500 2000 2500 3000 3500 Perpendicular distance in meters Figure 3 Plot of the detection funct ion of pooled sightings ofTursiops truncatus in the northern U.S. Gulf of Mexico. Stenella frontalis 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Perpendicular distance in meters Figure 4 Plot of the detection function of pooled sightings of Stenella frontalis in the northern U.S. Gulf of Mexico. 23 dolphins (Waring et al., 2001). Using the abundances, we estimated that the annual PBR would be 204 dolphins for T. truncatus and 246 dolphins for S. frontalis (Table 2). Although these changes in both PBRs are large, the annual fishery-related mortality and serious injury for each spe- cies is estimated to be <3 dolphins in the U.S. GOM OCS (Waring etal., 2001). they were achieved after four years of effort. In cases where there is human-caused mortality in a cetacean stock, abun- dance estimates with a CV < 0.50 are generally required to avoid incorrectly classifying a cetacean stock as "stra- tegic" under the U.S. MMPA (i.e. annual human-caused mortality>annual PBR) less than 10% of the time (Wade and DeMaster, 1999). Precision The precision of the abundance estimates for T. truncatus (CV=0.26) and S. frontalis (CV=0.27) was good, although Bias The surveys were designed to meet the assumptions of line- transect theory (Buckland et al., 2001). However, the abun- 930 Fishery Bulletin 101(4) dance estimates were negatively biased because the central assumption that all cetacean groups on the transect line are detected (i.e.^(G)=l), certainly was not met, and data were not collected to correct estimates for perception and avail- abihty bias (Marsh and Sinclair, 1989). Barlow (1995) esti- mated perception bias in a ship survey in the Pacific Ocean, and although the group-sizes were not estimated at close range, the majority of groups missed by the primary team were apparently small groups. From this. Barlow (1995) estimated g(0) to range from 0.73 and 0.79 for small groups of delphinids (<21 animals). Delphinids have relatively short dive-cycles but diving synchrony among members of a group can affect availability bias; if dives are asynchronous, the probability that at least one animal will be at the surface increases with group-size. Because availability bias varies by species due to differences in individual dive cycles, group diving behavior, and group-sizes, we were not able to address this potential bias based on Barlow's ( 1995) results. The use of the effective strip half-width [l//",(0)] from the 25x binocular sightings for the strip width for the strip-transect estimates (Table 2) was assumed to be con- servative and somewhat negatively biased. The distance from which animals will come to the ship to ride the bow is unknown and variable, depending on factors such as the animals' previous behavior, number of bowriding op- portunities, and the type of ship. If the strip width was too narrow, the strip-transect estimates of abundance would be positively biased. Our abundance estimates were for the entire U.S. OCS, but the surveys did not extend south of 26.0°N in the eastern Gulf Sightings from a 1994 survey of the eastern Gulf (Hoffstetter, 2002) indicated that the distribution of T. truncatus and S. frontalis does not change dramatically between 26.0°N and Key West; therefore we believe this potential bias is minimal. Because our estimates are from four combined years, an- other source of bias would occur if there were annual shifts in cetacean distribution, that is, if the majority of animals of any species occurred in a different part of the OCS in one year during fall compared to others years. However, there was no indication that this variation in distribution occurred and therefore potential bias is probably minimal. Potential bias due to the seasonality of the survey is also possible but cannot currently be addressed. Additionally, survey effort from the 2001 cruise was the most complete effort of all years and may have carried more weight than all the other cruises. However, the 2001 survey provided adequate eastern GOM coverage. Variable survey effort in the fall is common because tropical weather can create rough sea conditions. Additionally, fall surveys al- ways began in the west and terminated in the east. Because the same cruise track was always followed, we rarely had the opportunity to survey those areas not surveyed previ- ously during nighttime transit, and thus may have created both a spatial and temporal bias. Distribution The observed distributions of both T. truncatus and S. frontalis were not surprising given previous descriptions of their distributions. The greater number of S. frontalis in the U.S. GOM off Florida compared to the western GOM was suggested by Schmidley and Melcher (1974), and the distribution of sightings reported by Mills and Rademacher (1996) supported this finding. The density of S. fi-ontalis was much greater in the eastern GOM OCS than the west- ern GOM OCS but the density of T! truncatus was similar in the two regions (Table 2). The West Florida Shelf and Texas-Louisiana Shelf are very different marine environments, but how habitat dif- ferences specifically affect cetacean density patterns is not clear. The oceanography of the U.S. GOM continental shelf is complex, variable both spatially and temporally, and dif- ficult to characterize briefly. Nevertheless, there are some clear distinctions between eastern and western OCS. First, there are 3415 active oil and gas platforms in the U.S. GOM OCS (0-200 m); the vast majority of these platforms (with their attendant boat and helicopter traffic) occur in waters west of Mobile Bay (MMS^). Also, -95% of the U.S. GOM fisheries landings by weight occur west of Mobile Bay (10 years of NMFS^ data). Additionally, sediment- and nutri- ent-laden fresh water from the Mississippi River and its distributary, the Atchafalaya River, usually moves west and predominately affects the Texas-Louisiana and Mississippi- Alabama shelves. The bottom of the Texas-Louisiana Shelf is primarily clay-slit mud and sand, and that of the West Florida Shelf is a mosaic of sand, gravel, shell, and coral (Rabalais et al., 1999). Primary production associated with the Mississippi River outfiow is the highest measured in the GOM (Lohrenz et al., 1999). However, productivity on the West Florida Shelf can be enhanced by a variety processes (e.g. Gilbes et al., 1996). The deep eastern GOM is subject to the quasi-annual incursion of the Loop Current, which can extend to the Mississippi-Alabama Shelf (Wiseman and Sturges, 1999). This incursion can lead to upwelling episodes along the Loop Current front that may increase productivity along the shelf edge and on the West Florida Shelf (Paluszkiewicz et al., 1983; Gilbes et al., 1996). Baumgartner et al. (2001) reported greater sighting rates of cetaceans in the eastern GOM shelf-edge and oceanic wa- ters and suggested that greater feeding opportunities may occur because of the influence of the Loop Current. Griffin and Griffin (2003), whose study included coastal waters (<20 m), reported that S. frontalis on the West Florida Shelf was found in deeper, more saline, and less turbid water than those where T. truncatus was found. Demersal fish (e.g. sciaenids) are abundant and diverse on the western GOM OCS, but less abundant on the east- ern OCS (Darnell et al"; Darnell et al.^). The known prey of ^ Mineral Management Service, Gulf of Mexico Region website: http://www.gomr.mms.gove/hompg/fastfactsAVaterDepthAVater Depth.html. [Accessed on 7/8/2003.] ^ National Marine Fisheries Service web site: http://www. st.nmfs.gov/stl/commercial/. [Accessed on 8 July 2003.] ■ Darnell, R. M., R. E. Defenbaugh, and D. Moore. 1983. North- western Gulf shelf bio-atlas; a study of the distribution of demersal fishes and penaeid shrimp of the soft bottoms of the con- tinental shelf from the Rio Grande to the Mississippi River Delta. Open File Report No. 82-04, 438 p. Minerals Management Ser- vice, Gulf of Mexico OCS Region, New Orleans, LA 70123. NOTE Fulling et al : Abundance and distribution of cetaceans in the US, Gulf of Mexico 931 T. trunccitus from the GOM consist primarily of demersals, at least close to shore, but they also prey on pelagic spe- cies (Barros and Odell, 1990). The prey of S. frontalis are not well characterized but descriptions include epipelagic and mesopelagic fish and squid, and benthic invertebrates (Perrin, 2002). Richard and Barbeau ( 1994) observed "spot- ted dolphins" feeding on flyingfish (Exocoetidae) in waters 28-35 m deep on the West Florida Shelf This is not uncom- mon because S. frontalis have been routinely observed feed- ing on flyingfish at night during haulback of longline gear during NMFS fisheries assessment surveys (Grace^). Fertl and Wiirsig ( 1995) describe S. frontalis feeding on a school of small clupeid fish at the surface south of the Florida Panhandle. A S. frontalis satellite-tracked for 24 days off Texas stayed in waters 12-63 m deep (mean, 32.6 m) and 58.1% of its dives were <10 m (Davis et al., 1996). These shallow dives observed by Davis et al. may indicate feeding on epipelagic species. The occurrence of S. bredanensis in continental shelf waters of the U.S. GOM is interesting because this spe- cies is usually described as inhabiting oceanic waters (e.g. Jefferson, 2002). In the northern GOM, the estimated density of S. bredanensis was larger in OCS waters during fall (0.50 dolphins/100 km^; Table 2) than that estimated for oceanic waters during spring (0.32 dolphins/100 km'-) (MuUin and Fulling-). In fact, if there is no OCS-oceanic shift in distribution between spring and fall, there may be similar numbers of S. bredanensis in northern GOM shelf waters (1238; 0.65) as in oceanic waters (1231; 0.45). One of the groups sighted in OCS waters was near the shelf- edge 1 183 m) but the other two sightings were at depths of 31 m and 33 m off Texas (Fig. 2). The use of shelf waters in the U.S. GOM by this species may not be atypical; two sightings of S. bredanensis were made on the West Florida Shelf in waters <55 m deep during August 1994 (Hofstetter, 2002). Pitman and Stinchomb (2002) provide evidence that S. bredanensis may be specialized predators of dolphinfish [Coryphaena hippurus) in the Pacific Ocean. Dolphinfishes have a circumtropical distribution but occur in oceanic and shelf waters in the northern GOM commonly associated with Sargassiim and other drifting materials (Hoese and Moore, 1998). Steno bredanensis in the northern GOM are commonly found near flotsam, as they are in the Pacific — a place where dolphinfish tend to aggregate. The abundance estimates presented in this study are the first ship-based estimates of T. tursiops and S. fron- talis from Gulf of Mexico OCS waters. Although probably negatively biased, these estimates provide reliable data for the management of these species. Our results suggest that the diverse U.S. GOM environments provide an excellent natural experiment and opportunity to further understand » Darnell, R. M., J. A. Kleypas, and R. E. Defenbaugh. 1987. Eastern Gulf shelf bio-atlas; a study of the distribution of demersal fishes and penaeid shrimp of the soft bottoms of the continental shelf from the Mississippi River Delta to the Florida Keys. OCS Study 86-004 1 , 548 p. Mmerals Management Ser- vice, Gulf of Mexico OCS Region, New Orleans, LA 70123. ^ Grace, M. A. 2003. Personal commun. NOAA, 3209 Frederic Street, Pascagoula, MS 39567. the ecology of these sympatric cetacean species in OCS pe- lagic waters. Acknowledgments Many people made significant contributions to the success of the surveys including the officers and crews of NOAA ships Gordon Gunter and Oregon II and the field party chiefs. The marine mammal observers were H. Adams, N. Baertlein, C. Brown, J. Brusher, C, Burks, C. Gates, J. Con- tillo, L. Csuzdi, A. Debose, A. Beier-Engelhaupt, K. Maze- Foley, J. Henne, W. Hoggard, K. Hough, J. Litz, T. Martinez, M. Newcomer, C. Palmer, K. Rademacher, C. Roden, C. Sinclair, S. Stienessen, J. Tobias, CWH, KDM, and GLF. The comments of two anonymous reviewers improved the manuscript significantly. This work was conducted under Marine Mammal Research permit 779-1339 issued to the SEFSC and supported by Interagency Agreement 15958 between the NMFS, SEFSC and the Minerals Management Service, Gulf of Mexico Region. Literature cited Barlow, J. 1995. The abundance of cetaceans in California waters. Part I: Ship surveys in summer and fall of 1991. Fish. Bull. 93: 1-14. Barlow, J., S. L. Swartz, T. C. Eagle, and R R. Wade. 1995. U.S. marine mammal stock assessments: guidelines for preparation, background, and a summary of the 1995 assessments. NOAA Tech. Memo. NMFS-OPR-6:73. Barros, N. B., and D. K. Odell. 1990. Food habits of bottlenose dolphins in the southeastern United States. In The bottlenose dolphin (S. Leatherwood and R. R. Reeves, eds.), p. 309-328. Academic Press, San Diego, CA. Baumgartner, M. F. 1997. The distribution of Risso's dolphin (Grampus griseus) with respect to the physiography of the northern Gulf of Mexico. Mar. Mamm. Sci. 13:614-638. Baumgartner, M. F., K. D. Mullin, L. N. May, and T, D. Leming. 2001. Cetacean habitats in the northern Gulf of Mexico. Fish. Bull. 99:219-239. Blaylock, R. A., and W. Hoggard. 1994. Preliminary estimates of bottlenose dolphin abundance in the southern U.S. Atlantic and Gulf of Mexico continental shelf waters, 10 p. NOAA Tech. Memo. NMFS-SEFSC-356. Buckland, S. T, D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas. 2001. Introduction to distance sampling: estimating abun- dance of biological populations, 432 p. Oxford University Press, New York, NY. Davis, R. W., G. S. Fargion, N. May, T D. Leming, M. Baumgartner, W. E. Evans, L. J. Hansen, and K. D. Mullin. 1998. Physical habitat of cetaceans along the continental slope in the north-central and western Gulf of Mexico. Mar Mamm. Sci. 14:490-507. Davis, R. W., G. A. J. Worthy, B. Wiirsig, S. K. Lynn, and F I. Townsend. 1996. Diving behavior and at-sea movements of an Atlantic spotted dolphin in the Gulf of Mexico. Mar Mamm. Sci. 12:569-581. 932 Fishery Bulletin 101(4) Fertl, D.,andB. Wursig. 1995. Coordinated feeding by Atlantic spotted dolphins iStenella frontalis) in the Gulf of Mexico. Aquat. Mamm. 21{l):3-5. Gilbes, F., C. Tomas, J. J. Walsh, and F. E. Miiller-Karger. 1996. An episodic chlorophyll plume on the West Florida Shelf Cont. Shelf Res. 16:1201-1224. Griffin. R. B., and N. J. Griffin. 2003. Distribution, habitat partitioning and abundance of Atlantic spotted dolphins, bottlenose dolphins, and logger- head sea turtles on the eastern Gulf of Mexico continental shelf GulfMex.Sci. 21:23-34. Hersh, S. L., and D. A. Duffield. 1990. Distinction between northwest Atlantic offshore and coastal bottlenose dolphins based on hemoglobin profile and morphometry. In The bottlenose dolphin (S. Leatherwood and R. R. Reeves, eds.), p. 129-142. Academic Press, San Diego, CA. Herzing, D. L. 1997. The life history of free-ranging Atlantic spotted dol- phins (Stenella frontalis): age classes, color phases and female reproduction. Mar. Mamm. Sci. 13:576-595. Hoese, H. D., and R. H. Moore. 1998. Fishes of the Gulf of Mexico. Texas, Louisiana, and adjacent waters, 422 p. Texas A&M Univ. Press, College Station, TX. Hoffstetter, T. C. 2002. Distribution and abundance of marine mammals in the northeastern Gulf of Mexico relative to sea surface tem- perature and depth. M.Sc. thesis, 74 p. Univ. Southern Mississippi, Hattiesburg, MS. Jefferson, T. A. 2002. Rough-toothed dolphin, Steno bredanensis. In Ency- clopedia of marine mammals (W. F. Perrin, B. Wiirsig, and J. G. M. Thewissen, eds.), p. 1055-1059. Academic Press, San Diego, CA. LeDuc, R, G., and B. E. Curry. 1998. Mitochondrial DNA sequence analysis indicates need for revision of the genus Tursiops. Rep. Int. Whal. Comm. 47:393. Lerczak, J. A., and R. C. Hobbs. 1998. Calculating sighting distances from angular readings during shipboard, aerial, and shore-based marine mammal surveys. Mar Mamm. Sci. 14:590-599. Lohrenz, S. E., D. A. Wiesenburg, R. A. Arnone, and X. Chen. 1999. What controls primary production in the Gulf of Mexico? In The Gulf of Mexico large marine ecosystem (H. Kumpf, K. Steidinger, and K. Sherman, eds.), p. 151-170. Blackwell Science, Maiden, MA. Marsh, H., and D. F Sinclair 1989. Correcting for visibility bias in strip transect aerial surveys of aquatic fauna. J. Wildl. Manag. 53:1017-1024. Mills, L. R., and K. R. Rademacher. 1996. Atlantic spotted dolphins (Stenella frontalis) in the Gulf of Mexico. GulfMex.Sci. 14:114-120. Mullin. K. D., and L. J. Hansen. 1999. Marine mammals of the northern Gulf of Mexico. In The Gulf of Mexico large marine ecosystem (H. Kumpf K. Steidinger, and K. Sherman, eds.), p. 269-277. Blackwell Science, Maiden, MA. Paluszkiewicz, T, L. P. Atkinson, E. S. Posmentier, and C. R. McClain. 1983. Observations of a Loop Current frontal eddy intru- sion onto the West Florida Shelf J. Geophys. Res. 88: 9639-9651. Pitman. R. L., and C. Stinchcomb. 2002. Rough-toothed dolphins (Steno bredanensis) as predators of mahimahi (Coryphaena hippurus). Pac. Sci. 56: 447-450. Perrin, W. R 2002. Stenella frontalis. Mamm. Species 702:1-6. Rabalais, N. N., R. S. Carney, and E. G. Escobar-Briones. 1999. Overview of continental shelf benthic communities of the Gulf of Mexico. In The Gulf of Mexico large marine ecosystem (H. Kumpf, K. Steidinger, and K. Sherman, eds.), p. 171-195. Blackwell Science, Maiden, MA. Richard, K. R., and M. A. Barbeau. 1994. Observations on spotted dolphins feeding nocturnally on flying fish. Mar. Mamm. Sci. 10:473-477. Schmidley, D. J., and B. A. Melcher 1974. Annotated checklist and key to the cetaceans of Texas waters. Southwest. Nat. 18:453-464. Torres, L. G., P E. Rosel, C. D'Agrosa, and A. J. Read. 2003. Improving management of overlapping bottlenose dolphin ecotypes through spatial analysis and genetics. Mar. Mamm. Sci. 19:502-514. Wade, P R., and R. P Angliss. 1997. Guidelines for assessing marine mammal stocks: report of the GAMMS workshop April 3-5, 1996, Seattle, WA, 93 p. NOAA Tech. Memo. NMFS-OPR-12. Wade, P R., and D. P DeMaster. 1999. Determining the optimum interval for abundance surveys. In Marine mammal survey and assessment methods (G. W. Garner, S. C. Amstrup, J. L. Laake, B. F. J. Manly, L. L. McDonald, and D. G. Robertson, eds. ), p 53-66. A. A. Balkema. Rotterdam, Netherlands. Waring, G. T, J. M. Quintal, and S. L. Swartz (eds.l. 2001. U.S. Atlantic and Gulf of Mexico marine mammal stock assessments— 2001, 310 p. NOAA Tech. Memo. NMFS-NE-168. Wiseman, W. J., Jr., and W. Sturges. 1999. Physical oceanography of the Gulf of Mexico: processes that regulate its biology. In The Gulf of Mexico large marine ecosystem (H. Kumpf, K. Steidinger, and K. Sher- man (eds. ), p 77-92. Blackwell Science, Maiden, MA. Wiirsig, B., S. K. Lynn, T A. Jefferson, and K. D. Mullin. 1998. Behavior of cetaceans in the northern Gulf of Mexico relative to survey ships and aircraft. Aquat. Mamm. 24: 41-50. 933 Abundance of horseshoe crabs (Limulus polyphemus) in the Delaware Bay area David Hata Jim Berkson Department of Fisheries and Wildlife Sciences Virginia Polytectinic Institute and State University Blacksburg, Virginia 24061-0321 E-mail address (for J Berkson, contact author) iberksoniffivtedu III recent years, increasing commercial landings of horseshoe crabs {Limulus polyphemus) along the Atlantic coast of the United States have raised concerns that the present resource is in decline and insufficient to support the needs of its user groups. These concerns have led the Atlantic States Marine Fisher- ies Commission ( ASMFC ) to implement a fishery management plan to regulate the harvest (ASMFCM. In order to properly manage any species, specific management goals and objectives must be established, and these goals depend on the resource users involved (Quinn and Deriso, 1999). Horseshoe crabs present a distinct resource management challenge because they are important to a diverse set of users (Berkson and Shuster, 1999). Horseshoe crabs lay their eggs on sandy beaches in spring and summer, and migrating shorebirds rely heav- ily on the eggs to supply the energy required to complete their migration (Rudloe, 1980; Shuster and Botton, 1985; Castro and Myers, 1993; Botton et al., 1994; Myers, 1996; Thompson, 1998; Tsipoura and Burger, 1999). Biomedical companies catch horseshoe crabs for their blood, from which they produce Limulus Amebocyte Lysate (LAL) (Novitsky 1984; ASMFCi). LAL is used to detect contamination of in- jectable drugs and implantable devices by Gram-negative bacteria and is the most sensitive means available for detecting endotoxins (Novitsky, 1984). Finally, horseshoe crabs are harvested commercially for bait in the American eel (Anguilla rostrata), catfish (Ictci- lurus spp.), and whelk (Busycon spp.) fisheries (ASMFC). The goal of the ASMFC fishery man- agement plan is to ensure a sustainable population level that will support the continued use by these diverse ecologi- cal, biomedical, and fishing interests (ASMFC). Proper management of the resource requires information on the status and dynamics of the horseshoe crab population (Berkson and Shus- ter, 1999). However, the status of the population is poorly understood, and there is currently no reliable informa- tion on which to base any management scheme. Available fishery-independent surveys were not designed for horse- shoe crabs, and are of little or no value in assessing their status (ASMFC'^). Towards this end, the states of New Jersey, Delaware, and Maryland in conjunction with the ASMFC and the National Fish and Wildlife Foundation, funded a pilot benthic trawl survey for the fall of 2001. Data collected during this pilot trawl survey were used to es- timate the horseshoe crab population size in the Delaware Bay area. Methods This study was conducted in the vicin- ity of Delaware Bay, which is the center of abundance for horseshoe crabs on the Atlantic coast (Shuster, 1982). The study area extended from north of Cape May, New Jersey, to south of Ocean City, Maryland (39°10'N to 38°10'N), and from shore out to 22,2 km (Fig. 1). The area was divided into four strata based on distance from shore and topography, both of which influence crab distribu- tion. Distance from shore was con- sidered important because horseshoe crab abundance decreases with depth (Botton and Ropes, 1987a). Therefore, the area was split into an inshore zone from 0 to 5.6 km (0 to 3 nautical miles [nmi] ) from shore and an offshore zone from 5.6 to 22.2 km (3 to 12 nmi) from shore. Topography was also consid- ered important because commercial fishermen stated that crabs are more abundant in troughs (Burke^; Eutsler'*; Munson^). For this study, troughs were defined as at least 2.4 m deep, no more than 1.8 km wide, and more than 1.8 km long. These dimensions are common for troughs identified as important by the fishermen. The inshore and off- shore zones were both further divided into trough and nontrough areas. The resulting strata were inshore trough, inshore nontrough, offshore trough, and offshore nontrough. The study area was divided into grids of one-minute latitude by one-minute longitude. A grid was considered in- shore if the majority of its area was in water and inshore of the 5.6-km divid- ing line. A grid was considered offshore if the majority of its area was offshore of the 5.6-km dividing line and inshore of the 22.2-km boundary. A grid was also considered a trough if the long ax- is of a trough passed through the grid. • ASMFC (Atlantic States Marine Fisheries Commission). 1998. Interstate fishery management plan for horseshoe crab. Fishery management report no. 32, 58 p. Atlantic States Marine Fisheries Com- mission. 1444 Eye Street, NW, Sixth Floor, Washington, DC 20005. 2 ASMFC. 1999. Horsehoe crab stock assessment report for peer review. Stock assessment report No. 98-01 (supplement), 47 p. Atlantic States Marine Fisheries Commission, 1444 Eye Street, NW, Sixth Floor, Washington, DC 20005. ■* Burke, C. 2001. Personal commun. 25 Cove Drive, North Cape May, NJ 08204. ■* Eutsler, J. 2001. Personal commun. 11933 Gray's Corner Road, Berlin, MD 21811. ^ Munson, R. 2001. Personal commun. Box 358, Newport, NJ 08345. Manuscript approved for publication 6 March 2003 by Scientific Editor Manuscript received 22 July 2003 at NMFS Scientific Publications Office. Fish Bull. 101:933-938 (2003). 934 Fishery Bulletin 101(4) 3830' Figure 1 Study area and sampling locations. Symbols indicate type and location of strata. Day and night tows were made at each location. A grid was considered nontrough if no trough long axis passed through it. Each grid was therefore assigned to one of the four strata. Twelve grids were randomly selected in each stratum, for a total of 48 unique sampling locations. The fishermen also stated that time of day influenced horseshoe crab catchability (Burke^; Eutsler"*; Munson^). Therefore, grids were sampled both in daylight and at night. The second tow in a grid (day or night) was made near the location of the first to reduce location variability, but slightly offset to avoid possible influence of the first tow on the catch of the second. The second tow was also made more than 24 hours after the first to avoid interac- tions, but no more than four days later, to avoid introducing other unknown variability Abundance estimates from the daytime and nighttime samples were calculated separately for comparison. Our study was conducted in the fall, between 10 Sep- tember and 16 October 2001. The stock assessment model adopted by the ASMFC requires abundance information on newly mature crabs, and identification requires that crabs have undergone a terminal molt. Crabs reportedly molt in the late summer and fall in the Delaware Bay area (Burke^; Eutsler"*; Munson-''). Sampling was conducted from a chartered 16.8-meter commercial fishing vessel. For capturing horseshoe crabs, commercial fishermen typically use a flounder trawl equipped with a Texas sweep (Burke'^; Eutsler'*; Munson^; Michels''). This modified sweep consists of a chain line in- ^ Michels, S. 2001. Personal commun. Delaware Department of Natural Resources and Environmental Control, Division of Fish and Wildlife, 89 Kings Hwy, P.O. Box 1401, Dover, DE 19901. NOTE Hata and Berkson: Abundance of Limulus polyphemus in the Delaware Bay area 935 stead of rope, which runs from wing to wing of the net (Fig. 2). The net ropeUne is attached behind the sweep chain. In addition, usually three rows of weight chain are attached behind the sweep chain. The chain sweep is considered more effective in digging crabs out of the bottom than the typical ground gear of most research trawls. We used a standard two-seam flounder trawl with an 18.3-m headrope and 24.4-m footrope. The net consisted of 14-cm stretched mesh polypropylene throughout and was equipped with chafing gear on the bag. The net was attached to the trawl doors by 91-m ground cables wrapped in rubber cookies. Tow duration was usually 15 minutes (bottom time), except for one tow in the Delaware Bay ship- ping channel, which was reduced to 7.5 minutes. We assumed that density was not affected by tow dura- tion (e.g. gear saturation was not a factor). All horseshoe crabs were culled from the catch, and either all or a subsample were examined. For subsamples of a large catch, 50 crabs greater than 150 mm prosomal width were examined, as well as all small, soft, and shedding crabs. Horseshoe crabs that were not examined were counted separately by sex. Examined crabs were measured for prosomal width and identified to sex and maturity. Maturity classifications were as follows: immature; primipa- rous (mature horseshoe crabs that had not spawned yet); and multiparous (crabs that had spawned at least once (Table 1| ). When catches were subsampled, characteristics of examined crabs were extrapolated to all crabs in that tow. Abundance was estimated for each demographic group as well as for the total. Tow distances were determined for most tows from beginning and ending positions and recorded by using Loran C. These are minima because they do not consider any deviations from a straight path. Distances were not recorded for three tows; therefore they were estimated as the mean distance of all other tows. Net width was estimated as half of the mean of the headrope and footline lengths (Fridman, 1986). The tow distance and net width were used to calculate the swept area to deter- mine the density of horseshoe crabs. We assumed that the ground cables and trawl doors were not effective in catch- ing crabs; therefore all fishing was done only by the net. No information is available on the efficiency of the ground cables or doors for horseshoe crabs, but we do not believe horseshoe crabs are mobile enough, nor swim fast enough, to be effectively herded by them. The mean density (crabs/km^) and variance in each stratum were calculated by assuming a zl-distribution (Aitchison and Brown, 1957; Pennington, 1983), and these estimates were combined by using formulas for a stratified random design (Cochran, 1977). The /i-distribution model is applicable to skewed data that consist of a portion of zero catches when the frequency of nonzero catches follows a lognormal distribution (Pennington 1983; Pennington 1996). With such skewed data, the estimator of the mean as defined for the 4-distribution model is more efficient than the sample mean estimator derived from the normal distri- bution (Smith, 1988). Areas by stratum and total area were \ to ground cable 1 •■"■'^^ OTrai^^^v WvVAW. ^^''k. i;,;::'^m,^. ■ ! ■ X '! ■ X' C<)OvM>i»«******"**!a***^'*'*"**^^»^*"****^**'****^^^ 280 - ^V^ ** 26,0 - 240 - A 220 - on n 0 10 20 30 40 50 60 70 80 90 100 110 120 340 - Tag 5D-2 320 - ^T**^***^ ^ It /***^ K »r*«***nr*n m 30 0 - Y^ 7 V A V L / \r^ 280 - i\ ^ /V \r V 26,0 - i^ Y i Y 240 - B 22 0 - on n 0 10 20 30 40 50 60 70 80 90 100 110 120 34 0 - Tag 5D-4 32 0 - 5 30 0 - 2 280 - Q. 26 0 - a> 24 0 - H 22 0 - 20 0 ~ c 0 10 20 30 40 50 60 70 80 90 100 110 120 34 0 - Tag 5D-5 32 0 ^ A^*VVVV********N 7"********^ /***-******,»♦*** 30 0 - 280 - 260 - 240 - D 22 0 - 20 0 - 0 10 20 30 40 50 60 70 80 90 100 110 120 340 -| Tag 5D-7 320 - V******^K f**^ AK/^f f***^ ti